or Connect
AVS › AVS Forum › Blu-ray & HD DVD › HDTV Software Media Discussion › The Digital Bits: grain is not a defect on the disc!
New Posts  All Forums:Forum Nav:

The Digital Bits: grain is not a defect on the disc! - Page 8

post #211 of 338
Quote:
Originally Posted by bjmarchini View Post

And that is the real debate then isn't it. Is the 4K scanned image distinguishable from the 4K raw image using a digital camera? And if so at what angle?

I understand that you can in theory get a better transfer scanning film... at the moment, but would this difference be perceivable at 25 degrees, 30 degrees, 35 degrees, 40 degrees?

Is this increased better transfer worth introducing noticeable image degradation with the introduction of scanned screen grain if the difference in resolution and clarity is not a perceivable difference from a 25-35 degree angle?

It is a trade off in the end. The question is which poison do you prefer?

10bit log film scan up against RAW...okay you go do the research . See what one is bigger.
post #212 of 338
The big difference.

Electronic capture of light vs Chemical capture of light.
post #213 of 338
Quote:
Originally Posted by Mr.D View Post

Okay that's it for me . I 'm not wasting any more time trying to explain these issues to someone who can't even be bothered to make a stab at understanding the fundamentals.

I apologize for my apparrent lack of knowledge. I am sorry for wasting your time. You are right, screen grain does make the picture look better. Film will always be better than digital capture.

And I appreciate you insulting me with your statement, and appreciate being put back in my place.
post #214 of 338
Quote:
Originally Posted by MovieSwede View Post

The big difference.

Electronic capture of light vs Chemical capture of light.

There are many people that would argue that the former is getting to the point of being indistinguishable from the latter in real world use. My photographer friend at the paper says that for them, it was when they hit 8 megapixels. Of course, those are stills and not motion scenes. He also freelances alot for a number of magazines which need higher resolution pictures.... still uses an his digital for those as well. He doesn't know anyone in the media that still uses film. Of course his CCD is really good in his nikon. He has a D200 I think, but he has been talking about this D3 he was thinking of picking up.

From what I remember, they switched in 2003 I think to all digital. At first he was really reluctant to use the digital, but it so much easier for him to work with and develop. He told me another dark secret of the film is all the chemicals that are used in it. He jokes that the EPA only knew what alot of photographers do....
post #215 of 338
Quote:
Originally Posted by bjmarchini View Post

There are many people that would argue that the former is getting to the point of being indistinguishable from the latter in real world use. My photographer friend at the paper says that for them, it was when they hit 8 megapixels.

I think they are saying that 35mm motion picture film is usually better at capturing brightness ranges than a digital HD video camera.

eg. with a HD video camera the sky could look too white (bleached out?) whereas with film it should be able to go up more gradually and capture more brightness ranges (more similar to a how the eye sees?).
post #216 of 338
Quote:
Originally Posted by bjmarchini View Post

He told me another dark secret of the film is all the chemicals that are used in it.

Really !

Art
post #217 of 338
Quote:
Originally Posted by bjmarchini View Post

There are many people that would argue that the former is getting to the point of being indistinguishable from the latter in real world use. My photographer friend at the paper says that for them, it was when they hit 8 megapixels. Of course, those are stills and not motion scenes. He also freelances alot for a number of magazines which need higher resolution pictures.... still uses an his digital for those as well. He doesn't know anyone in the media that still uses film. Of course his CCD is really good in his nikon. He has a D200 I think, but he has been talking about this D3 he was thinking of picking up.

From what I remember, they switched in 2003 I think to all digital. At first he was really reluctant to use the digital, but it so much easier for him to work with and develop. He told me another dark secret of the film is all the chemicals that are used in it. He jokes that the EPA only knew what alot of photographers do....

No question that digital will replace film sooner or later, but its not just like you need a 8 megapixels and everything is just fine.

So give me 2K camera with 35mm size sensor, with global shutters and some form of RAW-video format with latitude that rivals film that also can shoot 18-250fps. And I would be very happy.
post #218 of 338
For people who want to refresh/upgrade their technical knowledge of digital Camera; Panavision have a 7 part masterclass video on their web site.

Quote:


A master class focused on understanding picture quality in the Digital Age, and the scientific concepts underlying the performance of modern digital cameras

An in-depth discussion on the science behind today’s digital cinema cameras, and its affect on recording motion picture images.

