I watched the HD DVD of THE FOUNTAIN last night, and I was bummed to see kind of dull blacks that I associate with excessive noise-reduction. Likewise I'm getting really sick of reviews sighting the appearance of actual film-grain as a flaw in new HD transfers. I'm tired of a films image being dumbed-down for people who can't handle that a 35mm image is composed of grain, I'm not talking about some artificially added noise, but the actual grain of the films neg.

Film grain isn't something that rides on top of the image, it is the image, and to insist on having it scrubbed away, is to willingly promote the loss of detail. IMHO it's the cart driving the horse and the opposite of what I signed on to HDM for.


Dammit! :mad:

Grain is detail, if they remove it, they remove detail!

Actually, it's not necessary to have grain to have detail. The observations by us humans are sometimes strange. In an experiment I read about many years ago, two pictures had exactly the same resolution, but one was grainy (noisy?). People thought the grainy picture was sharper, although the measured detail was the same.

Having a noise (grain) free image makes video compression easier. Similar subsequent images are the heart of modern compression techniques. Noise may be the biggest dissimilarity in many adjacent frames. More work for the compressor - and for what?

I don't understand why grain is a "feature" of film, but noise is a "flaw" in video!

Dave

The point is if grain is there (on purpose or not) you can't remove it without looseing detail.

If film is so much higher resolution than high definition media (HD-DVD, Blu-ray etc.), and film is made up of grain, should we actually be able to see the grain? Shouldn't they be using finer grained film/larger format film or better filming techniques (or perhaps using very high res video - red cameras?). I'm not saying they should remove all grain with processing but they should use better filming techniques so it's such high resolution that there is no grain when they film scan it and put it on high definition media.

I agree that if they totally remove grain that already exists in the source they'll probably remove some detail too :( (though I'm sure there are some forms of grain removal that are better than others - eg. by looking at info in multiple frames when removing grain for the current frame)

shouldnt this belong in the HD-DVD section?

shouldnt this belong in the HD-DVD section?

Isn't there some film grain in all formats of High Definition Media?

Isn't there some film grain in all formats of High Definition Media?

A lot of Blu Ray movies filter out grain (and detail)

There are some fancy techniques that can keep detail, while removing noise. They have been around for some time, and they just keep getting better. Such a technique might be to remove detail which changes randomly from frame to frame (noise), while retaining detail which has a correlation from frame to frame (image).

Dave

There are some fancy techniques that can keep detail, while removing noise. They have been around for some time, and they just keep getting better. Such a technique might be to remove detail which changes randomly from frame to frame (noise), while retaining detail which has a correlation from frame to frame (image).

Dave

This is just as bad as EE IMO.

There are some fancy techniques that can keep detail, while removing noise. They have been around for some time, and they just keep getting better. Such a technique might be to remove detail which changes randomly from frame to frame (noise), while retaining detail which has a correlation from frame to frame (image).

Dave

Well, so far some of those "fancy techniques" are making flat darks and neutrals, and giving a generally inorganic quality that I could do without. I always see close-up frame grabs of well lit hair, clothing and trees showing off these techniques, when the areas they screw over are in the shadows.

If film is so much higher resolution than high definition media (HD-DVD, Blu-ray etc.), and film is made up of grain, should we actually be able to see the grain? Shouldn't they be using finer grained film/larger format film or better filming techniques

Resolution and grain are different issues. A film stock like Kodak 7201 (50D) will have virtually no grain. Unfortunately it's not very sensitive to light. If you're shooting in darker situations you need something like Kodak 7218 (500T), which has grain. Both are exactly the same resolution, but one ha grain, one doesn't.

The same is true of the RED camera. If you set it to ISO 320 you have a pretty clean image. Change that to ISO 2000 and all of a sudden there is a TON of grain.

As time has gone on companies like Kodak have been able to reduce the amount of grain in their film stocks, and the sensitivity of digital cameras is increasing, but for now, grain is a part of filmmaking.

Resolution and grain are different issues. A film stock like Kodak 7201 (50D) will have virtually no grain. Unfortunately it's not very sensitive to light. If you're shooting in darker situations you need something like Kodak 7218 (500T), which has grain. Both are exactly the same resolution, but one ha grain, one doesn't.

The same is true of the RED camera. If you set it to ISO 320 you have a pretty clean image. Change that to ISO 2000 and all of a sudden there is a TON of grain.

As time has gone on companies like Kodak have been able to reduce the amount of grain in their film stocks, and the sensitivity of digital cameras is increasing, but for now, grain is a part of filmmaking.

Thanks for that clarification.

Resolution and grain are different issues. A film stock like Kodak 7201 (50D) will have virtually no grain. Unfortunately it's not very sensitive to light. If you're shooting in darker situations you need something like Kodak 7218 (500T), which has grain. Both are exactly the same resolution, but one ha grain, one doesn't.

The same is true of the RED camera. If you set it to ISO 320 you have a pretty clean image. Change that to ISO 2000 and all of a sudden there is a TON of grain.

As time has gone on companies like Kodak have been able to reduce the amount of grain in their film stocks, and the sensitivity of digital cameras is increasing, but for now, grain is a part of filmmaking.

If the grain is only mostly visible in dark scenes couldn't they film 'day for night' a bit more (or something similar - or is that not always realistic?) - also it may not work in scenes where there were dark areas (with grain) but most of the other areas of the scene are brighter (no grain).

If the grain is only mostly visible in dark scenes couldn't they film 'day for night' a bit more (or something similar - or is that not always realistic?) - also it may not work in scenes where there were dark areas (with grain) but most of the other areas of the scene are brighter (no grain).

"Day for night" can be difficult to pull off, and may cause the DP/Director to make compromises on shots and compositions (you typically avoid shooting the sky). Also, in order to sell the scene as night, you typically underexpose the image, say somewhere around 2 stops. And the bad thing about underexposing film (or video) is that it makes the image grainy.

EDIT: And trying to shoot 50D stock film inside would be cruel and unusual punishment to the crew. Trust me, it gets hot enough as it is.

I watched the HD DVD of THE FOUNTAIN last night, and I was bummed to see kind of dull blacks that I associate with excessive noise-reduction.
Dammit! :mad:
This title does not have excessive noise reduction.

"Day for night" can be difficult to pull off, and may cause the DP/Director to make compromises on shots and compositions (you typically avoid shooting the sky). Also, in order to sell the scene as night, you typically underexpose the image, say somewhere around 2 stops. And the bad thing about underexposing film (or video) is that it makes the image grainy.

EDIT: And trying to shoot 50D stock film inside would be cruel and unusual punishment to the crew. Trust me, it gets hot enough as it is.

Thanks. It sounds a bit complicated now. I thought that, instead of underexposing it in camera (which may lead to grain) you could alter the brightness/colour grade it in post production to make it look like night and that that way you'd have less grain (though there must be a reason they don't do that - maybe it would look less real that way).

Likewise I'm getting really sick of reviews sighting the appearance of actual film-grain as a flaw in new HD transfers.Ditto. I wish they'd just telecine them then leave them alone.

If the grain is only mostly visible in dark scenes couldn't they film 'day for night' a bit more (or something similar - or is that not always realistic?) - also it may not work in scenes where there were dark areas (with grain) but most of the other areas of the scene are brighter (no grain).

Day-for-night isn't a very ideal way of representing night scenes IMO. I can usually spot it a mile off.

What's so god-awful about a little grain? If new films lick grain, so be it, but I loathe the idea of trying to tweak old films with a fair amount of native grain to represent new ones. There are a lot of films from the 60's and 70's that took advantage of newer, fast film stocks which by their nature had a lot of grain (Godfather), I really hope they don't start scrubbing that out. It's actually part of the accepted aesthetic of these films at the time of their creation. To remove it skates close to colorization land in my book.

This title does not have excessive noise reduction.

So what's the deal then?

You can have my grain because I do not like it at all. It takes away from my movie experience.

Depends on the look of the film. If the director wants a gritty look, keep the grain, if they want a clean or sharp look, lose the grain. Pretty simple.

EE should never be in there. It's a defect.

I bet people would complain if the audio track included pops and hisses to simulate a vinyl record. We expect perfect audio recordings that emulate reality. So, why do people tolerate film grain? Film grain is the visual equivalent of those pop and hisses.

A lot of Blu Ray movies filter out grain (and detail)

Is this true? I am a HD DVD supporter but I never heard this before. Just give me the movie the way it was shot. I think we have a lot of younger folks coming up that are being exposed to HD video. They see a original movie source with grain and think it is flawed.

A lot of Blu Ray movies filter out grain (and detail)

OK, could you please post the proof of that. A link would be great.



P.S. I bet you can't find one......

I bet people would complain if the audio track included pops and hisses to simulate a vinyl record. We expect perfect audio recordings that emulate reality. So, why do people tolerate film grain? Film grain is the visual equivalent of those pop and hisses.

Funny you mention that.

Steven Spielberg believes that the "grain" in film is in essence the "life" of the medium.

I was joking about that with my brother once as we discussed the pops and crackles on vinyl records and we both joked that was "life" :)

I watched the HD DVD of THE FOUNTAIN last night, and I was bummed to see kind of dull blacks that I associate with excessive noise-reduction. Likewise I'm getting really sick of reviews sighting the appearance of actual film-grain as a flaw in new HD transfers. I'm tired of a films image being dumbed-down for people who can't handle that a 35mm image is composed of grain, I'm not talking about some artificially added noise, but the actual grain of the films neg.

Film grain isn't something that rides on top of the image, it is the image, and to insist on having it scrubbed away, is to willingly promote the loss of detail. IMHO it's the cart driving the horse and the opposite of what I signed on to HDM for.


Dammit! :mad:

It's like complaining about pixels. Some movies have grain as an important part of their cinematography (Eyes Wide Shut, Master and Commander: Far Side of the World, Big Momma's House 2).

ALTHOUGH sometimes if a crappy master is used then you get more grain than you should, in which case you can complain assuming you know you are getting more grain than you should due to a crappy master (i.e. you are familiar with how it looked in the theater).

I bet people would complain if the audio track included pops and hisses to simulate a vinyl record. We expect perfect audio recordings that emulate reality. So, why do people tolerate film grain? Film grain is the visual equivalent of those pop and hisses.

The "reality" is what reflects what the film looked like at the time of it's creation. Blotting out the grain that exists on the original pre-print material is tampering with reality.

OK, could you please post the proof of that. A link would be great.



P.S. I bet you can't find one......

How many examples do you need?

Depends on the look of the film. If the director wants a gritty look, keep the grain, if they want a clean or sharp look, lose the grain. Pretty simple.

EE should never be in there. It's a defect.

That's fine if the director is still alive to tell you so.

I could make the same argument about color. If color was as inexpensive as B/W in the 30's-50's, chances are most features would have been shot in color, does that sort of rationalization give license for colorization?

How is film-grain different from the texture on a canvas, or the strokes of a brush? Maybe the painter was striving for "realism," does that mean later generations should try and remove the presence of the medium from the piece?

Film is the medium that supports the art of photography, they are fused when the film is exposed and processed, why pull them apart after the fact? Especially if your taking detail along with it?

I like grain aswell, but for some older titles I have nothing against some restauration.

A lot of Blu Ray movies filter out grain (and detail)
And I thought a lot of BR movies are so grainy compared to HD-DVD. What is it? :rolleyes: Hint: Grain reduction is not related to the format, but the transfer and mastering processes which are the same for both formats.

I like grain aswell, but for some older titles I have nothing against some restauration.

When it's done with some care an awareness neither do I. Some times however too much license is taken.

For instance, when Warner Brothers released a DVD struck from the only surviving 2-strip TECHNICOLOR print of the (1933) MYSTERY IN THE WAX MUSEUM, they went way overboard. Unlike the full spectrum possible with 3-strip TECHNICOLOR, 2-strip was limited to a palette that derived all it's color from a mix of bluish-green and reddish-orange, but the DVD presented all kinds of deep blues that couldn't have been possible in 2-strip.

And I thought a lot of BR movies are so grainy compared to HD-DVD. What is it? :rolleyes: Hint: Grain reduction is not related to the format, but the transfer and mastering processes which are the same for both formats.

Have you compared Lionsgate releases like Rambo to the HD DVD equivalents?

And I thought a lot of BR movies are so grainy compared to HD-DVD. What is it? :rolleyes: Hint: Grain reduction is not related to the format, but the transfer and mastering processes which are the same for both formats.

Have you compared Lionsgate releases like Rambo to the HD DVD equivalents?

When it's done with some care an awareness neither do I. Some times however too much license is taken.

For instance, when Warner Brothers released a DVD struck from the only surviving 2-strip TECHNICOLOR print of the (1933) MYSTERY IN THE WAX MUSEUM, they went way overboard. Unlike the full spectrum possible with 3-strip TECHNICOLOR, 2-strip was limited to a palette that derived all it's color from a mix of bluish-green and reddish-orange, but the DVD presented all kinds of deep blues that couldn't have been possible in 2-strip.

Well thats pushing it.

My DVD version of the orginal king kong did include a colorized version, but it was clearly stated that this isnt the orginal, and they included a black and white version of it anyway.

A lot of Blu Ray movies filter out grain (and detail)

Yes, the Warner ones.

What movies does warner filters?

What movies does warner filters?

Every single one. In my opinion, of course.

Every single one. In my opinion, of course.


I doubt it (that warner does it, not that it isnt your opinion)

I doubt it (that warner does it, not that it isnt your opinion)

Of course you are entitled to your opinion just as much as I am entitled to mine. :)

As I am sure we both know, you can't prove they don't and I can't prove they do. ;)

Of course you are entitled to your opinion just as much as I am entitled to mine. :)

As I am sure we both know, you can't prove they don't and I can't prove they do. ;)


Well thats true, but I hope you dont claim they did DNR to 300. ;)

Well thats true, but I hope you dont claim they did DNR to 300. ;)

Less than usual. ;)

Putting that title aside, what is the reason for your doubt that they do it?

