Do You Like the Look of HFR? - Page 11 - AVS Forum
View Poll Results: Do You Like the Look of HFR?
Yes 166 46.50%
No 97 27.17%
I've never seen HFR 94 26.33%
Voters: 357. You may not vote on this poll

Forum Jump: 
Reply
 
Thread Tools
post #301 of 316 Old 02-12-2014, 04:07 AM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,601
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 174 Post(s)
Liked: 60
Quote:
Originally Posted by djbluemax1 View Post

If you stopped and thought about it, you'd see the flaw in the logic/premise above.

Whether it's native 48fps OR creative frame interpolation, NEITHER one actually speeds up the motion.
Interpolation could speed up motion.

eg. a car starts at point A and the next real frame it is at point B. If you want to interpolate between those two points at the time half way between the two, where do you place the car? If you assume it's in the middle - what if the car was accelerating from a stationary position - in reality it would be nearer point A - but if it wasn't accelerating and had been travelling at a constant speed between A and B, it should be placed in the middle (assuming 2D).

So that's one of the problems with interpolation - it doesn't know exactly where to place the interpolated objects - it can just estimate where to. If it gets it wrong it could be either sped up then slowed down or slowed down then sped up in comparison to how the object moved in reality.
Quote:
Besides logically speaking if anything was sped up, the movie's running time would shorten. It doesn't
The running time wouldn't change if there were slow downs that compensated for the sped-up parts.


Here's a sort of example, though it also shows other problems with interpolation:

An image from a 12 fps video:


An image from a 3 fps video interpolated to 12 fps:

As well as not interpolating very well, parts of the car at the same point in time are at different positions than their true position as shown in the 12 fps (not interpolated) video frame. So it could seem to speed up/slow down based on the interpolated position of the car in comparison to the actual non-interpolated version.

Joe Bloggs is offline  
Sponsored Links
Advertisement
 
post #302 of 316 Old 02-12-2014, 07:13 AM
Advanced Member
 
agnathra's Avatar
 
Join Date: Nov 2004
Location: Chattanooga, TN
Posts: 580
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 12 Post(s)
Liked: 17
Quote:
Originally Posted by Dan Hitchman View Post

I wonder if some of these motion artifacts are due to how current digital cinema projectors are handling the 48 fps signal from the digital files. Could some be unintentionally applying a kind of frame interpolation to the native signal rather than preserving the native 48 fps data? I ask this because 24 fps movies actually do look sped up using motion interpolation on TV sets with this feature because they are adding frames that were not there in the first place. This is different from just doubling or tripling frames in order to reduce flicker. Digital projectors, after all, are just giant TV's.

i've been thinking this all along...it's a brand new technology, and it's being implemented across the country by different theater chains with different setups. (for example, when i saw smaug in HFR it was actually letterboxed). it certainly seems possible that it might be being implemented improperly, and it might vary from theater to theater, explaining why not everyone sees the problem.

he also filmed this so that it could be shown in 24 fps or 48 fps...i don't know what processing he did (if any) to allow that, but i think if you use a higher shutter speed for 24 fps film, you get the saving private ryan effect...little or no blur in each frame. but if his shutter speed was perhaps slower than it ought to be for 48 fps so that he wouldn't have that effect at 24 fps, perhaps some time of interpolation-like effect is created at 48 fps playback.

the other reason i lean in this direction is that i've played hundreds and hundreds of hours of video games at 60 fps...no speed up. i've seen other examples of HFR 3d (including Star Tours at disney world, and that looked gorgeous)...no speed up effect.

this is a new filming approach and a weird new technology (for theater chains) and it looks too much like interpolation artifacts for me to think i just need to get used to it.
freemeat likes this.

