Jump to content


Photo

NAB: High frame rate 3D


  • Please log in to reply
3 replies to this topic

#1 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11939 posts
  • Other

Posted 19 April 2012 - 03:54 PM

Christie have a demo here showing stereoscopic 3D material at both the conventional 24fps, and at higher frame rates of 48 and 60. The whole thing is introduced by James Cameron, who insists the whole 24Hz update rate of conventional cinema is a nearly century-old throwback to the early 20th century (which is certainly true), and that increasing that frame rate will make 3D easier to watch, which is by no means certain.

I should probably start with the disclosure that I've never been a huge fan of 3D for little reason more than that it always - always - makes me feel unwell. There's a tendency for 3D apologists to scoff at this position, implying that in each case there was something wrong with the acquisition or exhibition technology, the techniques employed, or, frankly, that there's just something wrong with me. There are certainly a number of ways in which stereoscopic projection does not accurately match the way we perceive real three-dimensional objects, and there's medical issues as well, such as the small proportion of people whose brains coordinate focus and convergence distance so well that a disparity between the two - which is inevitable in 3D film projection - will always be uncomfortable. I may be one of these people, but either way, it did seem intuitively true that a higher frame rate, which would reduce visible lack of temporal resolution (judder, flicker, chatter, etc) might make it easier to view the edges of moving objects, where stereoscopy is perhaps the most visible of all.

The demo shows that this intuitive truth certainly is true, at least in part. Cameron apparently doesn't believe in the idea that edge violations are a problem, and I suspect that this contributed somewhat to the fact that I ended up with a slight feeling of nausea anyway. But subjectively - and what other assessment is possible, when considering comfort - it is somewhat better. I still don't think it's the final answer. There was also no discussion of the reasoning behind using a much wider shutter angle on some of the higher frame rate material, which they did do - presumably this is just an issue of preventing the short exposure intrinsic to high frame rates from having other, undesirable effects on motion rendering.

As an aside, it's worth mentioning that the demo also included downconversions of the 48fps and 60fps material to 24fps, which is critical in a world where we will still need to produce 24fps versions - even film prints - of major productions for some time. This is somewhat easier with the 48fps version, where we simply need to drop every other frame and render in some additional motion blur using optical flow. They also showed a 60-to-24 conversion using optical flow to derive correctly-time frames. This looked absolutely fine on the particular material we saw, which had been particularly chosen and very much directed to produce lots of fast motion to elucidate chatter and flicker, and would also therefore be a trying subject for any optical flow implementation. It's almost inevitable that it would fail on some subjects, even though it's a framerate decrease rather than increase, so I suspect that 48fps may be a good compromise for productions that do want to shoot at something faster than 24, as it avoids this issue entirely. I wouldn't say there was absolutely no visible difference between 48 and 60, but I suspect the common number may end up being 48 just for this reason.

The other issue covered was overcranking for slow motion. While many modern cameras (such as Alexa, which shoots up to 120fps) would still allow overcranking even on a 48 or 60fps production, simple step printing is a lot more acceptable when the output frame rate is high to begin with - 2:1 step printing is practically invisible, because it simply reduces the effective update rate on a 48fps show back to the 24fps we've been used to for a century.

I still don't think 3D is fixed, I don't think this solves all the problems, and I am extremely used to having people tell me all the problems are fixed. I think this does make a real difference, though - not a huge difference, but a difference. With most D-cinema installations only a software update away from being able to display high-frame-rate material, I wouldn't say it's impossible that it'll catch on - but at the end of the day, we did all clamour for 24fps cameras for a reason and I suspect it may not be mainstream for a longish while, if ever.

P
  • 0

#2 John Rizzo

John Rizzo
  • Basic Members
  • PipPip
  • 56 posts
  • Other
  • New York N.Y.

Posted 21 April 2012 - 09:35 AM

I also saw this demo film and was not at all impressed, the higher frame rate looked better, but the look was much to live, it didn't look like a movie





Christie have a demo here showing stereoscopic 3D material at both the conventional 24fps, and at higher frame rates of 48 and 60. The whole thing is introduced by James Cameron, who insists the whole 24Hz update rate of conventional cinema is a nearly century-old throwback to the early 20th century (which is certainly true), and that increasing that frame rate will make 3D easier to watch, which is by no means certain.

I should probably start with the disclosure that I've never been a huge fan of 3D for little reason more than that it always - always - makes me feel unwell. There's a tendency for 3D apologists to scoff at this position, implying that in each case there was something wrong with the acquisition or exhibition technology, the techniques employed, or, frankly, that there's just something wrong with me. There are certainly a number of ways in which stereoscopic projection does not accurately match the way we perceive real three-dimensional objects, and there's medical issues as well, such as the small proportion of people whose brains coordinate focus and convergence distance so well that a disparity between the two - which is inevitable in 3D film projection - will always be uncomfortable. I may be one of these people, but either way, it did seem intuitively true that a higher frame rate, which would reduce visible lack of temporal resolution (judder, flicker, chatter, etc) might make it easier to view the edges of moving objects, where stereoscopy is perhaps the most visible of all.