What movies does warner filters?
Almost all of them, from what I've seen. They do have some stellar looking ones as well though (300)...

A lot of Blu Ray movies filter out grain (and detail)So do a lot of HD DVD ones :(

The first reason would be that why would they?

You could say because they wouldnt have bandwith to encode with grain, but 300 clearly shows that they can encode grain with modest bitrate. So I guess bandwith isnt really a good reason why they should use DNR.

Then we come to that some warner titles look softer then other. To easiest reason would be that the process gives different looks.

Thats why I asked what movies that were DNR. Because when we have different titles we can look on the process of that movie and conclude why the movie looks the way it does.

The first reason would be that why would they?

You could say because they wouldnt have bandwith to encode with grain, but 300 clearly shows that they can encode grain with modest bitrate. So I guess bandwith isnt really a good reason why they should use DNR.



The look of 300 is exceedingly stylized and distorted. The fact that some grain can be preserved at low bitrates with this kind of source material does not really show very much about more naturalistic looking source material.

The look of 300 is exceedingly stylized and distorted. The fact that some grain can be preserved at low bitrates with this kind of source material does not really show very much about more naturalistic looking source material.

I dont think the grain in 300 was easier to the encoder then more avarage filmgrain.

But look at the Alexander trailer in the WMVHD showcase. DVD bitrates and fair amount of grain.

Large amount of grain isnt something that is frequent on every type of film. Different types of film leaves different amount of grain.

Actually no film I have seen on HDM that I also saw in theaters have dissapointed me. They more or less loke the same.

I dont think the grain in 300 was easier to the encoder then more avarage filmgrain.

But look at the Alexander trailer in the WMVHD showcase. DVD bitrates and fair amount of grain.

Large amount of grain isnt something that is frequent on every type of film. Different types of film leaves different amount of grain.

Actually no film I have seen on HDM that I also saw in theaters have dissapointed me. They more or less loke the same.

Have you seen Alexander in full on either BD or HD DVD? No grain.

Could someone please edit the brush strokes and paint buildup out of my Picasso? I can't believe that given modern computer techniques, Picasso wouldn't use MS Paint and a color laser printer if he had the choice. These silly brush strokes aren't present in the latest computer paintings, why the heck should I put up with it on a Picasso (especially since I paid millions for it)? Oh, and when you are done with that, let's talk about that hack Van Gogh. Starry Starry Night doesn't look like any stars I've ever seen. Slap some CGI on that piece of crap right now. I want to see something that looks like the Big Dipper or else!!

How many examples do you need?

One, would be nice. It was actually HD DVD that has a "grain remover" in spec. Also, VC1 is rumored to show less grain than AVC/MPEG2.

Every single one. In my opinion, of course.

I saw Gothika in HD DVD last night. Great resolution with nice preservation of grain. So, I don't think so...:rolleyes:

I saw Gothika in HD DVD last night. Great resolution with nice preservation of grain. So, I don't think so...:rolleyes:

I haven't seen that one.

I suppose I'll have to get it and see for myself.

One, would be nice. It was actually HD DVD that has a "grain remover" in spec. Also, VC1 is rumored to show less grain than AVC/MPEG2.

It aint grainremover in spec but grainadding as metadata.

The process is simple, remove grain then the decoder add grain with help of metadata. But this is only in AVC and hasnt been used on any title.

The reason why VC1 sometimes shows less grain is how the image is build up. So VC1 may show slight less grain, mpeg2 on the other hand shows more.

Thats the problem with lossy encodes, they will not show a lossless video. So either you see slight more or slight less. The question would be whats true to the master?

Grain/Noise that has come thanks to compression isnt more true to the master.

And what 300 shows is they can handle very large amount of grain so I wonder why they should have any problem to show less?

It aint grainremover in spec but grainadding as metadata.

The process is simple, remove grain then the decoder add grain with help of metadata. But this is only in AVC and hasnt been used on any title.

The reason why VC1 sometimes shows less grain is how the image is build up. So VC1 may show slight less grain, mpeg2 on the other hand shows more.

Thats the problem with lossy encodes, they will not show a lossless video. So either you see slight more or slight less. The question would be whats true to the master?



Do you not think that removing the grain and then adding it would lose some amount of detail?? I know it hasn't been used, Thompson was in hd dvd spec though.

I agree that VC1 appears to have less grain and mpeg2 appears to have more.

Is AVC closer to the master as far as grain?

Compression can accent grain.

Never buy the "that's the way it's supposed to look" line.

Compression can accent grain to a level that was not seen on the original.

Do you not think that removing the grain and then adding it would lose some amount of detail?? I know it hasn't been used, Thompson was in hd dvd spec though.

I agree that VC1 appears to have less grain and mpeg2 appears to have more.

Is AVC closer to the master as far as grain?

I have never done any scientific test on the matter wich one of the 3 codecs is most true to the master.

But Its easy to guess that AVC VC1 should be better then the older Mpeg2.

But it isnt just about the codec in itself but also how welldeveloped the encoding tools are.

You have surley seen that different mpeg2 encoder have been very varied in terms of quality.

But still, 300 seems to be very true to the source.

Compression can accent grain.

Never buy the "that's the way it's supposed to look" line.

Compression can accent grain to a level that was not seen on the original.

I would think compression would reduce grain. If you take a picture in photoshop. Add some noise to it. Then compress it by say 50%. You will notice the image becomes softer and a lot of grain is lost.

Compression can accent grain.

Never buy the "that's the way it's supposed to look" line.

Compression can accent grain to a level that was not seen on the original.

I was under the impression it took MORE bitrate to preserve grain.

Compression can accent grain.

Never buy the "that's the way it's supposed to look" line.

Compression can accent grain to a level that was not seen on the original.

Maybe so, but it isn't rocket science to know when your scaling it to a level that best approximates the appearance of your source, and when you're deliberately crossing that line smooth it out entirely.

I have never done any scientific test on the matter wich one of the 3 codecs is most true to the master.

But Its easy to guess that AVC VC1 should be better then the older Mpeg2.

But it isnt just about the codec in itself but also how welldeveloped the encoding tools are.

You have surley seen that different mpeg2 encoder have been very varied in terms of quality.

But still, 300 seems to be very true to the source.

I did see 300 in the theater. To the best of my recolection, the Warner release appears to be close to the source.

Now, Transformers appeared to have quite a bit LESS grain than I remember seeing in the theater.

I have seen mpeg2 look incredible, Underworld Evolution/Crank/Black Hawk Down, but I have seen it look horrible also.

Could someone please edit the brush strokes and paint buildup out of my Picasso? I can't believe that given modern computer techniques, Picasso wouldn't use MS Paint and a color laser printer if he had the choice. These silly brush strokes aren't present in the latest computer paintings, why the heck should I put up with it on a Picasso (especially since I paid millions for it)? Oh, and when you are done with that, let's talk about that hack Van Gogh. Starry Starry Night doesn't look like any stars I've ever seen. Slap some CGI on that piece of crap right now. I want to see something that looks like the Big Dipper or else!!

My point exactly, you could argue that any medium or support isn't part of the artists intent and therefore expendable. I say bullsh!t, what represents the film best on the whole is that which renders it as close to how the pre-print dictated when it was new.

I did see 300 in the theater. To the best of my recolection, the Warner release appears to be close to the source.

Now, Transformers appeared to have quite a bit LESS grain than I remember seeing in the theater.

I have seen mpeg2 look incredible, Underworld Evolution/Crank/Black Hawk Down, but I have seen it look horrible also.

Well actually since the theatercopy is some generations away from the negative you should see more grain in theaters then you do when you watch HDMs.

Underworld looks good, the Spanish HD DVD looks to be about the same.
Crank is shoot digital and should provide a very clean source.
Black Hawk Down, there is a scene in the beginning that seems to be more noisy then the theatrical version, but I could remember wrong.

My favourite Mpeg2 would be Kingdom of heaven. Tears of the sun aint bad either but its hard to match the footage from a Ridley Scott film. (Well BHD is shoot more gritty)

Actually, it's not necessary to have grain to have detail. I don't understand why grain is a "feature" of film, but noise is a "flaw" in video!

Dave

You're right you don't understand. Just ask Robert Harris the foremost film restoration expert extant, his opinion is the diametric opposite of yours.

Art

You can have my grain because I do not like it at all. It takes away from my movie experience.

Certainly! I love grain! I'm having, grain, grain, grain, grain, eggs, and grain!:p

You're right you don't understand. Just ask Robert Harris the foremost film restoration expert extant, his opinion is the diametric opposite of yours.

Art

I don't know how they relate to photoshop filters, but when I'm designing an image, I sometimes use noise to cloak the jagged or soft appearance of an inferior image, or knock a background further back so it doesn't compete with the foreground. It's a simulated grain tool that flattens the playing field and pulls out detail, detail which never comes back with sharpening or saturation tweaks. That's a lot different than the grain I get out of the gate on a high-rez scan of a negative or transparency which is the image itself.

Have you compared Lionsgate releases like Rambo to the HD DVD equivalents?
I don't have to, because it's not a format issue. Everyone can put everything on both formats. Grainy and clean, DNRed to death and no DNR. If you want to bring in the format BR has higher bandwidth for video so it can deal with more noise before falling apart by using higher bit rates than HD-DVD.

You're right you don't understand. Just ask Robert Harris the foremost film restoration expert extant, his opinion is the diametric opposite of yours.
Art
Grain is not required for detail as any grain free highly detailed digital acquisiton shows. Is grain required for detail from film elements? Again, no. You can temporally filter film and remove the grain while keeping the detail. Is it easy? Hell, no. You need top notch motion estimation and huge number cracking ressources.
You can also use fine grain film and use more light, giving you lots of detail with almost no visible grain.

Grain is not required for detail as any grain free highly detailed digital acquisiton shows. Is grain required for detail from film elements? Again, no. You can temporally filter film and remove the grain while keeping the detail. Is it easy? Hell, no. You need top notch motion estimation and huge number cracking ressources.
You can also use fine grain film and use more light, giving you lots of detail with almost no visible grain.

But none of that applies to whether the HDM version closely replicates the film version. You seem to want grain eliminated, if not by the choice of film stock or lighting, then at least by post-processing the HDM transfer. This is anathema to those who wish the replication of the artwork be as close to the original film version as possible. We all know that "(g)rain is not required for detail". What the purists are stating is that film grain, as present in the original print, is as important to the original artwork as brush strokes are to an original painting. As I stated before, they could make the HDM version of The Godfather look like a Discovery Channel filmed-on-HD documentary about sunshine and roses, but would it really be The Godfather?

Re: Grain

Just pray that we don't have any more DNR ruined titles like this:


http://www.cif-forums.com/png/starfighter_1.png

http://www.cif-forums.com/png/starfighter_2.png

http://www.cif-forums.com/png/starfighter_3.png

DNR, the killer of classic 1980s films.

That example clearly shows how some type of DNR looks like.

And very few HDMs shows that amount of DNR.

But none of that applies to whether the HDM version closely replicates the film version. You seem to want grain eliminated, if not by the choice of film stock or lighting, then at least by post-processing the HDM transfer. This is anathema to those who wish the replication of the artwork be as close to the original film version as possible. We all know that "(g)rain is not required for detail". What the purists are stating is that film grain, as present in the original print, is as important to the original artwork as brush strokes are to an original painting. As I stated before, they could make the HDM version of The Godfather look like a Discovery Channel filmed-on-HD documentary about sunshine and roses, but would it really be The Godfather?

That's a pretty excellent summation in my view too. In short don't fix it if it's not broken, and the Godfather with grain isn't broken.

Grain is not required for detail as any grain free highly detailed digital acquisiton shows. Is grain required for detail from film elements? Again, no. You can temporally filter film and remove the grain while keeping the detail. Is it easy? Hell, no. You need top notch motion estimation and huge number cracking ressources.
You can also use fine grain film and use more light, giving you lots of detail with almost no visible grain.


But it does not eliminate grain... again deferring to Robert Harris on this one. He is of the opinion that there is no way to eliminate grain without negatively impacting detail to some dgree even under best case circumstances. Of course DNR etc is not going to touch best case as it is presently implemented.

Art

But none of that applies to whether the HDM version closely replicates the film version. You seem to want grain eliminated, if not by the choice of film stock or lighting, then at least by post-processing the HDM transfer. This is anathema to those who wish the replication of the artwork be as close to the original film version as possible. We all know that "(g)rain is not required for detail". What the purists are stating is that film grain, as present in the original print, is as important to the original artwork as brush strokes are to an original painting. As I stated before, they could make the HDM version of The Godfather look like a Discovery Channel filmed-on-HD documentary about sunshine and roses, but would it really be The Godfather?

Agreed !!

Re: Grain

Just pray that we don't have any more DNR ruined titles like this:

DNR, the killer of classic 1980s films.

That's nasty, Darkman?

That's a pretty excellent summation in my view too. In short don't fix it if it's not broken, and the Godfather with grain isn't broken.

Agreed !!

Awww, shucks. I feel like I'm in a WWII movie, the Star Spangled Banner is playing and the sky is filled with planes. :o

That's nasty, Darkman?

The Last starfighter.

The Last starfighter.

The big "STARFIGHTER" printed on the side of the video game was my first clue. ;)

I don't have to, because it's not a format issue. Everyone can put everything on both formats. Grainy and clean, DNRed to death and no DNR. If you want to bring in the format BR has higher bandwidth for video so it can deal with more noise before falling apart by using higher bit rates than HD-DVD.

I never said it was a format issue.

Art,

It wasn't that someone might prefer to have grain. What boggles my mind is that that same person hates noise in video.