- Chris
agnathra is offline  
post #303 of 316 Old 02-12-2014, 10:47 AM
AVS Special Member
 
Dan Hitchman's Avatar
 
Join Date: Jul 2000
Location: Northern Colorado
Posts: 9,196
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1015 Post(s)
Liked: 521
Quote:
Originally Posted by agnathra View Post

i've been thinking this all along...it's a brand new technology, and it's being implemented across the country by different theater chains with different setups. (for example, when i saw smaug in HFR it was actually letterboxed). it certainly seems possible that it might be being implemented improperly, and it might vary from theater to theater, explaining why not everyone sees the problem.

he also filmed this so that it could be shown in 24 fps or 48 fps...i don't know what processing he did (if any) to allow that, but i think if you use a higher shutter speed for 24 fps film, you get the saving private ryan effect...little or no blur in each frame. but if his shutter speed was perhaps slower than it ought to be for 48 fps so that he wouldn't have that effect at 24 fps, perhaps some time of interpolation-like effect is created at 48 fps playback.

the other reason i lean in this direction is that i've played hundreds and hundreds of hours of video games at 60 fps...no speed up. i've seen other examples of HFR 3d (including Star Tours at disney world, and that looked gorgeous)...no speed up effect.

this is a new filming approach and a weird new technology (for theater chains) and it looks too much like interpolation artifacts for me to think i just need to get used to it.

I think you're on to something.

Listen up, studios! Just say "NO" to DNR and EE!!
Dan Hitchman is offline  
post #304 of 316 Old 02-12-2014, 11:59 AM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,601
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 174 Post(s)
Liked: 60
Quote:
Originally Posted by agnathra View Post

he also filmed this so that it could be shown in 24 fps or 48 fps...i don't know what processing he did (if any) to allow that, but i think if you use a higher shutter speed for 24 fps film, you get the saving private ryan effect...little or no blur in each frame. but if his shutter speed was perhaps slower than it ought to be for 48 fps so that he wouldn't have that effect at 24 fps, perhaps some time of interpolation-like effect is created at 48 fps playback.
He shot it at ~48 fps with a 270 degree shutter. Normal live TV (eg. sports) will normally be shot with an open (ie. 360 degree) shutter, at 50Hz (50i) in Europe - broadcast in 720p50 in some places I think. In the US I'm sure they'll do something similar with sports at 60i / 720p60. Having an open shutter shouldn't create an interpolated look (though it will create more of a blur - and you can see more of a blur type look in my interpolated car example - though not like accurate motion blur) but shooting with a more open shutter shouldn't create the "wrong" motion that interpolation does - eg. incorrectly positioned objects/incorrect images/having a sped-up/slow down appearance because of the incorrect images.
Joe Bloggs is offline  
post #305 of 316 Old 02-14-2014, 09:55 PM
AVS Special Member
 
djbluemax1's Avatar
 
Join Date: Oct 2004
Location: MI
Posts: 2,278
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 28 Post(s)
Liked: 287
Quote:
Originally Posted by Dan Hitchman View Post

But frame interpolation and natively shot frame rates are different animals. One inserts "theoretical" frames in order to go from 24 to 30 (in U.S. TV's), which is then doubled to 60 to match NTSC standards for flicker and to match the older power supplies (or 120 Hz or 240 Hz depending on the display). It looks weird because of the funky way the software interprets the fill in frames. The Hobbit was shot at 48 fps with actual frames, not "faked" insert frames.

Now take a look at something shot on video at frame rates higher than 24, such as something on TV that isn't a "scripted" show since they're usually shot at 24 fps and then converted, and again the motion looks "normal" and doesn't have a "soap opera effect." It may not have motion judder artifacts like a film at 24 fps, but it also doesn't have that fake motion look either.

I'm not saying that interpolation is making images go faster, it just looks weird.

The Hobbit shouldn't have that strange frame interpolation look at all. If it does, something's potentially being processed wrong in the digital cinema playback chain.
Couple of things to note here: First, I think you're conflating 3;2 pulldown and frame interpolation.

The conversion from 24p to the broadcast 60Hz is called 2:3 pulldown and is not a form of motion interpolation, it's simply a way to translate 24fps into 60fps, i.e. the sequential frames ABCD are simply displayed as AABBBCCDDD.

Frame interpolation displays generally use at least 120Hz because it allows the selection of frame interpolation OR multiples of 24p (i.e. 24 x 5 = 120), so they can either display the native 24p as a multiple like 24x5 AAAAABBBBBCCCCCDDDDD for 120Hz (reducing the appearance of strobing due to the low rate) or a simple doubling of 60Hz content.