On April 3, John Galt, Panavision Senior Vice President, Advanced Digital Imaging, and Larry Thorpe, National Marketing Manager, Canon Broadcast & Communication Division, presented their views on a realistic, scientifically valid way to gauge the quality of recorded pictures that belies the theory that “it’s all about the pixels.” The sessions were held at Panavision’s world headquarters in Woodland Hills, and taped in high definition video with three cameras before a live audience of cinematographers, and other craftsmen, studio executives and the media.




Quote:


Part 1 discusses the following topics: A short history of CCD/CMOS development, Sub-sampling and Super-sampling, HD lens and camera design strategies, Photosites - the tradeoff between resolution, dynamic range, and noise, Digital output signals from cameras, UHDTV, 2K and 4K cameras, and DCI Standards for 2K and 4K.

Part 2 discusses the following topics: The Nyquist Theorem, linking optical and digital sampling, The implications of Fill factor and optical low pass filtering, The Nyquist boundaries of of motion pictures, and Limiting resolution.

Part 3 discusses the following topics: Introduction to MTF, Introduction to an individual element’s contribution to final MTF, Transferring contrast, Cascading MTFs, Introduction to Otto Schade, Perceived picture sharpness and MTF, Edge Sharpness, image textures, and resolving power.

Part 4 discusses the following topics: Pixels are not resolution, Practical measurements of MTF, Introduction to sinusoidal MTF charts, The Panavision sinusoidal MTF Chart, MTF benches for measuring lenses, Cascaded MTF of a DI, Sensor MTF response, Depth of Field, MTF measurements of real world lenses.

Part 5 discusses the following topics: Line pairs per Millimeter, Differences between HDTV and SDTV lens standards, a demonstration of the MTF-Aliasing dilemma using three chip cameras, Loading the imager MTF into the digital container.

Part 6 discusses the following topics: The resolution metric for a Bayer pattern sensor, Diagonal sampling Bayer pattern sensors, Optical lowpass filter options for single Bayer pattern sensors.

Part 7 discusses the following topics: Spectral response, Camera color balance: Daylight v. Tungsten, Digital Intermediate MTF comparison between Genesis and 5218, Bayer vs. RGB striped sensors.
post #219 of 338
Quote:
Originally Posted by tbrunet View Post

The presence of noise (random grain) in an image severely degrades compression efficiency. In MPEG, for example, noise degrades interframe compression by adversely impacting the performance of "motion estimation," which achieves interframe compression by cross-referencing matching blocks of pixels in neighboring frames. In motion estimation, noise interferes with the identification of such matching blocks. In addition, noise also affects intraframe compression by reducing the correlation among neighboring pixel values, thereby reducing the compression efficiency achieved by quantizing video data transformed under a discrete cosine transform.

All true but nonetheless, inappropriate to quote others without attributions: http://www.patentstorm.us/patents/60...scription.html

"Under many compression standards, e.g. MPEG, the presence of noise in the video data severely degrades compression efficiency. In MPEG, for example, noise degrades interframe compression by adversely impacting the performance of "motion estimation," which achieves interframe compression by cross-referencing matching blocks of pixels in neighboring frames. In motion estimation, noise interferes with the identification of such matching blocks. In addition, noise also affects intraframe compression by reducing the correlation among neighboring pixel values, thereby reducing the compression efficiency achieved by quantizing video data transformed under a discrete cosine transform (DCT). Under fixed quantization, i.e., a variable output bit-rate approach, the noise degradation of interframe and intraframe compressions often leads to an increased output or encoded bit-rate1 by as much as 100%. The increased encoded bit-rate results in higher transmission and storage costs. Alternatively, in a storage medium of a fixed capacity, e.g. in a compact disk (CD-ROM) or a digital video disk (DVD), such increased encoded bit-rate results in content reduction. "

I am puzzled why you quoted this if you are trying to defend film grain. It says that you don't want to have grain in the source if you can help it....
post #220 of 338
Quote:
Originally Posted by amirm View Post

I am puzzled why you quoted this if you are trying to defend film grain. It says that you don't want to have grain in the source if you can help it....