Some years ago, we had some video productions transferred to 16mm film for projection to large audiences. (Video projection was pretty bad back then.) People worked hard to diminish the flaws of video (scan lines, low res, etc.) from the film transfer. Should we not do the same thing when going the opposite direction? Each format contributes its own set of flaws that keep it from a perfect copy of the original live scene. Now granted, An artist (director, whatever) might want those flaws for some artistic reason. But if there is grain in a movie that could have easily been made without grain (because we know that a lower grain film stock was then available), are we to assume the grain is an artistic decision? Perhaps! If the grain only appears on night scenes, are we to assume the director was forced to use a grainy stock, and would have preferred not to have the grain? Possibly. Grain is a FLAW of film, but if a director prefers it that's certainly an option.

I recorded, and made an HD DVD of a Tony Bennett special on NBC last fall. There are sections made to look like an old dirty, grainy film. This was obviously the director's choice. To this viewer though, it was a distraction.

Dave

Art,

It wasn't that someone might prefer to have grain. What boggles my mind is that that same person hates noise in video.

Some years ago, we had some video productions transferred to 16mm film for projection to large audiences. (Video projection was pretty bad back then.) People worked hard to diminish the flaws of video (scan lines, low res, etc.) from the film transfer. Should we not do the same thing when going the opposite direction? Each format contributes its own set of flaws that keep it from a perfect copy of the original live scene. Now granted, An artist (director, whatever) might want those flaws for some artistic reason. But if there is grain in a movie that could have easily been made without grain (because we know that a lower grain film stock was then available), are we to assume the grain is an artistic decision? Perhaps! If the grain only appears on night scenes, are we to assume the director was forced to use a grainy stock, and would have preferred not to have the grain? Possibly. Grain is a FLAW of film, but if a director prefers it that's certainly an option.

I recorded, and made an HD DVD of a Tony Bennett special on NBC last fall. There are sections made to look like an old dirty, grainy film. This was obviously the director's choice. To this viewer though, it was a distraction.

Dave

I agree. While some grain is a direct result of artistic vision, I would guess most, specially older or lower budget films, are byproducts of low budget and care. If you see lousy acting in a film or cheesy special effects, you wouldn't call it director's intent even though the director signed off on the film, it's lousy acting or cheesy effects. Preservation of grain can be a slippery slope much like restoration of any painting or artifacts. You have to walk a fine balance between preservation and restoration.

my take on this:

i don't see grain in everyday life, unless of course maybe my sunglasses are dirty...

and i don't see people/objects/the world moving along at 24/30/60fps

so, anything else is one step removed from reality :D

/flame suit on

I agree. While some grain is a direct result of artistic vision, I would guess most, specially older or lower budget films, are byproducts of low budget and care. If you see lousy acting in a film or cheesy special effects, you wouldn't call it director's intent even though the director signed off on the film, it's lousy acting or cheesy effects. Preservation of grain can be a slippery slope much like restoration of any painting or artifacts. You have to walk a fine balance between preservation and restoration.

It does not matter that if was "artistic intent" or not, it is the finished art - warts and all. Otherwise, we can argue that the colors of the Mona Lisa be changed because of advances in pigments since Leonardo's time and we off are off down the slippery slope that goes with that concept. Regardless, art restoration is left to the experts, not us "we wanna see it clearly, like outa window!!" plebes. Since we have the opinion of a world class film restorer to rely on, I'll stick with his opinion. I don't think anyone here can honestly say that their own uneducated, inexperienced, "I want what looks good to me!!" opinion trumps that of Robert Harris, can they?

my take on this:

i don't see grain in everyday life, unless of course maybe my sunglasses are dirty...

and i don't see people/objects/the world moving along at 24/30/60fps

so, anything else is one step removed from reality :D

/flame suit on

I would but your post doesn't even rise to the level of deserving being flamed.;):D

Art

I watched the HD DVD of THE FOUNTAIN last night, and I was bummed to see kind of dull blacks that I associate with excessive noise-reduction. Likewise I'm getting really sick of reviews sighting the appearance of actual film-grain as a flaw in new HD transfers. I'm tired of a films image being dumbed-down for people who can't handle that a 35mm image is composed of grain, I'm not talking about some artificially added noise, but the actual grain of the films neg.

Film grain isn't something that rides on top of the image, it is the image, and to insist on having it scrubbed away, is to willingly promote the loss of detail. IMHO it's the cart driving the horse and the opposite of what I signed on to HDM for.


Dammit! :mad:

just want to add the blu-ray version sucks too.and the movie itself sucks.
i like movies of this type,but i couldnt get it on this one.

Re: Grain

Just pray that we don't have any more DNR ruined titles like this:

[The Last Starfighter]

DNR, the killer of classic 1980s films.

Ugh, I was so excited to get The Last Starfighter on HD-DVD, it's one of my favorites (I've even seen the Musical). And I was so disappointed when I actually popped it in and saw those shifting smears of color from the DNR. :(

This is just as bad as EE IMO.

oh come on jeff dont be a party pooper:)

I always thought the grainy look of night scenes and the "film grain" seen in normal light scenes was two different things. I thought grain was introduced by the camera when trying to obtain the best picture with low light.

I know as a simple test, using my camcorder in sunlight produces a crystal clear life like picture. Using it at night, the camera automatically changes several settings to produce a grainy picture. If I try to correct the grain by changing the shutter settings etc. the grain is reduced, but so is the amount of detail in the picture. The only way to correct it is to add artificial light which then makes the 'night scene' look more like a 'lit room scene'.

I know movie studios have some pretty insane camera technology.. but I still think that the physics of lenses and photography still have some side effects when filming in low light conditions.

The only dark/night scenes I've seen on HDM that were virtually grain free... were those using heavy CG. Which of course, is not affected by lens physics.

I guess I have to ask how we would know whether the grain represented on disc is close to the master. Some have commented that movies have less grain than they remembered in theaters. After 35+ years of watching movies in theaters, I can't say I remember the level of grain in a movie well enough to compare to a home viewing 6-8 months or even years later. Unless grain was a feature of the movie, as in 300, why should it leave such an impression so that we would have a clear recognition of its levels.

FWIW, I noticed the grain much more prevalently on PotC CBP on BD than I remembered in theaters. Was it there in theaters? Maybe, but it registered strongly with me in HD, and not in a terribly good way.

I believe HDM should be as close to the master as possible, but how do we know what the master looks like?

Depends on the look of the film. If the director wants a gritty look, keep the grain, if they want a clean or sharp look, lose the grain. Pretty simple.

EE should never be in there. It's a defect.


Grain is essentially white noise at a certain level and frequency spectrum. All the comments about losing grain losing detail are a bit off the mark. It takes bandwidth to recreate the grain and some actual detail is lost. Sometimes you don't notice, sometimes you do. If you had infinite bandwidth or saved the image losslessly then keeping the grain preserves detail.

Ideally you subtract out the grain measuring its level and frequency spectrum then add it back in after decoding. Some purists scream about this approach, but it is the best for quality and allows for the option of switching off the grain or adjusting its level if you like. Your eye/brain is NOT going to be able to tell the difference between the original grain and the artificial grain (assume done right of course). HD-DVD has this grain reproduction method I believe, of course with possible bandwidth and space constraints this is probably a good thing (though they don't allow you to adjust the grain). I support Blu-Ray, but Blu-Ray should allow this also. In the future when grain is strictly an artistic decision since the original HD video image image will be essentially grain free, why not keep the grain and detail separate?

I guess I have to ask how we would know whether the grain represented on disc is close to the master. Some have commented that movies have less grain than they remembered in theaters. After 35+ years of watching movies in theaters, I can't say I remember the level of grain in a movie well enough to compare to a home viewing 6-8 months or even years later. Unless grain was a feature of the movie, as in 300, why should it leave such an impression so that we would have a clear recognition of its levels.

FWIW, I noticed the grain much more prevalently on PotC CBP on BD than I remembered in theaters. Was it there in theaters? Maybe, but it registered strongly with me in HD, and not in a terribly good way.

I believe HDM should be as close to the master as possible, but how do we know what the master looks like?

There is grain in the original negative AND also in the film positive released to the theater. Different final positives could have wildly different grain levels.

More importantly video tends to exaggerate contrast in dark scenes. Ever seen the original VHS release of STAR WARS? There were HUGE mask outlines on all the spaceships in space scenes. This because contrast levels in low light scenes were lower and thus the images over-saturated on purpose to fool the eye into see only light and dark.

my take on this:

i don't see grain in everyday life, unless of course maybe my sunglasses are dirty...

and i don't see people/objects/the world moving along at 24/30/60fps

so, anything else is one step removed from reality :D

/flame suit on

Oh well, do you see aliens, zombies, CGI sunsets, mice that cook, and pirates in your typical day?

Grain is essentially white noise at a certain level and frequency spectrum. All the comments about losing grain losing detail are a bit off the mark. It takes bandwidth to recreate the grain and some actual detail is lost. Sometimes you don't notice, sometimes you do. If you had infinite bandwidth or saved the image losslessly then keeping the grain preserves detail.

Ideally you subtract out the grain measuring its level and frequency spectrum then add it back in after decoding. Some purists scream about this approach, but it is the best for quality and allows for the option of switching off the grain or adjusting its level if you like. Your eye/brain is NOT going to be able to tell the difference between the original grain and the artificial grain (assume done right of course). HD-DVD has this grain reproduction method I believe, of course with possible bandwidth and space constraints this is probably a good thing (though they don't allow you to adjust the grain). I support Blu-Ray, but Blu-Ray should allow this also. In the future when grain is strictly an artistic decision since the original HD video image image will be essentially grain free, why not keep the grain and detail separate?

Sorry, but I think all this tinkering is bullsh!t. I want an HD that presents the film as film, not the film as video simulating the film.

That's like a MacDonald's hamburger having to be recolored and reflavored because for the purpose of uniformity the ground-beef has been processed beyond recognition.

But none of that applies to whether the HDM version closely replicates the film version. You seem to want grain eliminated, if not by the choice of film stock or lighting, then at least by post-processing the HDM transfer.
No. I want grain to stay. I hate grain removal unless it's artifact free, director approved and the result looks like low grain film and not processed video.

But it does not eliminate grain... again deferring to Robert Harris on this one. He is of the opinion that there is no way to eliminate grain without negatively impacting detail to some dgree even under best case circumstances. Of course DNR etc is not going to touch best case as it is presently implemented.
Art
Depends on the case in question. Eliminating grain 100% is probably not possible, but to a large extent. It depends how much noise/grain there is and how many frames in sequence you can combine during filtering. The more the higher the SNR you can achieve. I would not worry about detail but artifacts because the motion estimation is not 100% right. You can actually get more detail out of temporal filtering that there is on any individual frame.

my take on this:

i don't see grain in everyday life, unless of course maybe my sunglasses are dirty...

and i don't see people/objects/the world moving along at 24/30/60fps

so, anything else is one step removed from reality :D

/flame suit on


Sigh....

Films are a medium , they are an abstraction and experiencing a film is nothing like experiencing reality. To criticise film for differing from reality is meaningless. You might as well criticise a vase because its not a banana.

No. I want grain to stay. I hate grain removal unless it's artifact free, director approved and the result looks like low grain film and not processed video.

Be careful. I think I get your points but it sounds like you are talking out of both sides of your mouth. I agree if there was a way to do it well, consistantly, I'd say no grain too but to date that isn't happening. You know as well as I do that what we have gotten ,thus far, has looked like someone polished off high frequency information right with the grain.

I say keep it as close to the original as possible, for now, as the technique most likiely to result in the most superior images.



Art

I say keep it as close to the original as possible, for now, as the technique most likiely to result in the most superior images.
Art
I agree. But the grain bitching trend works against us here... :mad:

I agree. But the grain bitching trend works against us here... :mad:

That's why we must continue to speak out against the grain complaining at every opportunity. ;)

I agree. But the grain bitching trend works against us here... :mad:

True !!



That's why we must continue to speak out against the grain complaining at every opportunity. ;)


Circle the wagons !:D

Art

The best thing would be if they builded grainremoval in the HDM players (optional), so they kept grain in the encoding, but still let J6P have their smoothed picture.

The best thing would be if they builded grainremoval in the HDM players (optional), so they kept grain in the encoding, but still let J6P have their smoothed picture.

When did J6P (are we somewhat condescending toward him/her?) ask for the artificial reduction of grain? When did the amount of grain become a measure for good taste?

Again, I ask, how do we know when we're getting the correct amount of grain in a transfer when we don't see the master?

When did J6P (are we somewhat condescending toward him/her?) ask for the artificial reduction of grain? When did the amount of grain become a measure for good taste?

Again, I ask, how do we know when we're getting the correct amount of grain in a transfer when we don't see the master?

They dont ask, they complain -"hey that look worse then DVD..."

They dont ask, they complain -"hey that look worse then DVD..."

Pearls before swine.

Art

Pearls before swine.

Art

Indeed , seems to be a recurrent problem with good quality HD generally.

When did J6P (are we somewhat condescending toward him/her?) ask for the artificial reduction of grain? When did the amount of grain become a measure for good taste?

Again, I ask, how do we know when we're getting the correct amount of grain in a transfer when we don't see the master?

We assume it is a correct representation until someone in the know says otherwise. We do not immediately assume that the transfer is terrible and then challenge people who have no more information than we do to prove us wrong. There are enough film experts out on the net to be fairly assured that a bad transfer will eventually be brought to light. Robert Harris has certainly been vocal about this very subject on more than one release.

Of course, you may believe in the evil film restorers society that increases grain on purpose and then hides the master so we won't ever find out . . . MUUUUHAHAHAHAH!!!! In that case, I'm buying a roll of Heavy Duty Reynold's Wrap and opening a haberdashery. Would you like to see something in a modified pie plate? ;)

Let's inject some science behind film grains to have a meaningful discussion.

Film grain is a byproduct the materials used for photographic film, specifically, silver halides that are randomly exposed during shooting a film. It's there not because directors wanted them to be there, simply because they are there no matter what. Grains have also been changing due to film making (NOT movie making) improvement. Today's grains are certainly smaller and not as distracting as it used to be.