For motion interpolation though, they simply use an algorithm that analyzes the sequence of frames for moving objects and creates intermediate frames to smooth the motion eg. A,A1,A2,A3,A4,B,B1,B2,B3,B4,C,C1,C2,C3,C4 etc. The time to display all frames A through D are the same though.

Now examine your bolded paragraph above. It's amusing when you say that something shot natively on video doesn't have 'Soap Opera Effect'. You DO realize that the term SOE stems from the fact that higher frame rates (from motion interpolation) by reducing the motion judder and blur, made 24p look more like Soap Operas shot at 60Hz right? By definition, anything shot in video is an example of 'Soap Opera Effect' in comparison to the juddery, blurry mess of 24p.


Max
djbluemax1 is offline  
post #306 of 316 Old 02-14-2014, 10:13 PM
AVS Special Member
 
djbluemax1's Avatar
 
Join Date: Oct 2004
Location: MI
Posts: 2,278
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 28 Post(s)
Liked: 287
Quote:
Originally Posted by Joe Bloggs View Post

Quote:
Originally Posted by djbluemax1 View Post

If you stopped and thought about it, you'd see the flaw in the logic/premise above.

Whether it's native 48fps OR creative frame interpolation, NEITHER one actually speeds up the motion.
Interpolation could speed up motion.

eg. a car starts at point A and the next real frame it is at point B. If you want to interpolate between those two points at the time half way between the two, where do you place the car? If you assume it's in the middle - what if the car was accelerating from a stationary position - in reality it would be nearer point A - but if it wasn't accelerating and had been travelling at a constant speed between A and B, it should be placed in the middle (assuming 2D).

So that's one of the problems with interpolation - it doesn't know exactly where to place the interpolated objects - it can just estimate where to. If it gets it wrong it could be either sped up then slowed down or slowed down then sped up in comparison to how the object moved in reality.
Quote:
Besides logically speaking if anything was sped up, the movie's running time would shorten. It doesn't
The running time wouldn't change if there were slow downs that compensated for the sped-up parts.


Here's a sort of example, though it also shows other problems with interpolation:

An image from a 12 fps video:


An image from a 3 fps video interpolated to 12 fps:

As well as not interpolating very well, parts of the car at the same point in time are at different positions than their true position as shown in the 12 fps (not interpolated) video frame. So it could seem to speed up/slow down based on the interpolated position of the car in comparison to the actual non-interpolated version.

I have no idea what that animation is supposed to show, but it doesn't appear to be what happens with a good interpolation algorithm.

In your example of an accelerating car, your theory might appear sound to a layman... until you actually examine it, at which time it falls apart.

First off, motion interpolation only creates intermediate frames for movement between frames, i.e. if I go back to my example of a 100 unit wide screen, and frame A shows a car at unit 1 (going left to right), frame 2 shows the car at unit 2, frame 3 at unit 3.5, frame 4 at unit 5.5, frame 5 at unit 8 etc. the car is accelerating. A decent algorithm simply adds intermediate frames between each successive real frame i.e. if they add 4 additional frames, then the created frames between frames 1 and 2 simply show the car at unit 1.2, 1.4, 1.6, 1.8. Between frames 4 and 5, they simply add frames that place the car at unit 6, 6.5, 7, 7.5 etc.

The fact that the car is accelerating and not traveling at a constant rate is not a problem for any decent algorithm.

The other consideration is that directors are well aware of how bad the blurring and judder is with any motion in 24p, and as such, they film things so that there's only so much movement on screen (which is why you rarely see panning faster than a certain rate, because it would be so blurry it would be pointless).

To date, Sony has the best motion interpolation algorithms I've seen. They produce the least artifacts, and as processor power increases, hopefully they'll eventually be able to do per-pixel interpolation which should eliminate most visible artifacts.

Also, for those wondering about so-called 'improper implementation' or display of HFR: Do you have ANY evidence whatsoever, that ANY commercial systems have any kind of frame interpolation? It's one thing to display a higher rate, but some of you appear to be claiming that what you're seeing with HFR is possibly some kind of frame interpolation. I haven't heard ANY evidence of this.


