4k Projection Questions!

[Jay Drose]

We are in the final stages of projecting my latest short "Old Money." We finally got to see a test screener in 4k (video only) today and a few questions/concerns have arisen. Anyone who has experience projecting in 4k, feel free to chime in.

We noticed a good amount of shakiness in the Steadicam shots (something akin to strobing/juttering). What could this be coming from? Is it a frame rate issue? Too much information and the projector can't keep up? The slow pans and dolly shoots look good but the sweeping, steadicam shots have a lot of juttering issue. We did a TIFF Image sequence output (the total image sequence was 18,000+ frames, 650+ GIGS of data.)

I saw this Vimeo and found that this might be a fix: http://vimeo.com/1792533
What do you guys think?

Hope this hasn't been asked several times on the board already. All in all, the 4k projection is brilliant. It's such a pleasure to see the images we shot back in December up on the big screen after such a long post-production process.

Sidenote for those who were curious:

We found that the less color corrected the images were, the more pristine we found them to be. The more that the color was modified, the worst it looked on screen. Maybe this goes without saying, but it was VERY dramatic difference.

Underexposed images (there was only one major shot) were VERY noisy when blown up 4k with light added. We're sticking with the original exposure and dealing with the darkness the way it was shot.

 

[Noah Kadner]

Proper grading should improve footage. If it looks worse after the grading you're doing something wrong creatively or technically.

Noah

 

[Sarah C.]

I warned him!

 

[Marc Wielage]

What projector? What LUT? How was it graded? What is it playing back from?

We found that the less color corrected the images were, the more pristine we found them to be. The more that the color was modified, the worst it looked on screen.

That will happen if you're trying to make a drastic change, whether you're shooting on film or Red or anything else. I've often said in color-correction: "Darker is easy; brighter is hard." If you're trying to brighten up a dark-lit scene by a stop or two, yes, that will amplify the noise and make things grainy -- in any medium. It takes a skilled colorist to know how to coax the right look out of problematic footage, and even then, there are limitations. One hopes there's enough range in the original Red Raw files that there shouldn't be any problem.

The judder issue is troubling. I wonder if there's a throughput issue where the server just can't keep up. This is an issue in many high-end color correction systems, especially when working with 4K. I personally think it works fine in the color-correction process (particularly when there are tons of layers, windows, and keys going on) to just do the initial passes in 2K, and once they're approved, render everything using the original 4K files for the final output. It goes much, much faster this way, at least with current technology. Your miles may vary, and this is strictly my opinion.

Also, not all systems can perform optimally with native R3D files. Consult with the post facility to see if they need to debayer them to a more-compatible format first (like DPX). This may prove to be a more effective workflow with some systems.

 

[Jay Drose]

What projector? What LUT? How was it graded? What is it playing back from?


That will happen if you're trying to make a drastic change, whether you're shooting on film or Red or anything else. I've often said in color-correction: "Darker is easy; brighter is hard." If you're trying to brighten up a dark-lit scene by a stop or two, yes, that will amplify the noise and make things grainy -- in any medium. It takes a skilled colorist to know how to coax the right look out of problematic footage, and even then, there are limitations. One hopes there's enough range in the original Red Raw files that there shouldn't be any problem.

Maybe I exaggerated the non-color corrected files looking much more worse.

But it was like you said--the drastically CC'ed files seemed to project a muddier image than the naturally tinted originals. I understand that there are limitations and we are working to quiet the problems as much as possible.

The jitter effect is perplexing though...would love to hear more ideas on why this might be. I have a feeling it has to do something with the export. Each individual TIFF frame averages 30 mbs. Is it just a question of the system? I think not. I have a feeling its with the encoding or the export because we are projecting on a Sony SRX projector that has many 4k encoded projects with motion like ours.

 

[Sarah C.]

re: jitters

Shortest/quickest test of your TIFF sequence is to transcode/output it to a good-ol standard playable DVD and see if your original export did anything to it.