Photographic film generations (older folks here:)) are just conditioned to see film grains and associated it with artistic expression, I guess that's what art is. anyway:)

Digital generations would certainly see it differently, such as film vs. video debate, it's just different taste. As more and more movies are to be shot with digital cameras, I expect more complaints about lack of grains and loss of artistic value :)

Many people confuse film grains with noise. They are different. Film grains are always there with film camera, even film shot in bright day light. While noises usually only happen in low light condition when not enough light for film and digital sensors to generate meaningful signal.

Be careful. I think I get your points but it sounds like you are talking out of both sides of your mouth. I agree if there was a way to do it well, consistantly, I'd say no grain too but to date that isn't happening. You know as well as I do that what we have gotten ,thus far, has looked like someone polished off high frequency information right with the grain.

I say keep it as close to the original as possible, for now, as the technique most likiely to result in the most superior images.



Art

I vehemently agree with this. The goalis accuracy to the original first, with the exceptions being to degradations to the sources available (scratches, noise etc) which differ from how the film first would have appeared. Grain has a significant aesthetic affect, and whether it was a technologically imposed limitation of a particular shot, film stock, lighting condition, etc, OR whether it was an express desire of the film-maker, that original image should be preserved as closely as possible.

The ability to remove grain well does not yet exist, and the methods we have now can end up quite hideous. In the absence of that type of option, the negatives are simply unacceptable. And quite frankly, the film-aesthetic that grain provides is a significant part of the viewing experience, and even if there was a magical way to eliminate grain while preserving ALL detail, I would prefer not to use it.

Let's inject some science behind film grains to have a meaningful discussion.

Film grain is a byproduct the materials used for photographic film, specifically, silver halides that are randomly exposed during shooting a film. It's there not because directors wanted them to be there, simply because they are there no matter what. Grains have also been changing due to film making (NOT movie making) improvement. Today's grains are certainly smaller and not as distracting as it used to be.

Photographic film generations (older folks here:)) are just conditioned to see film grains and associated it with artistic expression, I guess that's what art is. anyway:).
Yes sweetheart, but if it's there in the original camera negs, that is the image it's not some dirt riding on the surface to be wiped off or diluted with digital soap and water.

When they restored the Sistine Chapel many were repulsed that the warm, subdued patina of years of incense resins & grime had warmed over what in fact were paintings with a very vibrant almost garish palette. My thoughts were "that's what Michael Angelo painted, that's what a legitimate restoration should reveal"

Grain may be have been a byproduct of certain filmstocks, but if it was, so be it, it's only a flaw to those who can't handle anything presented on it's own terms, those terms were established the day those cameras rolled.

I don't know if anybody ever noticed this before, but studios love to obscure the age of films over a decade old on new packaging, fearing that the consumer won't relate to and buy anything out of the realm of their immediate memory, or that seems old. That's kind of how I see grain-scrubbing, an effort to blanch fingerprints off a film of a different era.

Grain isn't even linear in the progression of cinema, fast film-stocks from the 60's-90's could be much more grainy than that of the 20's-50's if were talking about having access to original pre-print nitrates, and not safety prints and internegs which have picked up two to three generations just on film.

Sorry, but I think all this tinkering is bullsh!t. I want an HD that presents the film as film, not the film as video simulating the film.

That's like a MacDonald's hamburger having to be recolored and reflavored because for the purpose of uniformity the ground-beef has been processed beyond recognition.

Since Video is a different medium than Film and you are dead-set to have the original grain and not a simulacrum I suggest you invest in a 35mm film projector and buy the original prints.

Grain is something the Artist may or may not choose to have, be if he chooses to have it he doesn't choose the exact random pattern you see. Insisting to see the exact original grain is like insisting you can tell the difference in your shower every morning because the pattern of drops is different (assuming the same temp and water pressure). Because video codecs throw out info you can't perceive automatically you are already not getting the original picture. Worse, the grain you love is the least faithfully reproduced part of the image, TA-DA because it is grain and you can't discern much of it's detail. If it helps preserve other details, I say simulate the grain.

Since Video is a different medium than Film and you are dead-set to have the original grain and not a simulacrum I suggest you invest in a 35mm film projector and buy the original prints.

Grain is something the Artist may or may not choose to have, be if he chooses to have it he doesn't choose the exact random pattern you see. Insisting to see the exact original grain is like insisting you can tell the difference in your shower every morning because the pattern of drops is different (assuming the same temp and water pressure). Because video codecs throw out info you can't perceive automatically you are already not getting the original picture. Worse, the grain you love is the least faithfully reproduced part of the image, TA-DA because it is grain and you can't discern much of it's detail. If it helps preserve other details, I say simulate the grain.

Well, correct me if I'm wrong, but what your describing is smooshing out grain in the master because it's pesky to work with, and then taking the smooshed image and tossing some noise back in so that it looks somewhat more organic again? How is that not effectively adding a generation on?

Is that the master being altered to suit the limitation of the pre-print, or the pre-print being altered to suit the limitation of the video-master?

Yes sweetheart, ...:confused:
The rest of your post has nothing to do with mine, I am neutral on grains:)

It looks like ffdshow with the de-noise controls all cranked up. Pasty and flat with no texture definition. It looks like a painting. Not at all film-like and not at all acceptable for a HD format.

I actually detest excessive film grain. It is semi-random noise and just a artifact of the film media we have to work with. It obscures detail and takes huge bit rate to compress and store electronically, making most electronic delivery more expensive and less detailed. I admire the generations of filmmakers who have incorporated grain into our movie culture as an art form but still wish they could find something else.

Unfortunately I probably detest the softness from over-zestful attempts to remove grain even more so I am sort of stuck between a rock and a hard place for this generation of the technology. :(

- Tom

Quote:
Originally Posted by Timothy Ramzyk View Post
Yes sweetheart, ..:confused:
The rest of your post has nothing to do with mine, I am neutral on grains:)


Just being wierd;)

I actually detest excessive film grain. It is semi-random noise and just a artifact of the film media we have to work with. It obscures detail and takes huge bit rate to compress and store electronically, making most electronic delivery more expensive and less detailed. I admire the generations of filmmakers who have incorporated grain into our movie culture as an art form but still wish they could find something else.

Unfortunately I probably detest the softness from over-zestful attempts to remove grain even more so I am sort of stuck between a rock and a hard place for this generation of the technology. :(

- Tom

looks like we do need holographic after all :D

Seriously though how can a films original grain be obscuring it's detail? It's not like the un-grained detail is hidden underneath somewhere.

looks like we do need holographic after all :D

Seriously though how can a films original grain be obscuring it's detail? It's not like the un-grained detail is hidden underneath somewhere.

Actually it is exactly like that. Removing grain noise isn't just smearing out the existing detail, but actually using previous and following frames to determine edge transitions and sharpen them along with other signal processing tricks to tease out hidden detail, which ironically is what good DVD upscaling does. I suspect you have no problem with watching your older DVDs in upscaled mode. Most likely DVD upscaling completely throws out grain in order to extrapolate detail.

Actually it is exactly like that. Removing grain noise isn't just smearing out the existing detail, but actually using previous and following frames to determine edge transitions and sharpen them along with other signal processing tricks to tease out hidden detail, which ironically is what good DVD upscaling does. I suspect you have no problem with watching your older DVDs in upscaled mode. Most likely DVD upscaling completely throws out grain in order to extrapolate detail.

Thats not the same thing. A DVD is already "degrained" and when you upconvert they do diffferent things to hide noise etc.

But removing grain and try to interpolate isnt to going bring out more detail.

And yes it isnt fun to watch upscaled DVD anymore ;)

Actually it is exactly like that. Removing grain noise isn't just smearing out the existing detail, but actually using previous and following frames to determine edge transitions and sharpen them along with other signal processing tricks to tease out hidden detail, which ironically is what good DVD upscaling does. I suspect you have no problem with watching your older DVDs in upscaled mode. Most likely DVD upscaling completely throws out grain in order to extrapolate detail.

Some films upscale better than others, I use a DVDO to upscale, it does a nice job, but I pretty-much leave it on 1080-i for everything (my projectors max is 1080-i).

I can't use my XA2 on the projector, because it's pre-HDMI

The ability to remove grain well does not yet exist, and the methods we have now can end up quite hideous.
You have to make distinction between real time HW solutions and non real time software systems. The latter can remove grain without any artifacts with a subset of all footage and for the rest you can get very good results as well when finetuning by hand. The real time systems used for DVD or HD mastering give the process a bad name. They are so far useless for high quality work.

I had to clean up some 16mm stock footage for use in a mainstream 35mm film a few weeks ago.

First I ran it through two seperate grain removal tools : a vector based interpolation one produces ringing artifacts on strong detail but does an excellent job of preserving sharpness. The other filter based one softens significantly but doesn't have ringing artifacts. I then maxed the two images together ( combined the maximal pixel values , this has the advantage of losing the ringing yet preserving important sharpness in the lower tonal regions but it leaves the bright detail soft and maintains any bright grain structure : bright grain structure tends to be less objectionable than dark though.).

I then tracked and stabilised the image ( handheld camera so I had to use numerous tracks and average them together to get a rock solid track) and averaged the stabilised frames together ( at least 8 frames are needed for a good average). I normally use a 3d tracker called boujou for this sort of thing as its automated and tracks thousands of features at the same time but the motionblur on the plate was too severe to get a reliable automated track.

I then reversed the stabilise for the resulting averaged frame ( free of noise but maintained detail sharpness) and comped in the clean plate in brighter areas of the plate where I felt the grain was still too evident.

I had to do this in three stages as the camera move was large enough to require 3 seperate reference frames to give me enough coverage.

I then animated the colour correction on the clean plates to match the colour variation across the original footage. I also had to rotoscope and paint the areas of the foreground that passed in front of the clean plated areas. I also had to match the motionblur of the original camera move on the clean plates

Then I manually painted out remaining large bits of dirt and neg damage.

Then I colour corrected the entire image to counter some fading and flatness on the original footage and match the contempory shots in the scene.

Then I resized the image to match a 35mm frame size.

Then I applied a little sharpening , pulling it back in certain areas through another set of hand animated mattes.

Then I matched and applied grain that matched the rest of the scene.

Took me two days to do one shot ( I had two to do ) both shots were probably about 8 secs long. End results probably looked a little soft compared with actually shooting the footage on 35mm but they no longer looked like 16mm and would be accepted as "normal" looking in the context of the scene.

I've got over 10 years experience doing this sort of thing and it took me 4 days to do 16secs worth using fairly state of the art tools at 2k resolutions.

I think we can live with a little bit of normal film grain.

I had to clean up some 16mm stock footage for use in a mainstream 35mm film a few weeks ago.

First I ran it through two seperate grain removal tools : a vector based interpolation one produces ringing artifacts on strong detail but does an excellent job of preserving sharpness. The other filter based one softens significantly but doesn't have ringing artifacts. I then maxed the two images together ( combined the maximal pixel values , this has the advantage of losing the ringing yet preserving important sharpness in the lower tonal regions but it leaves the bright detail soft and maintains any bright grain structure : bright grain structure tends to be less objectionable than dark though.).

I then tracked and stabilised the image ( handheld camera so I had to use numerous tracks and average them together to get a rock solid track) and averaged the stabilised frames together ( at least 8 frames are needed for a good average). I normally use a 3d tracker called boujou for this sort of thing as its automated and tracks thousands of features at the same time but the motionblur on the plate was too severe to get a reliable automated track.

I then reversed the stabilise for the resulting averaged frame ( free of noise but maintained detail sharpness) and comped in the clean plate in brighter areas of the plate where I felt the grain was still too evident.

I had to do this in three stages as the camera move was large enough to require 3 seperate reference frames to give me enough coverage.

I then animated the colour correction on the clean plates to match the colour variation across the original footage. I also had to rotoscope and paint the areas of the foreground that passed in front of the clean plated areas. I also had to match the motionblur of the original camera move on the clean plates

Then I manually painted out remaining large bits of dirt and neg damage.

Then I colour corrected the entire image to counter some fading and flatness on the original footage and match the contempory shots in the scene.

Then I resized the image to match a 35mm frame size.

Then I applied a little sharpening , pulling it back in certain areas through another set of hand animated mattes.

Then I matched and applied grain that matched the rest of the scene.

Took me two days to do one shot ( I had two to do ) both shots were probably about 8 secs long. End results probably looked a little soft compared with actually shooting the footage on 35mm but they no longer looked like 16mm and would be accepted as "normal" looking in the context of the scene.

I've got over 10 years experience doing this sort of thing and it took me 4 days to do 16secs worth using fairly state of the art tools at 2k resolutions.

I think we can live with a little bit of normal film grain.

Sounds incredibly interesting,thanks for letting us inside a little ! :) Do you have access to any before and after frames that we could see ?

Art

Sounds incredibly interesting,thanks for letting us inside a little ! :) Do you have access to any before and after frames that we could see ?

Art


Nope but you can go and see "Fred Claus" when it comes out and keep an eye out for some footage of a red and white plane taking off and flying over a snowscape.

I've got over 10 years experience doing this sort of thing and it took me 4 days to do 16secs worth using fairly state of the art tools at 2k resolutions.

How reliable is motion estimation these days with the best tools? Do you get subpixel accuracy routinely everywhere, or are there still big problems with some types of footage that require a lot of tuning by hand?

Film grain is a byproduct the materials used for photographic film, specifically, silver halides that are randomly exposed during shooting a film. It's there not because directors wanted them to be there, simply because they are there no matter what.

Different film stocks have different densities of visible film grain. The choice of film stock can be very much an artistic decision. Witness the opening black & white sequence in Casino Royale, which is filled with heavy grain by purposeful design. If the director had wanted that sequence to be less grainy (or hardly grainy at all), he certainly could have shot it that way.