I'm NOT "defending" grain. Just illustrating the extremely complex random characteristic of film grain. Any compression codec will be challenged by the complexity, which btw is not really that difficult to grasp IMO
post #221 of 338
Quote:
Originally Posted by tbrunet View Post

I'm NOT "defending" grain. Just illustrating the extremely complex random characteristic of film grain. Any compression codec will be challenged by the complexity, which btw is not really that difficult to grasp IMO

How about addressing the fact that - once again - you have taken parts of other people's work and posted them as your own thoughts? Do you honestly see no problem with this?
post #222 of 338
Oh and btw amirm if you don't comprehend how image detail is embedded in the film grain pattern(s) maybe you could research the subject matter. Thanks in advance!
post #223 of 338
Having been caught multiple times ripping other people quotes without reference, is it really so "difficult to grasp" that nobody gives a hoot what you are saying anymore?

Diogen.
post #224 of 338
Quote:
Originally Posted by diogen View Post

.. is it really so "difficult to grasp" that nobody gives a hoot what you are saying anymore?

The big difference.

Comprehending the facts regarding temporal redundancy vs "I don't give a hoot what you are saying"
post #225 of 338
Quote:
Originally Posted by bjmarchini View Post

I love when folks look resort to rip others apart by insulting them or degrading them. Shows alot of class. I hope ripping me makes you feel better.

Nobody was even talking to, or about you. Hence the fact that you weren't quoted in any of the posts that you quoted in retaliation.
post #226 of 338
Quote:
Originally Posted by amirm View Post

All true but nonetheless, inappropriate to quote others without attributions: http://www.patentstorm.us/patents/60...scription.html

"Under many compression standards, e.g. MPEG, the presence of noise in the video data severely degrades compression efficiency. In MPEG, for example, noise degrades interframe compression by adversely impacting the performance of "motion estimation," which achieves interframe compression by cross-referencing matching blocks of pixels in neighboring frames. In motion estimation, noise interferes with the identification of such matching blocks. In addition, noise also affects intraframe compression by reducing the correlation among neighboring pixel values, thereby reducing the compression efficiency achieved by quantizing video data transformed under a discrete cosine transform (DCT). Under fixed quantization, i.e., a variable output bit-rate approach, the noise degradation of interframe and intraframe compressions often leads to an increased output or encoded bit-rate1 by as much as 100%. The increased encoded bit-rate results in higher transmission and storage costs. Alternatively, in a storage medium of a fixed capacity, e.g. in a compact disk (CD-ROM) or a digital video disk (DVD), such increased encoded bit-rate results in content reduction. "

I am puzzled why you quoted this if you are trying to defend film grain. It says that you don't want to have grain in the source if you can help it....


So if I am reading this right, film grain causes the film to take up more space and results in a higher bitrate. Seeing that the bitrate is capped on HDM, does this mean that high motion scenes that spiked over the threshold would have to caled down. Would this create motion blur simlar to when you use a lower bitrate when encoding to H264? There are some scenes in my BR disks that max out the bitrate when I have the onscreen diagnostics on in Nero 8 HD/BR.

Quote:
Originally Posted by BStecke View Post

Nobody was even talking to, or about you. Hence the fact that you weren't quoted in any of the posts that you quoted in retaliation.

I just do not like seeing posts where the only thing a person does is openly degrade the other person instead of replying with reasoning and/or facts why they believe they are in error. It is bad form and makes a discussion too personal IMO. When I disagree, I try to convey the reason I disagree without insulting someone. I think we all want to be treated that way, don't we?
post #227 of 338
Quote:
Originally Posted by bjmarchini View Post

I just do not like seeing posts where the only thing a person does is openly degrade the other person instead of replying with reasoning and/or facts why they believe they are in error. It is bad form and makes a discussion too personal IMO. When I disagree, I try to convey the reason I disagree without insulting someone. I think we all want to be treated that way, don't we?

Hey, well said!

i.e. pointing out that something is misspelled

How about trying to explain that exploiting temporal redundancies does NOT correlate to the debated subject matter of encoding film grain.
post #228 of 338
Quote:
Originally Posted by tbrunet View Post

How about trying to explain that exploiting temporal redundancies does NOT correlate to the debated subject matter of encoding film grain.

And where did you steal this sentence?

Diogen.
post #229 of 338
Quote:
Originally Posted by diogen View Post

And where did you steal this sentence?

The original logic I've posted above is all mine just like the sentence you quoted.

Do you even understand the concept of exploiting temporal redundancies and how it may correlate to the subject matter?
post #230 of 338
Quote:
Originally Posted by Mr.D View Post

Okay that's it for me . I 'm not wasting any more time trying to explain these issues to someone who can't even be bothered to make a stab at understanding the fundamentals.