Better:

I looked up Sony SRX and it will take DVI or HD-SDI depending on the board(s) it has? That's good news, you ought to be able to test playback of your show in "any" format you pipe to it. You could transcode your TIFF sequence into a compressed format and pump it out via DVI* or HD-SDI (with a production video card or external device from a Mac) to the projector and test that part out.

*DVI.. I did playback from my MacBookPro dualcore 2.2ghz to a fullsized 2K Christie** projector at Paramount Pictures once.

** Pretty sure it was that, been years now. The projectionist thought that was pretty darn cool to have a woman in a black dress, high heels, hacking video in such a manner!

~s

 

[M Most]

Maybe I exaggerated the non-color corrected files looking much more worse.

Color correction is inherently a destructive process. The more you twist the values of an image around, the more destructive it's going to be. You can't take any value and turn it into anything else you want without paying a price, regardless of whether you're working with RAW originals or not. The better the image processing chain, the more apparent this becomes, much like focus. The lesson to be learned (too late, in this case) is that the best way to get pristine images that look the way you want is to shoot them that way.

I have a feeling it has to do something with the export. Each individual TIFF frame averages 30 mbs. Is it just a question of the system? I think not. I have a feeling its with the encoding or the export because we are projecting on a Sony SRX projector that has many 4k encoded projects with motion like ours.

You still haven't said what you're using to play this sequence back. I don't know of very many systems that can play back 4K TIFF files in real time reliably (or at all). Why did you decide to use an uncompressed image sequence for projection? And why TIFF? And where are you doing this?

 

[Jay Drose]

@ M Most:

That's what I was saying about the original exposures (with limited CC'ing) looking the best. This kind of goes against Noah's argument. The proof is in the images, as subjective as they are.

I'm not sure what systems they are running nor is it my domain to start playing with settings. I have a strong feeling that it is an issue on our end as far as the jitters go.

What confuses me is the fact that they have over ten 4k films that are sequenced as TIFF images. They playback flawlessly. A lot of it looks like it was shot on the Dalsa Origin, though I'm just speculating. There is only one other RED project on their system and none of it has any jitter effect, though there is far less motion than in ours.

Could it be a frame rate issue? Not staying consistent with the frame rate? Like the difference between 23.97 and 24fps?

 

[David Mullen ASC]

What is their explanation? Surely they have an answer.

I agree with Mike, it's not uncommon for uncompressed real-time 4K to cause playback systems to stutter or freeze momentarily.

 

[Brandon Kraemer]

I have dealt with judder in footage before (with Red footage and other footage), and the issue really comes down to panning speeds in conjunction with shutter speed and distance to subject. If the sweeping steady-cam shots (pans I assume?) are the only shots your seeing judder on, then that is likely your culprit. Can you detect this judder in a different monitoring environment? A projection will likely magnify the judder just due to the pure scale, but you should be able to see it on an HD monitor.

Try scaling up your shots or center cropping them as a test and watching them on a normal monitor. This will tend to magnify the perceived judder, might help you uncover what your dealing with.

Judder is very difficult to remove, but lowering contrast and detail can help. If you've added lots of contrast, detail or color contrast into the shot you might be amplifying the judder. You can try a plugin that does a optical flow retiming and force motion blur or vector blur into the shot, but this can have some nasty side effects.

Also check the metadata on that shot, what was the shutter speed and the frame rate. Were you over-cranked?

 

[Ravi Kiran]

Thought I'd piggyback on this thread.

We're having a screening of our film at a theater with a Sony SRX T110 4K projector. I'd love to screen it in 4K, but we'll probably screen it at 1920x1080 24p from a ProRes HQ file (or even H.264, if that would look okay). What are our options for this? Would this work okay with no hiccups from the internal drive of a MacBook Pro? What are some options for playing the video? Could an entire feature film be played off P2 cards from a Panasonic HPG-20 using the HD-SDI output? Would converting the video to 60i help ease the judder or just introduce interlacing artifacts? Could we get 5.1 sound from a P2 card reader or a file read from a MBP?