Your implication that all grain is a filmic flaw that no director really wants is way off base.

To me grain is like a noisy TV channel. I consider Kubric the master in movie visuals and he avoided grain. If you want to reproduce a newsreel or TV news effect, grain has a place, or if you've under-exposed the film and need to boost it. Maybe dial-a-grain?

Also, no grain means I'm there, while grain means I'm watching a picture by someone else who was there.

(Ducking from movie purists)

To me grain is like a noisy TV channel. I consider Kubric the master in movie visuals and he avoided grain. If you want to reproduce a newsreel or TV news effect, grain has a place, or if you've under-exposed the film and need to boost it. Maybe dial-a-grain?It's up to the director, but I would say if you want to reproduce a TV new effect, NOISE has a place - not grain.

To me grain is like a noisy TV channel. I consider Kubric the master in movie visuals and he avoided grain. If you want to reproduce a newsreel or TV news effect, grain has a place, or if you've under-exposed the film and need to boost it. Maybe dial-a-grain?

Also, no grain means I'm there, while grain means I'm watching a picture by someone else who was there.

(Ducking from movie purists)

I don't know what is sadder, the fact that you actually believe what you are saying or the fact that judging from your "(Ducking from movie purists)", you obviously revel in your own lack of appreciation for art. But even sadder than either of those is using Kubrick to support your nonsense. :rolleyes:

I don't know what is sadder, the fact that you actually believe what you are saying or the fact that judging from your "(Ducking from movie purists)", you obviously revel in your own lack of appreciation for art. But even sadder than either of those is using Kubrick to support your nonsense. :rolleyes:
Realism....2001, eyes wide shut... Since when is grain art?

...
Your implication that all grain is a filmic flaw that no director really wants is way off base.
What nonsense is this? Are you disputing that grain is not a film byproduct?

Where did I say no director wants it? "It's there not because directors wanted them to be there" certainly does not mean directors do not want it. It is there no matter what directors want or not, some directors may want them, some may not.

Where did I say it's a flaw? Let me repeat one more time, it's a byproduct, byproduct does not mean flaw, dictionary is your friend here.

For someone who was caught lying just to win an argument, you need (1) learn how to read; (2) don't put words into other people's mouth. Talk about sad state of American education:)

How reliable is motion estimation these days with the best tools? Do you get subpixel accuracy routinely everywhere, or are there still big problems with some types of footage that require a lot of tuning by hand?

Nowhere near sub-pixel in fact you usually get more artifacts the closer you get to pixel accurate. Vector field usually work best at half the resolution of the image they relate to.

It requires a lot of manual intervention and cleanup.

Realism....2001, eyes wide shut... Since when is grain art?


So I guess Dr. Strangelove was realistic too? I didn't know the 60's Cold War was devoid of color. :rolleyes: As far as your very small list goes (your lack of familiarity with Kubrick explains your use of him to defend your taste), Kubrick made the decision to use low grain film in those movies. That does not mean every film should or can strive for elimination of film grain. Film grain is part of the medium of film. It is no more separable from the art than paint, brush and canvas are separable from the art of painting or clay/stone are from the art of sculpture. To be truthful, grain is one of the very essences of the medium. If you wish to reproduce the art in the closest way possible to the original, the best way is not to immediately remove an essence of the medium just because some clown "lahks his movin pickchures to look jest lahk re-al life."

So I guess Dr. Strangelove was realistic too? I didn't know the 60's Cold War was devoid of color. :rolleyes: As far as your very small list goes (your lack of familiarity with Kubrick explains your use of him to defend your taste), Kubrick made the decision to use low grain film in those movies. That does not mean every film should or can strive for elimination of film grain. Film grain is part of the medium of film. It is no more separable from the art than paint, brush and canvas are separable from the art of painting or clay/stone are from the art of sculpture. To be truthful, grain is one of the very essences of the medium. If you wish to reproduce the art in the closest way possible to the original, the best way is not to immediately remove an essence of the medium just because some clown "lahks his movin pickchures to look jest lahk re-al life."
OMG, all those directors choose their grain size, and beat the DP when they don't get it. Are they adding grain when the film is shot in digital?

Do you add grain to your digital camera shots? :rolleyes:

No more bronze or laquered art pieces either? Is there grain in your auto paint?

Strangelove wasn't trying for reality IMO - parody. Grain could be ignored and it wouldn't change that story. Grain is there whether they like it or not. If they come out with a zero grain film it won't be used?

To me grain is like a noisy TV channel. I consider Kubric the master in movie visuals and he avoided grain. If you want to reproduce a newsreel or TV news effect, grain has a place, or if you've under-exposed the film and need to boost it. Maybe dial-a-grain?

Also, no grain means I'm there, while grain means I'm watching a picture by someone else who was there.

(Ducking from movie purists)

What on earth are you talking about? Have you ever seen a Kubrick film? They are by NO means devoid of grain.

What on earth are you talking about? Have you ever seen a Kubrick film? They are by NO means devoid of grain.
Dang. My 2001 HD DVD should be arriving today, and now I'll be watching the damn thing for grain.

So for the anti-grain people:

Do you also want all your B&W films to be colorized?
Do you watch all foreign films with dubbed english audio?
Do you never watch science-fiction movies at all because it is not realistic?
Do you fast-forward through all slow-motion segments because it is not realistic?
Do you not watch any movies with stylized cinematography?
Do you not watch any fantasy-based movies (LOTR, etc)?
Do you avoid Saving Private Ryan because Spielberg chose some grainy and streaky film and lenses for certain shots, and that's not realistic?
Do you never watch any action films for any reason, since they are almost never realistic?
Do you never watch any animated films for the same reason?
Do you never watch any film with special effects? CGI?
Do you never watch any film that has any foley sound effects?
Do you only watch documentaries?
Do you only watch C-SPAN?
And C-SPAN 2?
Do you never watch any films with actors?
Do you even care about film at all?

Dang. My 2001 HD DVD should be arriving today, and now I'll be watching the damn thing for grain.

Have you ever seen that film in 70mm?

Have you ever seen Full Metal Jacket? That has tons of grain.

So for the anti-grain people:

Do you also want all your B&W films to be colorized?
Do you watch all foreign films with dubbed english audio?
Do you never watch science-fiction movies at all because it is not realistic?
Do you fast-forward through all slow-motion segments because it is not realistic?
Do you not watch any movies with stylized cinematography?
Do you not watch any fantasy-based movies (LOTR, etc)?
Do you avoid Saving Private Ryan because Spielberg chose some grainy and streaky film and lenses for certain shots, and that's not realistic?
Do you never watch any action films for any reason, since they are almost never realistic?
Do you never watch any animated films for the same reason?
Do you never watch any film with special effects? CGI?
Do you never watch any film that has any foley sound effects?
Do you only watch documentaries?
Do you only watch C-SPAN?
And C-SPAN 2?
Do you never watch any films with actors?
Do you even care about film at all?
I'm happy with your (assumed) POV on all the above. I don't think most grain is intended or necessary.

Have you ever seen that film in 70mm?

Have you ever seen Full Metal Jacket? That has tons of grain.
Saw 2001 in 70mm. IIRC the space shots were as crystal clear as they could be made. Didn't mention FMJ because the public knew Vietnam through grainy news photography, so Kubrick may have wanted that. That made it more real in that case. Grain has its place but where there is no artistic purpose, get rid of it.

To me HDM should have the same grain that was present in the source. No more, no less. If a codec is found to introduce grain then that is not something I want. If a codec is found to reduce grain, I don't want that either. The problem as others have pointed out is that we never have access to the masters. So statements like MPEG-2 introduces grain or VC-1 removes it, are baseless without the original to compare with.

Take a film like Miami Vice. Many panned it for it's appearance. The director, Mann, decided to use a high grain medium to capture the night shots without external lighting. People cried bloody murder about it. Personally I think the transfer is exactly what it should be: a faithful representation of what I saw in the theater.

I'm afraid I can't agree with the people who wish to see grain reduced from the original. The spirit of home theater, to me, will always be to reproduce what I saw in the theater as accurately as possible. Any enthusiast that agrees with this should have only one answer on film grain.

Realism....2001, eyes wide shut... Since when is grain art?

Movies are art, film (for now) their medium. Remove the grain, remove the medium. You suffer from the arrogance that the standards of the moment in which you live trump all those that proceed it, and that your standards should be retrofit on to the past ones.

Don't fix what isn't broken.

Just curious, when you channel surf, do you just naturally pass by anything in B/W?

Movies are art, film (for now) their medium. Remove the grain, remove the medium. You suffer from the arrogance that the standards of the moment in which you live trump all those that proceed it, and that your standards should be retrofit on to the past ones.

Don't fix what isn't broken.

Just curious, when you channel surf, do you just naturally pass by anything in B/W?
I've been around long enough to see, real time, many movie standards. Just like some don't like 24 fps stutter in pans, or 3/2 judder, I think grain is a distracting artifact where it doesn't have a purpose. Just watched all of War in cable definition (shudder), so I do look for substance not style. Unlike some around here.

Edit P.S. but I feel as strongly about OAR as you do in this

OMG, all those directors choose their grain size, and beat the DP when they don't get it. Are they adding grain when the film is shot in digital?

Do you add grain to your digital camera shots? :rolleyes:

No more bronze or laquered art pieces either? Is there grain in your auto paint?

Strangelove wasn't trying for reality IMO - parody. Grain could be ignored and it wouldn't change that story. Grain is there whether they like it or not. If they come out with a zero grain film it won't be used?

You just don't get it. Just because a digital camera doesn't have grain does not mean copies of a film should not have grain. Why? Because we aren't talking about digital camera shots!!! We are talking about film!!! When grain (or black and white, or color, or CGI effects, or a silent soundtrack or a French soundtrack, or cheesy, obvious F/X shots, or a pimple on Marilyn Monroe's behind) is present in the final release to theater, then we want it in the release to (pick your consumer format of choice). If the final release is a zero grain film, the consumer format release should have ZERO GRAIN!!! If they end up with a grainy film, the consumer format release should BE GRAINY!! In other words, the "limitations" (yeah, right :rolleyes:) of film should not be seen as something to eliminate in the subsequent copies to a consumer format.

Nowhere near sub-pixel in fact you usually get more artifacts the closer you get to pixel accurate.
Why is that? In theory in the image parts where motion estimation is 100% right there are no artifacts from filtering.

Saw 2001 in 70mm. IIRC the space shots were as crystal clear as they could be made. Didn't mention FMJ because the public knew Vietnam through grainy news photography, so Kubrick may have wanted that. That made it more real in that case. Grain has its place but where there is no artistic purpose, get rid of it.

So for the other shots in 2001, you advocate altering the film to suit your PERSONAL preferences? That's pretty selfish.

I'm happy with your (assumed) POV on all the above. I don't think most grain is intended or necessary.

Ted Turner argued that old B&W films could have been shot in color if the technology was available to justify his colorization of B&W films. He assumed (like you) that B&W was not "intended or necessary" if color had been available. Your assumption carries the same lack of insight and/or information of the director's intent. Besides, who cares what the director "intended???" Whether he intended it or not, the final artwork is the final artwork; limitations and compromises are inseparable from "intent." Unless you agree that Greedo really did shoot first?

Keep Ted Turner and his goddamned crayolas away from my movie!!
- Orson Welles

I think grain is a distracting artifact
Think that if you like, but it's not, a films grain is the very composition of the original image itself. As soon as light hit those camera negs the filmstock, responded in accordance to what it's properties were, and that is the image.

If in 1975 a director chose to use a fast low-light stock for a night scene, they knew that grain would be the compromise for the convenience. It may not have been their aesthetic desire to have grain, but just an acceptable trade-off. In fact, at the time seeing grain may have just been par for the course and an excepted look to audience and filmmakers alike. Seeing that grain now is a record or true representation of what was the standard in 75, which was actually more accepting of film looking grainy than 30 years earlier. Rub that all out and your watching a film made in 1975 and tweaked to appear as if it were made in 2007.

If for some reason they hit on a sure-fire way for filming and comfortably watching 3-D down the road, and it replaced 2-D as the standard, do we start manipulating previous films to appear as if the were also 3-D?

If for some reason they hit on a sure-fire way for filming and comfortably watching 3-D down the road, and it replaced 2-D as the standard, do we start manipulating previous films to appear as if the were also 3-D?

Something tells me I really don't want to read his answer to that question. :confused:

Why is that? In theory in the image parts where motion estimation is 100% right there are no artifacts from filtering.

Because its never 100% right , you need a bit of averaging of the target vectors. Consistently tracking a single pixel in a moving image is practically impossible. There are always going to be times when the track loses lock because the relative differences between features/pixels is beneath the threshold of discernability.

If you had a handful of vectors you could correct them but it becomes unrealistic to adjust the thousands of vectors resulting from an optical flow process. Hopefully if you have less vectors they will be accurate enough to give the overall motion required to get a good result.

Contrary to the demos , its not a panacea.

So for the anti-grain people:
...
...
Do you even care about film at all?

What an elitist final summation. Watch it and enjoy it the way you do or be a schmuck evidently. :rolleyes:

I for one would like the option to adjust the graininess while watching -- Just as I enjoy the option of using subtitles or dubbed audio. Timothy Ramzyk started off this whole mishagosh with the implicit assumption detail was being lost by removing grain. Anyone that has seen signal processed video from a surveillance camera or a space probe can tell you that removing grain and noise can bring out detail that was TOTALLY unobservable before.

Video is not film, some compromises must be made in order to keep storage size reasonable, else we would record every pixel 48bits deep losslessly. Large as Blu-Ray is it still won't hold a terabyte.

What an elitist final summation. Watch it and enjoy it the way you do or be a schmuck evidently. :rolleyes:

I for one would like the option to adjust the graininess while watching -- Just as I enjoy the option of using subtitles or dubbed audio. Timothy Ramzyk started off this whole mishagosh with the implicit assumption detail was being lost by removing grain. Anyone that has seen signal processed video from a surveillance camera or a space probe can tell you that removing grain and noise can bring out detail that was TOTALLY unobservable before.