Max
djbluemax1 is offline  
post #307 of 316 Old 02-14-2014, 10:56 PM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,601
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 174 Post(s)
Liked: 60
Quote:
Originally Posted by djbluemax1 View Post

I have no idea what that animation is supposed to show
The first video shows what the car is supposed to be doing and where it is supposed to be at 12 fps. The interpolated version (3 fps -> 12 fps) shows that interpolation puts the car at the wrong places and also that interpolation isn't precise and causes bad artefacts. If you don't see the car in the interpolated video appearing to speed up/slow down, look at the still frames of the car. The top image shows the front of the car not much after position 600. The interpolated version shows the front of the car (even though part faded) starting past position 625. The back of the car also is at the wrong position in the interpolated version (the faded part starts before the back of the car in the non-interpolated version). So you could say interpolated version of the car shows it both sped up and slowed down.
Quote:
doesn't appear to be what happens with a good interpolation algorithm.
Quote:
hopefully they'll eventually be able to do per-pixel interpolation which should eliminate most visible artifacts
The algorithm used for my example was a per-pixel algorithm "pixel motion" and a good algorithm but like all algorithms for interpolating video it will generate artefacts. It might have done better if the car had more points (detail) it could match. I interpolated from 3 fps to 12 fps. Obviously if it had more frames to work with it would have done a better job, but doing it that way showed the artefacts more clearly as well as the motion artefacts. It still had less of a ratio of frames to generate than 24->120 fps interpolation - though with the lower frame rate you see the artefacts more clearly.
Quote:
A decent algorithm simply adds intermediate frames between each successive real frame i.e. if they add 4 additional frames, then the created frames between frames 1 and 2 simply show the car at unit 1.2, 1.4, 1.6, 1.8. Between frames 4 and 5, they simply add frames that place the car at unit 6, 6.5, 7, 7.5 etc.

The fact that the car is accelerating and not traveling at a constant rate is not a problem for any decent algorithm.
You're assuming and possibly basing it off the promotional material of TVs that feature interpolation instead of basing it on evidence of what actually happens with and the results of interpolation algorithms. I've shown what actually happens with interpolation - using pixel motion - the type of interpolation you think will give the best interpolation - and shown - it causes motion artefacts - as well as can interpolate things wrong in other ways (eg. the shape of the objects, as well as ghosting type artefacts - and the ghosting is one thing that could make something appear to be at the wrong position/sped up - by placing it ahead of where it should be because it is showing part of something from a frame ahead of the current frame).
Quote:
then the created frames between frames 1 and 2 simply show the car at unit 1.2, 1.4, 1.6, 1.8
And if the true position of the car or other thing in real life wasn't 1.2, 1.4, 1.6., 1.8 at those points in time between the real frames it's still either sped up or slowed down (or both) the true motion of the car/object - even if no other artefacts (such as ghosting - placing the object ahead of it's true position) are generated.

Here's another link which shows some of the artefacts of interpolation from the Red camera site (see interpolated 24 fps example towards the bottom of the page) - though they don't say what frame rate it was interpolated from or what algorithm they used (might be just frame blending):
http://www.red.com/learn/red-101/high-frame-rate-video
There will also be youtube videos that show actual TV motion interpolation artefacts.

It proves that motion interpolation can speed up/slow down the true motion of something in video (as proved by the start/end positions of the car in the example frames, both from the same point in time, which show the time code and an X coordinate) - which you said didn't happen with frame interpolation, as well as shows the sort of artefacts it can add.
Joe Bloggs is offline  
post #308 of 316 Old 02-15-2014, 12:33 AM
AVS Special Member
 
djbluemax1's Avatar
 
Join Date: Oct 2004
Location: MI
Posts: 2,278
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 28 Post(s)
Liked: 287
Just because you can show some random video claiming to show how interpolation works doesn't make it so. I still don't see how the front of the car (in actual real world motion interpolation) would end up being further in front than the original image as the interpolated images produce motion in-between the successive frames, i.e the interpolation looks at frame A and B and creates intermediate frames where the object is between A and B, NOT looking at the positions in A and B and then putting the object in front of B.


