Compression Artifacts

[Daniell Simmons]

Hi all,

I just got back from a helicopter shoot with a Red in Africa. I've been looking at the footage for a week now, and although it looks great, I've been noticing some pretty severe compression artifacts

http://www.reduser.net/forum/uploaded/5825_1196738630.jpg

[Matthew Rogers]

Have you done a tiff export to make sure it's just not how RedCine displays the footage? Someone earlier today said there wasn't as much noise after they exported the footage from it. What about RedAlert?

Matthew

[Chris Nuzzaco]

I would second an export, at first, when I played with Jim's dragster clips, I wasn't too hot on how they looked, but it was just a viewing setting.

[Daniell Simmons]

I would second an export, at first, when I played with Jim's dragster clips, I wasn't too hot on how they looked, but it was just a viewing setting.

Yes, have exported tiff, .exr and dpx and get the same results each time. Admittedly, it's been done with the same video card (NVIDIA Quaddro FX 4400). Would a different card give better results?

B

[Greg M]

No this is an issue that is being addressed. Try the latest firmware (Build 11) it improves the problem a bit, but it still needs more optimization.

[Rick Darge]

It's amazing to know that this stuff will keep getting better.

It's blowing my mind right now..

[Gavin Greenwalt]

REDCode might improve with time, but I'm sorry to say the footage you've shot right now is what it is. That looks like too low of bitrate wavelet compression and once that data has been discarded it's gone.

[dino g]

also remember we are shooting moving images, these issues and artifacts are not present at 24 fps on a 30' screen.

have you ever put a feature film that has played more than a day at a multiplex through a simple magnifying glass...you want to see artifacts, dusts, cuts, scratches...

i know it is a not the best answer, but when you look at the options for watching this stuff, it will never be seen like you are dissecting it.

i have done digital projection screening with 2k and 4k projectors for 4 years for all the major studios, distributors and festivals and this footage we are shooting now is unlike anything that has come before it.

again, yeah it might look a bit compressed when blown up in a still image, but that is not how it will be seen by the audience & it is compressed.

[Justin Anderson]

also remember we are shooting moving images, these issues and artifacts are not present at 24 fps on a 30' screen.

have you ever put a feature film that has played more than a day at a multiplex through a simple magnifying glass...you want to see artifacts, dusts, cuts, scratches...

i know it is a not the best answer, but when you look at the options for watching this stuff, it will never be seen like you are dissecting it.

i have done digital projection screening with 2k and 4k projectors for 4 years for all the major studios, distributors and festivals and this footage we are shooting now is unlike anything that has come before it.

again, yeah it might look a bit compressed when blown up in a still image, but that is not how it will be seen by the audience & it is compressed.

well said.

[Daniell Simmons]

also remember we are shooting moving images, these issues and artifacts are not present at 24 fps on a 30' screen.

have you ever put a feature film that has played more than a day at a multiplex through a simple magnifying glass...you want to see artifacts, dusts, cuts, scratches...

i know it is a not the best answer, but when you look at the options for watching this stuff, it will never be seen like you are dissecting it.

i have done digital projection screening with 2k and 4k projectors for 4 years for all the major studios, distributors and festivals and this footage we are shooting now is unlike anything that has come before it.

again, yeah it might look a bit compressed when blown up in a still image, but that is not how it will be seen by the audience & it is compressed.

That may be true, but these days, not everyone is going to shoot and go straight to projection with no digital adjustment of the image.

[Oliver Koeppel]

did you try to blur the blue channel? The waveletcompression from redcode reduce data the most in blue channel.
Blur it first and resharpen it after.
I don´t think it will be possible in redcine. I did it in combustion with good results

[Álex Montoya]

Or denoise the blue channel.

[Brian Broz]

Great tip about blurring the blue channel.
But most of us (camera people especially) will not be too crazy bout adding yet another step just to process footage.
Again, I'm sure these noise issues will improve once RED has time to address them.

[ChrisLyon]

I'm investigating using the camera to shoot plates for backgrounds for a vfx-heavy feature and the ability to stabilise, grade, zoom, sharpen etc is critical.

Well his comments still are true. If you are rendering out a plate to be used in an image comp with smoke and cc and other VFX, that makes his point even more valid. Once you are moving your image in the final comp, this pixels are going to be re-rendered anyway.

My question to you is why you are wanting to make comps to zoom in on with footage rather than, say, a digital camera like a 10 MP Canon Rebel. You get more resolution which is better for comp work anyway, plus you don't have to constantly worry about motion blur in your plate frames. That would be my number one concern.