Video is not film, some compromises must be made in order to keep storage size reasonable, else we would record every pixel 48bits deep losslessly. Large as Blu-Ray is it still won't hold a terabyte.

What you see from surveillance camera is not film grain, it's chroma noise.

Actually, the thing about film grain is that film itself is made of up grains, it's a chemical thing with how light hits the spectral sensitizer sitting on the surface of a silver-halide grain. IE: it is the nature of film that causes this grain effect.

If the movie wasn't made with film, we could get a different look, cleaner in some eyes, but artificial for others. In reality, people can have their preferences, but if the movie is shot on film, if you want to be accurate to the original, you're going to have to preserve the grain, you can't really remove it unless you're going for a look different than the original.

...
If for some reason they hit on a sure-fire way for filming and comfortably watching 3-D down the road, and it replaced 2-D as the standard, do we start manipulating previous films to appear as if the were also 3-D?

Actually this will probably happen, I'm sure you will be able to switch 3D off if you choose, just like you can watch a colorized film in B&W if you want. Ironic you should choose this example with "The Nightmare Before Christmas in 3D" now showing in theaters.

I think what you care about more is that people watch things the way you want them to watch them because you assume only the original unadulterated experience has any merit. I am a purist in somethings, but I don't force my preferences on others. Mostly I like choice. I think you will find in 10 years that grain is almost a thing of the past, only inserted for artistic reasons, and then mostly to denote an age gone by.

If for some reason they hit on a sure-fire way for filming and comfortably watching 3-D down the road, and it replaced 2-D as the standard, do we start manipulating previous films to appear as if the were also 3-D? I wonder what grain looks like in 3D :rolleyes:

Actually it looks flat.

Actually this will probably happen, I'm sure you will be able to switch 3D off if you choose, just like you can watch a colorized film in B&W if you want. Ironic you should choose this example with "The Nightmare Before Christmas in 3D" now showing in theaters.

I think what you care about more is that people watch things the way you want them to watch them because you assume only the original unadulterated experience has any merit. I am a purist in somethings, but I don't force my preferences on others. Mostly I like choice. I think you will find in 10 years that grain is almost a thing of the past, only inserted for artistic reasons, and then mostly to denote an age gone by.

I don't care it you watch it through the wrong end of a telescope sitting naked in a bowl of Jell-O. On the othe hand, the anti-grain people were requesting that because video is not film, the transfer to video should eliminate grain. So exactly who is forcing preferences on whom?

What you see from surveillance camera is not film grain, it's chroma noise.

Actually, the thing about film grain is that film itself is made of up grains, it's a chemical thing with how light hits the spectral sensitizer sitting on the surface of a silver-halide grain. IE: it is the nature of film that causes this grain effect.

If the movie wasn't made with film, we could get a different look, cleaner in some eyes, but artificial for others. In reality, people can have their preferences, but if the movie is shot on film, if you want to be accurate to the original, you're going to have to preserve the grain, you can't really remove it unless you're going for a look different than the original.

I'm not sure how many of my responses you have read. I am for preserving the grain look, but it doesn't need to be the exact original grain (though offering the option of dialing down the grain if you want). I know film grain can be cleaned up and see hidden details come out because I have seen demos of this signal processing technology, so while film isn't surveillance video it is still much the same signal processing that work.

I used this analogy earlier in this thread, but I will repeat it. If the temperature and pressure are identical in two morning showers can you tell them apart because the pattern of drops is different? Grain is like a shower, the experience isn't the individual drops but the stream.

What an elitist final summation. Watch it and enjoy it the way you do or be a schmuck evidently. :rolleyes:

I for one would like the option to adjust the graininess while watching -- Just as I enjoy the option of using subtitles or dubbed audio. Timothy Ramzyk started off this whole mishagosh with the implicit assumption detail was being lost by removing grain. Anyone that has seen signal processed video from a surveillance camera or a space probe can tell you that removing grain and noise can bring out detail that was TOTALLY unobservable before.

Video is not film, some compromises must be made in order to keep storage size reasonable, else we would record every pixel 48bits deep losslessly. Large as Blu-Ray is it still won't hold a terabyte.

You're right video isn't film, grain is film. The foremost expert on film restoration says point blank that present grain removal techniques all are detrimental to image detail.

People are seeing film grain for the first time, it seems, and are the same people ,it also seems , who have been weaned on video and therefore don't like the look of grain due to that fact.

The argument is just repeated over and over except there are way more video lovers each day. Until we have a technology that can reliably be used to reduce grain without reducing high frequency information then leave the grain. Even if technology exists, will they use it on a 1940's Bogart film for example.....very unlikely. What then ,another Citizen Kane mutilation.

Art

To me grain is like a noisy TV channel. I consider Kubric the master in movie visuals and he avoided grain...

EYES WIDE SHUT is probably the grainiest 35mm production of the last ten years. Kubrick shot it using Kodak's fast 500 ASA 5298 filmstock- already a grainy stock to begin with (in fact it was in the process of being discontinued when Kubrick began production on EYES, but he choose it over the newer, much finer grained "Vision" 500 stock because he didn't like how the Vision stock's color shifted when push-processed)- and had it pushed a full 2-stops in the lab during processing. Talk about a grainfest!

IMO, the heavy graininess (blacks in particular were SWIMMING with heavy grain in the 35mm prints) worked and added to the dreamy nature of the film, but I digress- to say Kubrick "avoided grain" is wrong, as proven at the very least by his final film.

Vincent

I think what you care about more is that people watch things the way you want them to watch them because you assume only the original unadulterated experience has any merit. I am a purist in somethings, but I don't force my preferences on others. Mostly I like choice. I think you will find in 10 years that grain is almost a thing of the past, only inserted for artistic reasons, and then mostly to denote an age gone by.

Sorry but I think your ass-backawards on this , your asking for an alteration of the original, I'm asking for a close rendition of the original. Why isn't manipulating the original more an imposition of taste on others than saying play it as it lays?

It may be one preference pitted against another, but at least my preference is true to the source, your's is based on an unwillingness to allow for anything but a homogenized uniform look for everything. I'm gonna venture to guess I'm interested in a lot more 30+ year-old titles than you are anyway?

EYES WIDE SHUT is probably the grainiest 35mm production of the last ten years. Kubrick shot it using Kodak's fast 500 ASA 5298 filmstock- already a grainy stock to begin with (in fact it was in the process of being discontinued when Kubrick began production on EYES, but he choose it over the newer, much finer grained "Vision" 500 stock because he didn't like how the Vision stock's color shifted when push-processed)- and had it pushed a full 2-stops in the lab during processing. Talk about a grainfest!

IMO, the heavy graininess (blacks in particular were SWIMMING with heavy grain in the 35mm prints) worked and added to the dreamy nature of the film, but I digress- to say Kubrick "avoided grain" is wrong, as proven at the very least by his final film.

Vincent
I admit I wasn't looking at the grain :) I'm stepping out of this red anthill -bye

Sorry but I think your ass-backawards on this , your asking for an alteration of the original, I'm asking for a close rendition of the original. Why isn't manipulating the original more an imposition of taste on others than saying play it as it lays?

It may be one preference pitted against another, but at least my preference is true to the source, your's is based on an unwillingness to allow for anything but a homogenized uniform look for everything. I'm gonna venture to guess I'm interested in a lot more 30+ year-old titles than you are anyway?

Great post and I bet you hit the nail on the head.

Art

I admit I wasn't looking at the grain :) I'm stepping out of this red anthill -bye

Gotta admit, that's pretty funny :)

I also have to say, too bad you weren't looking at the grain, 'cause the grainiest parts of the frame were, ahem, the "dirtiest" as well. Kubrick designed it that way...

(kidding of course)

Vincent

You're right video isn't film, grain is film. The foremost expert on film restoration says point blank that present grain removal techniques all are detrimental to image detail.

People are seeing film grain for the first time, it seems, and are the same people ,it also seems , who have been weaned on video and therefore don't like the look of grain due to that fact.

The argument is just repeated over and over except there are way more video lovers each day. Until we have a technology that can reliably be used to reduce grain without reducing high frequency information then leave the grain. Even if technology exists, will they use it on a 1940's Bogart film for example.....very unlikely. What then ,another Citizen Kane mutilation.

Art


Again, you make the mistake that assuming that leaving in the grain leaves in the detail. Codecs throw out detail in order to save space. Details hopefully that wouldn't normally be noticed. When a codec also has to deal with grain it throws out detail in the original picture (non grain detail) in order to compromise. So you don't get a sharp image with grain on it you get a dull image with grain on it. The main argument seems to be that Timothy Ramzyk wants the original grain not simulated grain because he thinks original detail is being lost if the grain is processed separately. Ironically I guarantee detail is being lost if it isn't.

Do you really think you can tell the difference between artificial grain and real grain? Because if you can't I don't understand what all the fuss is about.

Again, you make the mistake that assuming that leaving in the grain leaves in the detail. Codecs throw out detail in order to save space. Details hopefully that wouldn't normally be noticed. When a codec also has to deal with grain it throws out detail in the original picture (non grain detail) in order to compromise. So you don't get a sharp image with grain on it you get a dull image with grain on it. The main argument seems to be that Timothy Ramzyk wants the original grain not simulated grain because he thinks original detail is being lost if the grain is processed separately. Ironically I guarantee detail is being lost if it isn't.

Do you really think you can tell the difference between artificial grain and real grain? Because if you can't I don't understand what all the fuss is about.
Then essentially what your saying is that compression just can't handle and preserve the info on film? So just for S & G's how big of a file would one need to preserve the whole works. I mean apart from our home disks, would a lot of future digital preservation (hopefully) be uncompressed?

Also, back to my original whine, why does The Fountain look pretty weak? It had both a milky softness and noise at the same time.

Then essentially what your saying is that compression just can't handle and preserve the info on film? So just for S & G's how big of a file would one need to preserve the whole works. I mean apart from our home disks, would a lot of future digital preservation (hopefully) be uncompressed?

Also, back to my original whine, why does The Fountain look pretty weak? It had both a milky softness and noise at the same time.

The Fountain looks bad because it wasn't shot very well. Crumby film of the highest order a real disapointment from Aronofski.

Maintaining film grain is probably one of the most difficult things for encoding schemes like VC1 ( small transient detail is the thing you most notice going from D5 to a compressed version) but its not as impossible as some have suggested. Ben Wagoner would be the best person to comment on this.

Because its never 100% right , you need a bit of averaging of the target vectors. Consistently tracking a single pixel in a moving image is practically impossible. There are always going to be times when the track loses lock because the relative differences between features/pixels is beneath the threshold of discernability.

I see what you mean. You talk about several moving objects moving independently all in need of being tracked independently. I was talking about a simple case like one rigid object moving and nothing else (such as a rigid landscape in winter with no moving subparts) where you can get the motion right down to subpixel accuracy, everywhere.

I see what you mean. You talk about several moving objects moving independently all in need of being tracked independently. I was talking about a simple case like one rigid object moving and nothing else (such as a rigid landscape in winter with no moving subparts) where you can get the motion right down to subpixel accuracy, everywhere.


Tracking features and applying resulting sub-pixel transforms is trivial these days. (assuming you can get an accurate enough track in the first place).

I was referring to optical flow vector field generation where a vector is derived for ideally each pixel and is then used to drive a warp for things like tracking and retiming and deinterlacing in some cases. I used a segmented motion driven degrain , however it also generates artifacts. Most of the time this sort of technology does create artifacts but its just a question of whether the artifacts are visually acceptable as natural or not.

Do you really think you can tell the difference between artificial grain and real grain?

Yes.

I want to recreate a theater experience in my home aka. hometheater

Grain is an important aspect in that area and it sure feels good to watch that on my screen. :)

Im maybe more fanatical then some others but I even want this.

http://www.avsforum.com/avs-vb/showthread.php?t=833355

Im maybe more fanatical then some others but I even want this.

http://www.avsforum.com/avs-vb/showthread.php?t=833355

If you are serious then yes you are.

Art

I know film grain can be cleaned up and see hidden details come out because I have seen demos of this signal processing technology, so while film isn't surveillance video it is still much the same signal processing that work.
.


Hmm I've just been reading some of your comments and you are mistaken. With film , grain is the image.

You cannot degrain a film image without introducing significant softening unless its essentially a static image ( as essentially you average each frame with every other frame to remove detail that varies on a frame by frame basis) . You can alter the grain structure using manually intensive techniques but even the best automated systems produce imagery that is not really regarded as acceptable for viewing on its own.

If anything grain actually makes things look sharper and disguises posterisation.

I think you've been watching too much CSI.

If you are serious then yes you are.

Art


Its only metadata, you could always turn it of ;)

The Fountain looks bad because it wasn't shot very well. Crumby film of the highest order a real disapointment from Aronofski.


That's just kind of hard for me to buy, because it wasn't milky in the theater, the trailer on the same disk isn't even that flat.

Yes, I actually like :o The Fountain, though I can see why many could see it as a laborious exercise in philosophical bullshit trying to masquerade as something more artful and deep than the sum of it's parts.

I want to recreate a theater experience in my home aka. hometheater

Grain is an important aspect in that area and it sure feels good to watch that on my screen. :)

Im maybe more fanatical then some others but I even want this.

http://www.avsforum.com/avs-vb/showthread.php?t=833355

I think the opposite would be useful, give the grain-haters a button to posterize it out. In photoshop it's called "median" ;)

What nonsense is this? Are you disputing that grain is not a film byproduct?

Where did I say no director wants it? "It's there not because directors wanted them to be there" certainly does not mean directors do not want it. It is there no matter what directors want or not, some directors may want them, some may not.

As I said, the density and visibility of film grain is something very much under a director's control.