Max
djbluemax1 is offline  
post #309 of 316 Old 02-15-2014, 12:39 AM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,601
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 174 Post(s)
Liked: 60
Quote:
Originally Posted by djbluemax1 View Post

Just because you can show some random video claiming to show how interpolation works doesn't make it so.
Are you claiming that's not showing the result of video interpolation? Are you claiming I animated the whole interpolation myself and didn't use an interpolation algorithm?
Would you like to post a true interpolation of a similar video, with example frames, at similar frame rates to prove me wrong?
Quote:
I still don't see how the front of the car (in actual real world motion interpolation) would end up being further in front than the original image as the interpolated images produce motion in-between the successive frames, i.e the interpolation looks at frame A and B and creates intermediate frames where the object is between A and B, NOT looking at the positions in A and B and then putting the object in front of B.


Max
The original video the interpolation is working from is 3 fps (like I've said in previous posts). It's interpolating to create a 12 fps version. It doesn't see the true 12 fps version (which is shown above it) - it's interpolated from the 3 fps version of the same video. I haven't shown the result of interpolating the 12 fps video to a higher fps - I've shown the true 12 video, and a video interpolated from 3 fps to 12 fps (it shows "Interpolated to 12 fps" above the video not from 12 fps), showing how it puts things at incorrect positions (mainly due to interpolation artefacts, like ghosting, but even if there were no artefacts like that, interpolation never really knows the true positions of where something should be - it just interpolates (guesses)).

It puts it ahead of it's true position because of the artefacts (eg. ghosting - which is showing parts of the image from future frames in the current frame. Since the ghosting shows parts of the future picture in the current frame, it's shows it ahead of where it should be).

I'm not saying it looks at frames A and B and puts it ahead of B (though I'm not claiming that's not the case either). I'm saying the frames it generates between A and B can be at the wrong positions even though they are between A and B.
eg. A, A.5, B. Due to ghosting/interpolation errors A.5 could be at the wrong position ie. it's real position may not have been midway between A and B, and even if it should, artefacts including ghosting could take the picture from B and put it starting/ending at the wrong position than it's true position had it been shot at a true higher frame rate (as proved in my previous posts - where it shows the car at the wrong positions compared to the true (non-interpolated) 12 fps video, mostly due to artefacts like ghosting).
Joe Bloggs is offline  
post #310 of 316 Old 02-15-2014, 01:29 AM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,601
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 174 Post(s)
Liked: 60
Even if a TV using interpolation created no ghosting (and I think most/all of them will - which will also place things in the wrong place) or similar artefacts and was able to precisely place an object eg. exactly half way between 2 frames, since it doesn't know that that object really was at that position at that point in time, other than the times when it really should have been at that precise point, it will still have altered the speed of the object motion (eg. sped up then slowed down or slowed down then sped up) between those frames or put it in an incorrect path but at the correct speed.
Joe Bloggs is offline  
post #311 of 316 Old 02-15-2014, 08:30 AM
AVS Special Member
 
djbluemax1's Avatar
 
Join Date: Oct 2004
Location: MI
Posts: 2,278
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 28 Post(s)
Liked: 287
Using an interpolation of a 3fps sampling is a pretty poor example as that's a ridiculously slow rate first off. How about a 1-frame per minute example? Of course the argument then goes, "Well it's just an example, the principle holds true". And if we use a trafgic cam taking 1 photo per minute? Can we interpolate anything other than the movements of the sun across the sky at that rate?

Have you ever watched displays using motion interpolation. The algorithms must be pretty poor/old to cause effects like you're describing/illustrating.

I see more 'ghosting' on cheap slow LCD displays due to their native refresh rates than I EVER have due to motion interpolation.

As far as interpolation artifacts, with decent algorithms, the MOST common artifacts are haloing, which occur when an object moves past a complex background (eg a man walking past bushes/shrubs or a car driving past a chainlink fence). Decent algorithms lock on the important target object and maintain the solidity of it, but the algorithm results in a blurring of the background in a halo around the moving object precisely because they don't use a per-pixel algorithm.

What I DON'T see are the kind of artifacts you're indicating (and who the heck is watching a 3fps presentation anyway?).