And if you were using footage as the plate, then, again he is right- the artifacts won't be visible because of the motion.

I understand your concern but I think it's rather mute at this point.

(Unless you don't have a digital camera)

[John Tissavary]

Well his comments still are true. If you are rendering out a plate to be used in an image comp with smoke and cc and other VFX, that makes his point even more valid. Once you are moving your image in the final comp, this pixels are going to be re-rendered anyway.

My question to you is why you are wanting to make comps to zoom in on with footage rather than, say, a digital camera like a 10 MP Canon Rebel. You get more resolution which is better for comp work anyway, plus you don't have to constantly worry about motion blur in your plate frames. That would be my number one concern.

And if you were using footage as the plate, then, again he is right- the artifacts won't be visible because of the motion.

I understand your concern but I think it's rather mute at this point.

(Unless you don't have a digital camera)

I doublt Animal Logic is going to have a problem getting a 10mp still cam, being one of the premiere vfx houses in the world.

I agree, basically, that the blue channel noise is not going to be much of an issue with plates. But it is a concern, and definitely an issue that needs to be solved asap. As they say, shit in, shit out... especially when you start applying moves in post.

What I'm confused about is that some are reporting the blue channel noise as very noticeable, and others aren't . Is this a difference in critical observation or variance between cameras/firmware/lighting/subject matter etc...?

I believe it's a combination of subject matter and exposure more than just exposure. I definitely noticed noise in some of our night shots, but less so in others with nearly the same exposure.

Anyway, I know Red is aware of this and an improvement is at hand, so I can't wait to see it!

cheers,

john t.

[Michael Lindsay]

The shot that 'bgunsberger' presents is basically almost as demanding as it gets for a digital motion camera (with spatial lossy data reduction.. ) The minutia detail and the color balance are tricky. I mention this for perspective not because I don't want it addressed..

I personally (when there is storage that supports this.. redram?) would like to see higher data rate options.

Michael

[jbeale]

I think Michael Lindsay's point is right on- the Red compression excels on a lot of "typical" feature shots where some significant part of the frame is not in sharp focus. But if you've got a long shot full of foliage detail where the entire image is in focus, something has to give, and apparently it's the darker areas.

[Daniell Simmons]

Well his comments still are true. If you are rendering out a plate to be used in an image comp with smoke and cc and other VFX, that makes his point even more valid. Once you are moving your image in the final comp, this pixels are going to be re-rendered anyway.

Quite the opposite. As the image may be graded, stabilised, retimed, etc, starting with good material is even more important, as each of these image-processing operations can degrade image fidelity.



My question to you is why you are wanting to make comps to zoom in on with footage rather than, say, a digital camera like a 10 MP Canon Rebel. You get more resolution which is better for comp work anyway, plus you don't have to constantly worry about motion blur in your plate frames. That would be my number one concern.

The shots we're working on involve mid-air action above and through forests, requiring moving vegetation and lots of parallax motion. Static panoramic capture is not an option.


And if you were using footage as the plate, then, again he is right- the artifacts won't be visible because of the motion.

The artifacts are quite visible at 2k on a DP90 pretty much out of the box

[Gavin Greenwalt]

Ben to put it simply. REDCode is over compressed for any real serious manipulation of the blue channel. I would investigate the RAW port option to shoot uncompressed. Hopefully there'll be a hardware 4k Cineform option early next year which could enable aerial footage without compression. Or else you can join in the chorus (currently of one :) ) asking for 50MBs REDCode RAW.

[Gavin Greenwalt]

Also I'm seeing some compression glitches in that footage I haven't seen before. It looks like a de-mosaic issue.
http://i111.photobucket.com/albums/n134/im_thatoneguy/CompressionGlitches.jpg

Are those in the original or a result of online posting?

[Daniell Simmons]

Don't get me wrong, I'm still a big fan of the camera, I just don't think the argument of "yes there are artifacts, but they're not *that* bad and are acceptable for most people so don't worry about it", doesn't fit with the philosophy of Red and what is being done.

I'd also like to hear if anyone has any guidelines for getting the most out of the compression, in terms of lighting, exposure, composition, camera settings, etc. It seems that people are getting wildly different results, but does anyone have any specifics of the strengths and weaknesses of the codec?