The demo shows that this intuitive truth certainly is true, at least in part. Cameron apparently doesn't believe in the idea that edge violations are a problem, and I suspect that this contributed somewhat to the fact that I ended up with a slight feeling of nausea anyway. But subjectively - and what other assessment is possible, when considering comfort - it is somewhat better. I still don't think it's the final answer. There was also no discussion of the reasoning behind using a much wider shutter angle on some of the higher frame rate material, which they did do - presumably this is just an issue of preventing the short exposure intrinsic to high frame rates from having other, undesirable effects on motion rendering.

As an aside, it's worth mentioning that the demo also included downconversions of the 48fps and 60fps material to 24fps, which is critical in a world where we will still need to produce 24fps versions - even film prints - of major productions for some time. This is somewhat easier with the 48fps version, where we simply need to drop every other frame and render in some additional motion blur using optical flow. They also showed a 60-to-24 conversion using optical flow to derive correctly-time frames. This looked absolutely fine on the particular material we saw, which had been particularly chosen and very much directed to produce lots of fast motion to elucidate chatter and flicker, and would also therefore be a trying subject for any optical flow implementation. It's almost inevitable that it would fail on some subjects, even though it's a framerate decrease rather than increase, so I suspect that 48fps may be a good compromise for productions that do want to shoot at something faster than 24, as it avoids this issue entirely. I wouldn't say there was absolutely no visible difference between 48 and 60, but I suspect the common number may end up being 48 just for this reason.

The other issue covered was overcranking for slow motion. While many modern cameras (such as Alexa, which shoots up to 120fps) would still allow overcranking even on a 48 or 60fps production, simple step printing is a lot more acceptable when the output frame rate is high to begin with - 2:1 step printing is practically invisible, because it simply reduces the effective update rate on a 48fps show back to the 24fps we've been used to for a century.

I still don't think 3D is fixed, I don't think this solves all the problems, and I am extremely used to having people tell me all the problems are fixed. I think this does make a real difference, though - not a huge difference, but a difference. With most D-cinema installations only a software update away from being able to display high-frame-rate material, I wouldn't say it's impossible that it'll catch on - but at the end of the day, we did all clamour for 24fps cameras for a reason and I suspect it may not be mainstream for a longish while, if ever.

P


Edited by John Rizzo, 21 April 2012 - 09:39 AM.

  • 0

#3 Chris Millar

Chris Millar
  • Basic Members
  • PipPipPipPip
  • 1642 posts
  • Other

Posted 21 April 2012 - 03:58 PM

This is somewhat easier with the 48fps version, where we simply need to drop every other frame and render in some additional motion blur using optical flow.



Can you tell me more about this ? Any flow taken from one frame (as the other was dropped) will be an extrapolation, and hence prone to all the attendant error of any extrapolation process.

So a simple 50/50 mix of the two frames produces something nasty ? You'd get the 180deg shutter again but with two exposures 90deg apart (in 24fps timelines).

Maybe they use the dropped frames in analyzing the motion so the extrapolation is more an interpolation ?


  • 0

#4 Chris Millar

Chris Millar
  • Basic Members
  • PipPipPipPip
  • 1642 posts
  • Other

Posted 21 April 2012 - 04:21 PM

I'ved read up on it - sorry, my post was premature ...

It's all interpolation and of higher order. But I'd still maintain that throwing away every second frame is a waste of information, I don't mean for us to watch, but to use in the optical flow algorithms Posted Image
  • 0


Gamma Ray Digital Inc

rebotnix Technologies

FJS International, LLC

Media Blackout - Custom Cables and AKS

Wooden Camera

Rig Wheels Passport

Ritter Battery

CineTape

Aerial Filmworks

The Slider

Abel Cine

Tai Audio

Paralinx LLC

Glidecam

CineLab

Broadcast Solutions Inc

Metropolis Post

Opal

Technodolly

Willys Widgets

Visual Products

FJS International, LLC

Aerial Filmworks

The Slider

Metropolis Post

Opal

Abel Cine

Technodolly

Ritter Battery

rebotnix Technologies

CineLab

Tai Audio

Rig Wheels Passport

CineTape

Paralinx LLC

Visual Products

Gamma Ray Digital Inc

Media Blackout - Custom Cables and AKS

Broadcast Solutions Inc

Glidecam

Wooden Camera

Willys Widgets