Max
djbluemax1 is offline  
post #312 of 316 Old 02-15-2014, 08:46 AM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,601
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 174 Post(s)
Liked: 60
Quote:
Originally Posted by djbluemax1 View Post

Using an interpolation of a 3fps sampling is a pretty poor example as that's a ridiculously slow rate first off.
I chose a slow source frame rate to interpolate from to prove the point. Not to create a great looking interpolation but to easily prove to you that it can show objects at their incorrect positions and therefore speed up/slow down their motion. Which I proved. It would have been a lot less obvious if I had interpolated using something that didn't show it as clearly.

Also, if I had interpolated 24->120 fps or 24->48 fps, it wouldn't show properly as an anim smile.gif. Yes I could have uploaded to some other site, but doing it this way proved what I said was the case, as well as made it obvious.

Also, TVs interpolate 24->120 fps or 24->240 fps (as well as other rates). 120/24=5, 240/24=10, so they're interpolating to about 5 to 10 times the real frame rate. I only interpolated 12/3=4 times the original frame rate. So mine had to generate less of a ratio of fake frames than the TVs do.
Quote:
Have you ever watched displays using motion interpolation.
Yes. I own one.
Quote:
The algorithms must be pretty poor/old to cause effects like you're describing/illustrating.
The algorithms will be worse than the one I used (which used pixel motion interpolation - the very type you were wanting TVs to have, and was non-realtime). However they will usually have more frames to work with and the interpolated frames will be visible on screen for a lot less time so you won't be able to see the artefacts as clearly, even though they are there.
Quote:
What I DON'T see are the kind of artefacts you're indicating
Whether you see these kind of artifacts or not, the point is still proven. It can affect the motion and cause speed ups/slow downs due to the artefacting and the fact that it doesn't know the true position of objects at the frames it generates, other than where they were in the source video frames.
Joe Bloggs is offline  
post #313 of 316 Old 02-15-2014, 08:56 AM
 
Cyrano's Avatar
 
Join Date: Jul 2003
Location: Northwest Boonies
Posts: 5,738
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 90
Interesting.
Cyrano is offline  
post #314 of 316 Old 02-15-2014, 09:19 AM
AVS Special Member
 
djbluemax1's Avatar
 
Join Date: Oct 2004
Location: MI
Posts: 2,278
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 28 Post(s)
Liked: 287
While the multiples you quote are accurate, what folks need to take into account is the base rate too. At 3fps, the rate is so slow, anomalies are much easier to produce.

Take a shot of a break in a pool game for instance. The break is a hard shot (i.e. fast moving object), PLUS you suddenly get multiple objects moving in various directions. At 3fps, the rate is too slow to account for all this and anomalies are sure to abound.

Likewise with a table tennis game. The ball moves fast and arcs depending on the type of spin imparted. A super slow frame rate is going to produce anomalies. At 24fps, the frame rate is high enough to capture multiple movement positions, leaving the intermediate interpolated frames a fairly small 'window' of placement.

Take something shot at 120Hz and if you had the processing power to interpolate the sequence and ability to display it at 120,000fps, you wouldn't be able to see movement anomalies as the base 120fps rate is so high that interpolating it to produce 1000X as many images isn't going to cause huge errors (if the processing and display capabilities are there).

All this is moot and a red herring in any case because I still haven't seen any indication that commercial projection setups have ANY sort of frame interpolation available. If they did I'D WANT IT, because if they can interpolate 24fps with the complete lack of haloing/interpolation artifacts that I saw in The Hobbit HFR presentations, I WANT IT AT HOME TOO.

As far as HFR at home goes, I think what someone else suggested would be a great idea if it could be implemented, i.e. a dial-in blur and or judder setting on the equipment, eg. they could release AUJ (and any other HFR releases) in all their smooth motion super clarity/sharpness glory, and for folks who can't (or don't want to) get used to the look, have separate blur and judder settings allowing the user to dial-in "simulated 24p blur" and "24p judder" (or those folks could just stick to buying 24p versions). This could allow folks to gradually change the settings to adjust to their preference or adapt to higher frame rates.