Where did I say it's a flaw? Let me repeat one more time, it's a byproduct, byproduct does not mean flaw, dictionary is your friend here.

What was the point of your post, then?

For someone who was caught lying just to win an argument, you need (1) learn how to read; (2) don't put words into other people's mouth. Talk

:confused:

Excuse me? Care to elaborate on that? That's an awfully big accusation to drop without bothering to explain what you're talking about.

...
Excuse me? Care to elaborate on that? That's an awfully big accusation to drop without bothering to explain what you're talking about.
Here is a hint: Chronos
PS I am still waiting for the apology!

What was the point of your post, then?
Here is another hint, don't put your words into other's mouth.

Here is a hint: Chronos
PS I am still waiting for the apology!

You're going to have to refresh my memory more than that. I have a vague recollection of some pointless bickering over that title, but you'll have to forgive me for not memorizing every argument that happens on this forum.

You're going to have to refresh my memory more than that. I have a vague recollection of some pointless bickering over that title, but you'll have to forgive me for not memorizing every argument that happens on this forum.
Well, I guess you don't. Anyway, no need to stir up the past, I thought my memory is bad:)

However, I do have an advice for you, just read twice before you reply someone who seems to have a different opinion than yours, that would avoid a lot of pointless debate.

As I said in another post, I am neutral on grains. I don't like it when it's too excessive, but if it is really what a director intends, then it should stay.

I like multi-grain Cheerios. Grain IS a plus in cereal!

What an elitist final summation. Watch it and enjoy it the way you do or be a schmuck evidently. :rolleyes:


It's not elitist, but basically yes. The goal is REPRODUCTION. The same instinct that has people adding EE, removing grain like crazy, etc, is the same ignorant instinct that has people not care about color accuracy, calibration, greyscale etc, and prefer instead to adjust their display to however they personally like it. Those people do not care about quality images, and I simply discard their opinions entirely when it comes to imaging.

As I said, the density and visibility of film grain is something very much under a director's control.


.

You know I understand the point you make but film making is nowhere near as controlled a process as I suspect a lot of you guys think it is. The entire process could be described as a happy accident.

This is 2007. What I suggest ( and this isn't aimed at Josh) is you go out take all these great new easily accessible technologies that are available and make your own films. It should be a piece of cake these days.

Then we'll take a look at what you came up with and if all we are concerned about is the level of grain occasionally on your photography...I'll cut my nuts off. How would that be?

Tracking features and applying resulting sub-pixel transforms is trivial these days. (assuming you can get an accurate enough track in the first place).
I was referring to optical flow vector field generation where a vector is derived for ideally each pixel and is then used to drive a warp for things like tracking and retiming and deinterlacing in some cases. .
If you track the whole picture which is one rigid object you have by definition an optical flow vector field for every pixel in the image with subpixel accuracy. Problems arise when you have several objects moving independently, non rigid objects, objects with no surface (smoke etc.), multiple objects interacting (occlusion) and of course image noise complicating finding maxima in matching funtions. It can get so complex every software breaks down and produces nonsense. I'm not sure we will ever reach a point where human intervention is not needed any more no matter what image content is. It would require an AI approach where the software is actually aware of image content in detail and how it relates to physical reality. And even then, how do you teach sofware that this shape in the frame is a movie screen onto which an old film is projected which is supposed to have scratches while everything outside the shape is not? :D

Do you really think you can tell the difference between artificial grain and real grain?
Maybe not. But I'm quite sure I can tell the difference between well compressed original footage with original grain the footage with the grain removed and readded. Not so much because of the added grain but the cleaned image the grain has been added to.
Image with natural grain = image with grain removed and added again only IF image with grain removed = image with natural grain minus natural grain. But grain reduced image with natural grain = image with (no/less) grain plus grain reduction artifacts! So what you get in the end is image with no/less grain plus artificial grain plus grain reduction artifacts. Whether the artificial grain is able to mask the grain reduction artifacts effectively so they are invisible I doubt very much in the general case.

Hmm I've just been reading some of your comments and you are mistaken. With film , grain is the image.
You cannot degrain a film image without introducing significant softening unless its essentially a static image ( as essentially you average each frame with every other frame to remove detail that varies on a frame by frame basis) .

Static image or one where you can get accurate motion estimation for very pixel. As long as you are tracking the same surface area for every pixel over multiple frames you can remove noise and keep detail. Problem is you usually can't do that reliably due to many issues. So you resort to heuristics giving good results on average, and lots of manual intervention.

With film , grain is the image.Exactly, and since a characteristic of film grain is that its random, it's impossible to remove it without discarding the embedded (image) detail.

I sounds like we are coming to a consensus:

1) Film is grain or visa versa.

2) Even grain lovers don't love it per se but only the detrimental effects of it's removal.

3) To date there is no reliable effective technique in use for film transfers for grain removal that won't concomitantly reduce high frequency information.


Art

IMHO there are many fancy heuristics for grain removal but in the final result most simply decide they do not trust any high frequency information and thus end up mostly removing those high frequencies, including texture detail. The exceptions to this rule will guess at the right answer and allow a bit more detail through but also create visible artifacts in the tough cases.

Grain on film creates random noise at frequencies that both can and cannot be represented at 1080p. There is no doubt in my own mind that the higher frequencies above 1080p should be removed by using a precise high rez scan, filtering, and then downsampling to 1080p. Whether any frequencies below the Nyquist frequencies for 1080p also should be removed will depend upon the source material vs the codec efficiency and your available bit budgets.

Adding grain back in should probably be reserved for upscaling during display time if this is being done for any reason (say 720p downloaded movies).

Because of all this I think there will gradually be a shift to producing lower grain movies with the idea of nicer looking electronic delivery. For instance if the director specified parameterized random grain/noise to be added during display time and the technology supported passing on this information then that choice would indeed document the directors intent yet would be practically free as far as bit budgets. It would become just another tool in the artists palette.

- Tom

If you track the whole picture which is one rigid object you have by definition an optical flow vector field for every pixel in the image with subpixel accuracy. Problems arise when you have several objects moving independently, non rigid objects, objects with no surface (smoke etc.), multiple objects interacting (occlusion) and of course image noise complicating finding maxima in matching funtions. It can get so complex every software breaks down and produces nonsense. I'm not sure we will ever reach a point where human intervention is not needed any more no matter what image content is. It would require an AI approach where the software is actually aware of image content in detail and how it relates to physical reality. And even then, how do you teach sofware that this shape in the frame is a movie screen onto which an old film is projected which is supposed to have scratches while everything outside the shape is not? :D

You're getting perilously close to a bit of egg-sucking tuition Michel.;)
You get hardly any imagery where a single tracking point will give you a useful vector for the whole plate. You'll get x,y information for one point and any points that are coplanar to that point as long as there is no rotation , scaling or perspective shifts. You can normally isolate sufficient transforms for all the translations mentioned with a 4 point track assuming those points are co-planar. To solve a fully 3d camera for a given scene you need a miimum of 6 consistent trackable features.

The most sophisticated trackers attempt to track every single recognisable feature until the feature becomes untrackable or no longer makes mathematical sense with the global model that the software is calculating as the track progresses (Bouojou although thats primarily for resolving 3d cameras it relies on multiple image tracks).

These trackers can produce very good results for two reasons : they are automated , they allow you to resolve planar tracks for areas of the plate even if they move on and off screen. The downside is that because you have so many tracks if the actual solve is incorrect you can't really manually adjust them . They don't work on nodal camera moves all that well as they rely on parralax to solve the 3d model but nodal camera moves by their nature tend to be easy to track using point tracks and a vector field itself doesn't care if the camera is moving in 3d space or not as its just tracking image features.

Like any other tracker if the features are inconsistent ( motionblur for example) they will fail , again though most heavily motionblurred material doesn't require pixel perfect tracking for most situations.

But once you've got a track what do you do with it. For a degrain all you really do is use the track to stabilise the footage around a single frame either by applying a simple transform to the whole plate or by segmenting the frame and warping individual sections according to the vector field (optical flow techniques). Then the warped frames are averaged then unwarped back to correspond with the original footage after the averaging process.

Thats as sophisticated as a degrain process gets all the other techniques are just variations on filtering tricks.

I sounds like we are coming to a consensus:

1) Film is grain or visa versa.

2) Even grain lovers don't love it per se but only the detrimental effects of it's removal.

3) To date there is no reliable effective technique in use for film transfers for grain removal that won't concomitantly reduce high frequency information.


Art


I don't think we've reached a consensus on any of these points. What I think has been missed on both sides of the argument is that since Video cannot reproduce the Film experience exactly, so how to deal with grain on video should be left to the ARTIST. If they want to bump detail on video and reduce grain then that should be their prerogative. Of if they wish to remove and simulate or allow consumer choice -- again their prerogative.

Making flat assertions that removing grain is ALWAYS detrimental is absurd. The level of technical challenge varies. What to do when there is lots of grain, but its size is sub 1080-pixel? If you felt that type of grain was important for the experience then you would be forced to simulate it to a degree. Really fine grain tends to totally disappear; though you could add just enough noise artificially to give the same perception. Again it should be the Artist/Director/Producer that makes the call. As others have hinted at, I think you would be surprised how many go with making the image as sharp as possible. Movies in theaters get their punch from size, video from clarity. If the Artist is fine with simulated grain or removing grain, why should you carp? Granted some older movies may get a sharpening treatment outside of the control of the original creators and that might not be ideal, but in general I think tweaking the grain is acceptable for many reasons including, but not limited to: sharpening the image, saving disc space, adjusting the grain look to be more compatible with video presentation, artistic choice. Smoke and haze are also elements that sometimes need tweaking for video. What looks like Smoke in the movies will often look like fog on video.

Making flat assertions that removing grain is ALWAYS detrimental is absurd. The level of technical challenge varies. What to do when there is lots of grain, but its size is sub 1080-pixel? .

Exactly the same thing you do with a 2k film scan , the sample rate resolves subpixel detail as a pixel. That comment just shows how little you actually understand about this subject.




Smoke and haze are also elements that sometimes need tweaking for video. What looks like Smoke in the movies will often look like fog on video.

And how pray tell are you going to tweak that ?
And what do you mean by "video" ?


:rolleyes:

Originally Posted by Mr.D
Hmm I've just been reading some of your comments and you are mistaken. With film , grain is the image.
You cannot degrain a film image without introducing significant softening unless its essentially a static image ( as essentially you average each frame with every other frame to remove detail that varies on a frame by frame basis) .

I would not expect much detail in a moving image anyway with the 1/48 sec. exposure in typical movie cameras.

Dave

I don't think even for a single moment in time we should consider any one thing THE image. Any modern director these days is probably cognizant of the transformations that occur in the entire chain from the original scene through negatives, post production, theater projection, and yes even HD home viewing.

I am neither a director or an artist but I imagine the good ones have intents reaching all the way to the effects on the eyes, ears, and minds of all their sets of potential viewers. I just cannot imagine any competent director saying he was satisfied because he came up with a highly satisfying set of film negatives even if the final results all sucked.

As far as I am concerned, creating art and entertainment for humans is measured by the effects on those target humans, not encapsulated by any results stored in some intermediate step toward this goal.

Though of course Hollywood can more simply measure it in profit dollars. But even there it entails calculating the effects of offering some good to the minds and wallets of consumers of all types.

- Tom

Originally Posted by Mr.D


I would not expect much detail in a moving image anyway with the 1/48 sec. exposure in typical movie cameras.

Dave


Depends on what the movement actually is and won't have any bearing on the grain visibility anyway.

Mr D,

Yeah, but then removing the sharp grain can't hurt an already soft image much.

Dave

OK I'll take one last stab at this for you Grain purists. One frame scanned at 1080p takes 1080x1960x3 bytes of information to encode (double that if you want deep color) or 152409600 bytes per frame. 24x60x130 for typical movie run time for a grand total of 28,531,077,120,000 bytes. Or 28 Terabytes of information. To compress this to 20-30GB requires a lot of tricks. Smooth images compress more easily because the information is redundant. I won't go into the math of fast Fourier transform cosines, but the "detail" you think is being thrown out with the grain is definitely being thrown out if you leave the original grain in, this because only the most significant digits of the matrix transform are retained. With grain (noise) left in, all the rows have information, thus instead of truncating insignificant rows and retaining more information in the areas where high-bandwidth info would help resolve detail, all the rows are truncated to the same depth since there is no way for the transform to discern detail from noise.

Exactly, and since a characteristic of film grain is that its random, it's impossible to remove it without discarding the embedded (image) detail.

You may think subtracting out the grain is removing detail, but you fail to realize codecs are already removing detail and grain exacerbates this. It is a trade off. You cannot have a full faithful reproduction of the grain without sacrificing detail elsewhere. For those that think grain is massively different than the thermal image noise in deep space probes you are wrong -- it can be reduced or eliminated without sacrificing apparent detail. Have you ever seen before and after images from deep space probes? How dare those NASA technicians remove thermal noise, because by necessity some original information must also be being removed. Some claim the grain is the image, but that is just a rhetorical trick -- grain is the noise artifact of the capture process just as with our deep space probe thermal image noise.

I remember when DVDs came out some Laser-Disk purists asserting digital compression could never achieve an image equivilent to the original analog, because you were throwing information away in the digital transform. Some early DVD where indeed poor compared to Laser Disc, but that phase only lasted about a year before DVDs began to far exceed Laser Disc. To be fair many of those early DVDs were likely single layer and overly compressed, but motion estimation and other optimizations improved also.

The grain you see in your final result is already a simulation of the original, not an original, which as mentioned above would take close to 30TB-60TB to be sure we don't lose anything. If grain didn't drain HD encoding resources they wouldn't be doing workarounds to remove it and then add it back in. Again, the professionals do these things for a reason, not because the want to piss Timothy Ramzyk who wants his "original grain".