I used to get into these conversations a lot here. Many were very educational and I learned from people more knowledgable from me, as others learned from me. Since bjmarchini started participating, always prefacing his entry with "What do I care about the director's vision for, I know what I like!!" and "Reality doesn't have grain" strawman type posts, I have given up.

He'll post reams and reams of nonsense, justify it by sheer misinformation, play coy when proven wrong, and finally, claim victim when you call him out on his utter lack of understanding of any aspect of film production. All this adds up to my conclusion that his mind is already set, and although he claims to be learning something; he is really just waiting for the next thread on the subject to start, upon which he reboots and starts all over again.

Personally, I'd rather argue with a screen filler.
post #231 of 338
Quote:
Originally Posted by bjmarchini View Post

So if I am reading this right, film grain causes the film to take up more space and results in a higher bitrate. Seeing that the bitrate is capped on HDM, does this mean that high motion scenes that spiked over the threshold would have to caled down. Would this create motion blur simlar to when you use a lower bitrate when encoding to H264? There are some scenes in my BR disks that max out the bitrate when I have the onscreen diagnostics on in Nero 8 HD/BR.

Noise is very difficult to encode as the reference mentioned. What that translates into though, depends on many factors from codec choice, how easy it is to see compression artifacts, etc.

"Motion blur" tends to be caused by many things, with the top reason being real-time MPEG-2 encoders performing temporal filtering when they get in trouble. So if the above experience is with people re-encoding MPEG-2 broadcast, then I would put the fault there, before going after film grain.

But yes, there is no question that if you want to deliver things to people in compressed domain, the less noise the better. Digital distribution for example, looks a lot better without grain in the source than with. In case of HD optical, it is less of a concern (and potentially none). So I would not point the finger at it quickly, but all else being equal, there is no doubt you want less noise here also. It would as a minimum, reduce encoding time/effort. And at best, allow a BD-25 to look as good as BD-50 for example.

I should also point out something counter intuitive with grain. That is, presence of it makes you think the source is sharper. If you did a perfect grain removal, with nothing but grain removed, your eye will consider the resulting image softer than the source, even if zero detail is removed from the picture itself! Indeed, some photographers rescue soft images by adding a touch of noise to them. The presence of such "high frequency" component makes the picture seem sharper.

If there is interest and I find the time, I will come up with some demonstrations of above.
post #232 of 338
Quote:
Originally Posted by tbrunet View Post

The original logic I've posted above is all mine just like the sentence you quoted.

Do you even understand the concept of exploiting temporal redundancies and how it may correlate to the subject matter?

I do, and I am sure he does as well. This is an excellent point. It is where that video repeats many of the same elements over and over and in essence wastes size in doing this which is how most compression schemes work.

Using a digital capture instead of screen grain could drastically reduce the file size need to show the same shots as these screen grains inhibit the ability of the compression software to reduce the file size.

While more compression may not seem at first to be desired as the notion of artifacts comes to mind, it would in theory allow you to reduce motion blur which is inherent to most compressed video. So by increasing your ability to compress a video you can include more detail without increase or decreasing the file size and bitrate.

So in a sense you could actually add detail in video by at the same bitrate without screen grain as this would allow the video to be more efficiently compressed.

Does that make sense?
post #233 of 338
One other thing. Presence of noise in the source reduces banding effects. Banding is quantization noise and dither removes it. Natural random noise in the source serves the same purpose as dither, making problems downstream with scanners, etc. far less of an issue. This is why you see banding on computer generated logos and effects far easier than film sourced material.

Related to above, is why I am not a fan of 8-bit distribution systems. Proper 8-bit means dither noise added to the 10-bit source to arrive at 8-bits. While this gets rid of banding, the resulting image then is harder to encode. We are much better of with 10-bit systems because of this.
post #234 of 338
Quote:
Originally Posted by amirm View Post


I should also point out something counter intuitive with grain. That is, presence of it makes you think the source is sharper. If you did a perfect grain removal, with nothing but grain removed, your eye will consider the resulting image softer than the source, even if zero detail is removed from the picture itself! Indeed, some photographers rescue soft images by adding a touch of noise to them. The presence of such "high frequency" component makes the picture seem sharper.

I do not understand how grain could make something look sharper. I could understand how removing grain could, but not a lack of it.
post #235 of 338
Quote:
Originally Posted by bjmarchini View Post

Using a digital capture instead of screen grain could drastically reduce the file size need to show the same shots as these screen grains inhibit the ability of the compression software to reduce the file size.