Come to think of it, studios would probably prefer to release separate non-adjustable 24p and HFR versions, going by their proclivity for double/triple dipping. It would be their preference that folks used to the blurry juddery mess of 24p, buy the 24p versions, then if they eventually adjust to HFR, wind up double dipping for the HFR versions (or charge extra for the 5-pack HFR UHD, 24p UHD, BD, DVD, Digital DL, combo packs), rather than sell one version that different preferences can all adjust to taste and enjoy.


Max
djbluemax1 is offline  
post #315 of 316 Old 02-15-2014, 09:38 AM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,601
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 174 Post(s)
Liked: 60
Quote:
Originally Posted by djbluemax1 View Post

While the multiples you quote are accurate, what folks need to take into account is the base rate too. At 3fps, the rate is so slow, anomalies are much easier to produce.

Take a shot of a break in a pool game for instance. The break is a hard shot (i.e. fast moving object), PLUS you suddenly get multiple objects moving in various directions. At 3fps, the rate is too slow to account for all this and anomalies are sure to abound.

Likewise with a table tennis game. The ball moves fast and arcs depending on the type of spin imparted. A super slow frame rate is going to produce anomalies. At 24fps, the frame rate is high enough to capture multiple movement positions, leaving the intermediate interpolated frames a fairly small 'window' of placement.
The video I showed was a simple object moving slowly across the screen. None of that complicated stuff with many objects/fast movements. Like I said it might have been harder for it because it didn't have much detail (was all one graduated colour) but it was a slow, simple motion.

But I'm not against motion interpolation or anything. I have a TV that has it even though I rarely use the option. And if we're going to say interpolation causes motion artefacts (which it does) obviously 24 fps does too (judder/strobing/blur etc.) and things in motion don't stay at the same point for 1/24th of a second or 1/48th or however long the display shows it/re-shows it. So, like the point of this thread, what we really need is true higher frame rates, that will be the most accurate - ie. a lot more films should be at >=48 fps. Ideally at least 60 fps.
Quote:
As far as HFR at home goes, I think what someone else suggested would be a great idea if it could be implemented, i.e. a dial-in blur and or judder setting on the equipment, eg. they could release AUJ (and any other HFR releases) in all their smooth motion super clarity/sharpness glory, and for folks who can't (or don't want to) get used to the look, have separate blur and judder settings allowing the user to dial-in "simulated 24p blur" and "24p judder" (or those folks could just stick to buying 24p versions). This could allow folks to gradually change the settings to adjust to their preference or adapt to higher frame rates.
That's what I suggested smile.gif. I'm sure something like this would be easier to do well than interpolation to create higher rates/algorithms that try to reduce blur/judder. Though ideally you'd want it shot at a higher fps than 48 fps for this to work as well as possible (so depends on the shooting frame rate and shutter speed).
Joe Bloggs is offline  
post #316 of 316 Old 03-09-2014, 07:31 PM
Advanced Member
 
billdag's Avatar
 
Join Date: Apr 2007
Location: Vancouver, BC, Canada
Posts: 524
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 84 Post(s)
Liked: 66
I think a couple of things need to be reiterated. HFR is simply "filming" at a higher frame rate. SOE is also higher frame rate but with interpolated frames inserted between the actual 24 frames per second. Yes, I know everyone knows that but the thing is - HFR gives you smoother motion and generally, less blur. I see no downside to this, in fact I love it and have been waiting for it for decades!! Interpolated 24 fps presentations don't do anything to reduce blur, but can have some pretty strange artifacts - usually some weird mottled distortion surrounding an object in motion and how it interacts with a static or slower moving background. I agree this can be overcome with more advanced techniques. Having said all that, I still find that watching movies at home with the enhanced movie mode (SOE) turned on, is superior to 24 FPS with movie mode turned off.
As a side note, my TV, the Sharp LC-80LE632U also has a 'Standard" mode in movie settings. What this does, is remove the irregular 3-2 cadence and gives a more pleasing regular 24 fps judder, similar to a film presentation. I'm pretty sure they do this by repeating each of the frames five times to arrive at 120. So, for people who HATE SOE you should see if your TV has a similar mode. The 3-2 on a 60 Hz set is pretty ugly.
billdag is offline  
Reply Community News & Polls

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off