Exactly the same thing you do with a 2k film scan , the sample rate resolves subpixel detail as a pixel. That comment just shows how little you actually understand about this subject.
:rolleyes:

Lets assume in your 2k scan you have a bright pixel of noise next to a dark pixel of noise, when averaged for 1080 it becomes the average -- you've just lost those two dots of grain. The finer the grain the more likely this averaging process will just remove most of the grain. It is unlikely normal picture grain disappears completely, but the averaging process of creating a video reproduction of film grain by definition is destroying the original grain and it is a non-linear process with fine grain and heavy grains being reproduced to different degrees than they appeared in the theater. There are many complaints about the video noise that the grain in the film 300 creates -- these complaints were largely not heard in the theater. So again I say leave it to the professionals, they have reasons for dealing with grain in various ways, some artistic, some technical. But to just DEMAND they leave in all the original grain, all other considerations be damned, is naïve and arrogant.

:p

This is rediculous. I'm not taking about digital captures from space probes but ordinary 35mm film. The image on that film is captured with silver halide crystals (grain) depending on the processding and film speed these are more or less visible but always present.

I'm flabergasted when Robert Harris is quoted and it is just ignored as if the foremost film restoration expert is not up to a few of you guys.:rolleyes:

So also saying that those of you including me who want grain like it and for itself ? I've seen examples like Citizen Kane which show the evils of grain removal saying that technically it could be done is like saying technically there could be a 100 mpg car. It's possible but are we going top see it implemented for consumers?

This whole debate is hopeless if some of you saying that technology exists over and over knowing that it won't be used in HDM transfers is just mental masterbation.


Leave the grain unless it can be removed without image detriment you disagree with this?

Chemically grain is film I just don't get where the argument is here.


Art

The argument is the same argument that people put forward for full screen transfers, dubbed foreign soundtracks, mono soundtracks converted to 5.1, colorization of B&W films and CGI'ing new backgrounds and characters into older films. It is the argument of technology over art. It is the argument of those who love the capability of the gadgets they buy more than the films they watch. It is the argument of those who feel they know better than the artist what they want to see and, as if that is not enough of a travesty, they think they know better than the artist what everyone else should see. In short, it is the argument of the Philistine, dressed up with techno-babble, along with much more than a fair share of pure BS.

...
I can't use my XA2 on the projector, because it's pre-HDMI

As an aside to this discussion, my 1997 NEC1351 if definitely pre-HDMI and doesn't do component either, but for about $40 bucks you can buy a Component to RGB adapter on ebay. You may need a BNC VGA cable as well for $5. But my PS3 drives the NEC just fine in 1080I and 1080P from RGB converted component. To my surprise the PS3 will do 1080P in component and converted RGB. You might want to make sure you buy and adapter rated for 1080P, many/most will smear a little if only rated for 1080I, but 1080I is all you need...

Lets assume in your 2k scan you have a bright pixel of noise next to a dark pixel of noise, when averaged for 1080 it becomes the average -- you've just lost those two dots of grain. The finer the grain the more likely this averaging process will just remove most of the grain. It is unlikely normal picture grain disappears completely, but the averaging process of creating a video reproduction of film grain by definition is destroying the original grain and it is a non-linear process with fine grain and heavy grains being reproduced to different degrees than they appeared in the theater.
:p


Lets not assume anything. I know what 2k film scans look like and I know how film grain resolves on a 2k scan.

Mr D,

Yeah, but then removing the sharp grain can't hurt an already soft image much.

Dave


No but if you are advocating motion estimated grain removal (which does actually happen with frame segmenting compression schemes anyway) the sort of imagery where you tend to notice film grain isn't usually moving enough to produce large scale motionblur.

Grain removal does not just soften an image it produces artifacts .

You're getting perilously close to a bit of egg-sucking tuition Michel.;)
You get hardly any imagery where a single tracking point will give you a useful vector for the whole plate. You'll get x,y information for one point and any points that are coplanar to that point as long as there is no rotation , scaling or perspective shifts. You can normally isolate sufficient transforms for all the translations mentioned with a 4 point track assuming those points are co-planar. To solve a fully 3d camera for a given scene you need a miimum of 6 consistent trackable features.

I never suggested you should track ONE point. I said when you track ONE rigid object which is identical to the whole image you can get everywhere correct motion estimation with subpixel accuracy and an optical flow field with flow vectors for every pixel. Naturally you track far more than one or 6 points, hundreds or more, so the resulting equations are massively overdefined and you can push the error from noise close to zero when minimising the target function.

The most sophisticated trackers attempt to track every single recognisable feature until the feature becomes untrackable or no longer makes mathematical sense with the global model that the software is calculating as the track progresses (Bouojou although thats primarily for resolving 3d cameras it relies on multiple image tracks).

Yup. I have written my own software in the past tracking objects: Matrices, target functions and solving overdefined equation systems for minimal error. But that was long ago. ;)

But once you've got a track what do you do with it. For a degrain all you really do is use the track to stabilise the footage around a single frame either by applying a simple transform to the whole plate or by segmenting the frame and warping individual sections according to the vector field (optical flow techniques). Then the warped frames are averaged then unwarped back to correspond with the original footage after the averaging process.

Something like that. If you can track the same surface point over several frames and know about lighting changes etc. you can in theory effectively remove random noise. Averaging or whatever the noise type requires. That is what Lowry DTS is supposed to do. If and how well we'll see when their HD stuff gets released on high quality discs.

http://www.thefoundry.co.uk/index.aspx

Is probably the best vector based degrain tool . The denoise tool is the one to use , the degrain tool is just a filtering based degrain.

Grainkiller is without a doubt the best filter based degrain tool I've seen :

http://www.twistedtreephoto.com/Article%20pages/grain%20surgery.html

ironically its useless at temporal averaging.

None of these tools work in realtime ( furnace is actually quite slow for 2k material) and they all produce their own artifacts that usually require manual intervention.

OK I'll take one last stab at this for you Grain purists. One frame scanned at 1080p takes 1080x1960x3 bytes of information to encode (double that if you want deep color) or 152409600 bytes per frame. 24x60x130 for typical movie run time for a grand total of 28,531,077,120,000 bytes. Or 28 Terabytes of information. To compress this to 20-30GB requires a lot of tricks.

First trick is not to multiply by 24 twice.

Grainkiller is without a doubt the best filter based degrain tool I've seen :
http://www.twistedtreephoto.com/Article%20pages/grain%20surgery.html.
Do you know what these local filtering based tools are actually doing? What principles are employed? How do they split pixels into the image and the grain part?

OK I'll take one last stab at this for you Grain purists. One frame scanned at 1080p takes 1080x1960x3 bytes of information to encode (double that if you want deep color) or 152409600 bytes per frame.
I thought both HD-DVD and Blu-ray had a pixel resolution of 1920x1080
(and the 'actual' pixel resolution will be less if the aspect ratio is >1.78:1 (ie. black bars encoded at top and bottom).

You may think subtracting out the grain is removing detail, but you fail to realize codecs are already removing detail and grain exacerbates this. It is a trade off. You cannot have a full faithful reproduction of the grain without sacrificing detail elsewhere. I’m factory trained to service the following equipment, so if you think I don’t comprehend DCT or the obligatory artifacts of COMPRESSION you are mistaken;)

Panasonic AJ-HD3700H
Panasonic AJ-HDC27
Sony HDW-F500
Sony HDW-F900

This debate has quickly degenerated into one of technology vs. art, which is a pointless argument that the technology lovers will never understand and can never win. Film grain is equivalent to the brush strokes on a painter's canvas. Any true art lover would rather see the Mona Lisa in person to appreciate every brush stroke, rather than a flat Xerox copy of it that smooths those brush strokes away. If you can't visit it in person, the goal is to find a reproduction that captures as much of that detail as possible, certainly not one that deliberately eradicates it. The notion is offensive.

Hollywood can only exist because of a partnership between those providing art and technology. It is the same thing that allowed movies to compete so successfully with Broadway plays.

No media of any form is really the artists idea or intent. Instead it is a way to communicate ideas to others, to deliver them. As new improved media technology becomes used we become better able to share ideas, to deliver art.

Art is an effective packaging of ideas. As tools get better artists will use them and we will all be better for it.

- Tom

Hollywood can only exist because of a partnership between those providing art and technology. It is the same thing that allowed movies to compete so successfully with Broadway plays.

No media of any form is really the artists idea or intent. Instead it is a way to communicate ideas to others, to deliver them. As new improved media technology becomes used we become better able to share ideas, to deliver art.

Art is an effective packaging of ideas. As tools get better artists will use them and we will all be better for it.

You still miss the point.

Paintings wouldn't exist if someone hadn't invented the paint brush. Over the years, paintbrush technology has improved to offer a wide variety of sizes and bristle types. These days, we have photographs and computer graphic technology that can capture an image much "cleaner" and more technically proficient than a paintbrush ever could.

Nonetheless, painters are still out there painting. Some of them use brushes with thick bristles to make their strokes more evident, some of them use brushes with fine bristles to make their strokes less evident, some of them simply splatter paint on the canvas willy nilly. Other artists use computers so that you never see paint or brush strokes at all, because that's what they like.

Technology is not art. Technology is a tool that the artist uses to realize his vision, whatever that should be. There is no one uniform definition of what art is, nor should there be.

Do you know what these local filtering based tools are actually doing? What principles are employed? How do they split pixels into the image and the grain part?


My understanding is that grain surgery isolates grain in terms of frequency characteristics but it matches the frequency response of a particular image against a library of real grain models to better isolate the grain. It can use temporal averaging but produces more artifacts than the furnace denoise in this mode. The great thing about it is you usually don't have to tune it too much and it seems to retain high frequency details quite well for a filter degrain.

However its still soft but you can usually get away with a bit of sharpening followed by a regrain as it retains enough detail to fool the eye. Its still nothing approaching perfect.

It was bought out by adobe a few years ago and is no longer available for Shake ( my particular package at the moment) but you can get it for AE and PS and it comes as standard with Mokey ( interesting software but in practice as useful as tits on a boar).

However its still soft but you can usually get away with a bit of sharpening followed by a regrain as it retains enough detail to fool the eye. Its still nothing approaching perfect.

Are you aware of any HD discs where a master was used that was degrained with this tool?

Are you aware of any HD discs where a master was used that was degrained with this tool?

Its not really used for that : render times would probably be prohibitive. Its really for taking grain off for other processes rather than managing end grain levels.

The argument is the same argument that people put forward for full screen transfers, dubbed foreign soundtracks, mono soundtracks converted to 5.1, colorization of B&W films and CGI'ing new backgrounds and characters into older films. It is the argument of technology over art. It is the argument of those who love the capability of the gadgets they buy more than the films they watch. It is the argument of those who feel they know better than the artist what they want to see and, as if that is not enough of a travesty, they think they know better than the artist what everyone else should see. In short, it is the argument of the Philistine, dressed up with techno-babble, along with much more than a fair share of pure BS.

:D !

Art

To be fair though, when the director uses a HDCAM to capture the video and then go back to artificially add fake grain to make it 'look' like film, I have yet seen a case where it looks natural.

In that case, I think the artist has lost some sense of aethestics and gone for the use of technology the wrong way. But that is still his right as he alone decides what he wants people to see.

To be fair though, when the director uses a HDCAM to capture the video and then go back to artificially add fake grain to make it 'look' like film, I have yet seen a case where it looks natural.

In that case, I think the artist has lost some sense of aethestics and gone for the use of technology the wrong way. But that is still his right as he alone decides what he wants people to see.

Yeah, we all know George Lucas is an ass when it comes to art. :D He and Spielberg should be locked in a room and made to watch the Southpark Indiana Jones episode on a month long tape loop, then brought outside and shot by a firing squad armed with CGI walkie-talkies.

Yeah, we all know George Lucas is an ass when it comes to art. :D He and Spielberg should be locked in a room and made to watch the Southpark Indiana Jones episode on a month long tape loop, then brought outside and shot by a firing squad armed with CGI walkie-talkies.

Right and Wrong are relative here, and not meant to be prejorative. Aesthetics certainly is a subjective thing. I like film and I also like video, when they are mixed and artists try to cover up the difference, it can get problematic, purely from a subjective personal aesthetics viewpoint. :)

I love grain, especially when it's obviously noise from the telecine process. The more digital noise the better, cause it reminds me of 800 speed 35mm film when I blow it up 100 times magnification. It's an artifact I not only can live with, but could not live without. Edge enhancement artifacts that make halos more than 10 pixels away from their source are pretty neato too.

Kidding here folks ;)

This, lossless and BD vs HD DVD are lose-lose discussions. Out of curiosity aren't theater films about 4th generation? Do you want to see grain upon grain upon grain upon grain HDM to be faithful to what you saw at the movies?

This, lossless and BD vs HD DVD are lose-lose discussions. Out of curiosity aren't theater films about 4th generation? Do you want to see grain upon grain upon grain upon grain HDM to be faithful to what you saw at the movies?

I never saw 90% of the films I buy in a theater, even at 40 I'm too young. So it boils down what's the best film element at your disposal and how can you bring it to our eyes in a way that is faithful to that element when it was new.

Yeah, we all know George Lucas is an ass when it comes to art. :D

I know he is, not sure about you.

I never saw 90% of the films I buy in a theater, even at 40 I'm too young. So it boils down what's the best film element at your disposal and how can you bring it to our eyes in a way that is faithful to that element when it was new.


Intermediate stocks are low grain : its not quite as simple as grain upon grain upon grain when the grain size is smaller than the grain in a subsequent stage.

The grain to be honest isn't really the problem , softening is which is very much mitigated these days by DI and of course the stages that cause softening don't really apply to a video master.

This, lossless and BD vs HD DVD are lose-lose discussions. Out of curiosity aren't theater films about 4th generation? Do you want to see grain upon grain upon grain upon grain HDM to be faithful to what you saw at the movies?
The reference is a print from the camera negative...

The reference is a print from the camera negative...

I'd agree with that.