No one is implying that digital capture is NOT the easier paradigm. The fact is film grain contains embedded in it the original image detail. Filter it out and you lose that information.

http://ftp3.itu.ch/av-arch/jvt-site/...tes_draft3.doc
Film grain has no temporal correlation since it results from the random creation of dyes during the development of photographic film. Because of its high entropy, film grain encoding using common prediction techniques without the use of post-decoding film grain modeling synthesis can only be accomplished at very high bit-rates.”

“In order to enable some film grain presence effect at lower bit-rates, Thomson proposed to the JVT a modeling-based solution where the film grain is filtered before encoding and simulated after decoding. This is accomplished by means of the transmission of a film grain SEI message, already accepted as part of the FRExt amendment draft for H.264/AVC.”
post #236 of 338
Quote:
Originally Posted by amirm View Post

One other thing. Presence of noise in the source reduces banding effects. Banding is quantization noise and dither removes it. Natural random noise in the source serves the same purpose as dither, making problems downstream with scanners, etc. far less of an issue. This is why you see banding on computer generated logos and effects far easier than film sourced material.

Related to above, is why I am not a fan of 8-bit distribution systems. Proper 8-bit means dither noise added to the 10-bit source to arrive at 8-bits. While this gets rid of banding, the resulting image then is harder to encode. We are much better of with 10-bit systems because of this.

Very interesting point
post #237 of 338
Quote:
Originally Posted by bjmarchini View Post

I do not understand how grain could make something look sharper. I could understand how removing grain could, but not a lack of it.

Take a look at this super well writen article on digital printing. In this part of it, he shows a demonstration of grain making the image look sharper. It is not the best demo I have seen, but should be useful to convey the message.

http://www.naturescapes.net/docs/ind...etter-prints#6

BTW, I highly recommend people read the entire article: http://www.naturescapes.net/docs/ind...-better-prints
post #238 of 338
Quote:
Originally Posted by tbrunet View Post

No one is implying that that digital capture is NOT the easier paradigm. The fact is film grain contains embedded in it the original image detail. Filter it out and you lose that information.

http://ftp3.itu.ch/av-arch/jvt-site/...tes_draft3.doc
Film grain has no temporal correlation since it results from the random creation of dyes during the development of photographic film. Because of its high entropy, film grain encoding using common prediction techniques without the use of post-decoding film grain modeling synthesis can only be accomplished at very high bit-rates.

In order to enable some film grain presence effect at lower bit-rates, Thomson proposed to the JVT a modeling-based solution where the film grain is filtered before encoding and simulated after decoding. This is accomplished by means of the transmission of a film grain SEI message, already accepted as part of the FRExt amendment draft for H.264/AVC.

But when you scan it for editing, aren't you losing that detail? In essence you are still digitizing an image, but in this case you are digitizing the image of an image like copying a copy and then you are putting it back on film which would be a copy of a copy of a copy. I know this is simplifying it, but sometimes it is better keep the concept simple so it doesn't get lost in the details. Regardless, I don't think filtering it is a realistic solution anyway. If it is there, I would just leave well enough alone. You can't go back and reshoot the film so this would leave removal the only solution for older films. But for currently made films, I would rather see them shot digitally
post #239 of 338
Quote:
Originally Posted by amirm View Post

Take a look at this super well writen article on digital printing. In this part of it, he shows a demonstration of grain making the image look sharper. It is not the best demo I have seen, but should be useful to convey the message.

http://www.naturescapes.net/docs/ind...etter-prints#6

BTW, I highly recommend people read the entire article: http://www.naturescapes.net/docs/ind...-better-prints

I will have to read this article further when I have time. I see what you mean though.

I still would have liked to see a movie like Caddyshack without the screen grain though. I am not advocating going back removing it. But if they made a similar type of movie today, I would rather not see the grain as I think it distracts from it. Especially on big screens
post #240 of 338
Quote:
Originally Posted by amirm View Post


If there is interest and I find the time, I will come up with some demonstrations of above.

Isnt time something you got plenty nowdays

I would like to add that adding grain has the advantage of masking compression artifacts aswell.

Unless you working with a lossless videoformat.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: HDTV Software Media Discussion
AVS › AVS Forum › Blu-ray & HD DVD › HDTV Software Media Discussion › The Digital Bits: grain is not a defect on the disc!