[Oliver Koeppel]

Of course bluring/denoising blue channel is just a workaround.
But if you want 4k stored on cf you need to compress the data significantly. And they did it nearly 1:10. Red calls it visually losless. I will call it great images, mostly visually losless for not pros. But sometimes you will see artefacts. That´s fact.
Redteam is damn good, but they´re still from earth, managing data with without being really magic.
It will be better with raw-port. It could be better with higher datarates stored on hdd.
Maybe I´m wrong in near future. Maybe the new daughterboard will fix it. Maybe they will do it in new redcode without touching datarates.
But if, I will call them the codecsorcerers from outer space or - let me think a while - keeping simple by calling them codecgods.

Just in case --> anyone here wanna pray with me twice a day towards lake Forrest?

[Rick Darge]

Ben, can you post up the r3d file? I'd like to run some tests out of redcine with the original.

[Gavin Greenwalt]

With the atmospheric contrast loss and it being almost exclusively in the blue channel the only thing I could think of doing to possibly bring up the noise floor and extract some more data out of it would be to try using some CTB filtration and opening up the lens a bit more.

[Oliver Koeppel]

I´m wondering about knowone recognized these artecacts before. Do you remember the hdr.tiff from evin? has been from february `07. Look at this great image -
and after that mind this part of same pic showing wavelat scars in blue channel. (8x zoomed)
This is how they did the incredible small datarate.

And btw I didn´t realize it either. Laguun showed me these bluescars...

[Álex Montoya]

It has been discussed many times.

[Steve Sanacore]

I´m wondering about knowone recognized these artecacts before. Do you remember the hdr.tiff from evin? has been from february `07. Look at this great image -
and after that mind this part of same pic showing wavelat scars in blue channel. (8x zoomed)
This is how they did the incredible small datarate.

And btw I didn´t realize it either. Laguun showed me these bluescars...

That blue channel CU is very disappointing but why would there ever be a reason to look at an image above 100% magnification except for pixel by pixel retouching.

I would love to see the color image at full 4K res to compare it to the blue channel sample.

[Steve Sanacore]

Doesn't the sample aerial image look very soft? Or is that just normal for motion picture helicopter shot? I would think the less sharp an image is, the more artifacts you may get as the software is tying to find detail to render? I would love to hear an experts input on this.

[Gavin Greenwalt]

Tafkar this has been in discussion for a long long time :D It didn't just creep up on us.

[Justin Anderson]

Or else you can join in the chorus (currently of one :) ) asking for 50MBs REDCode RAW.

Maybe there could be both; SP 27MB/s and HQ 40MB/s.

[Mitch Deoudes]

There's another thread discussing this issue here:

http://www.reduser.net/forum/showthread.php?t=6346

that includes some pretty good images of the problem. Although the thread starts to wander, and at times breaks up into bouts of spasmodic defensiveness. (I honestly don't understand why technical discussion should provoke such defensiveness in some folks. If we talk about these things, they'll get fixed - or at least we'll generate workarounds. Everybody loves the camera - just chill out.)

I'm going to have to weigh in on the "yes, I've seen it" side. It's definitely noticeable - particularly with tungsten-balanced lighting, mostly in the blue, some in the red, and in areas of low detail (which is, paradoxically, high detail - due to noise... i.e. your eye is going to pick out noise/grain more easily over a smooth surface). And it's definitely visible in motion as well, contrary to what some have suggested.

The thing that really gets me is the *size*. Check out post #8 in the thread above, in the upper-right corner over the blurry background. I'd think that sensor noise would be pretty high frequency - it surprises me that the codec responds with artifacts of that size. Then again, I stopped taking math classes as soon as I was allowed to.

In motion (RGB), it appears as a subtle crawling/pulsing in the luminance of the image. And I think this is the crux of the problem... i.e. film can get plenty grainy in the blue channel as well - but it's an organic effect that we're all used to. Compression / video noise artifacts just look weird.

I'm not sure either of the workarounds already suggested are ideal: 1) blur the blue channel (will piss off the compositors), 2) shoot everything daylight balanced (will piss off DP's and/or producers), 3) wait for uncompressed RAW output.

But I'm hoping with further discussion, we'll come up with something. (I will say, a selection of data rates/compression ratios in the future would be, in the vernacular, "dopesauce". Be great to just crank the rate when necessary.)

[Stacey Spears]

Maybe there could be both; SP 27MB/s and HQ 40MB/s.

I like the idea of different data rates, but one bottleneck might be CF at this point. The RED drive probably has higher bandwidth capacity in RAID0, but is more sensitive to bumpy enviroments.

Does 2k use the 27 MB/s data rate? If so, a test between 4k and 2k would be worthwhile.

This might be an area where someone could develop some algorithms for post to improve the quality or reduce noise. I am not a fan of noise reduction myself, but people use it all the time. a SCATCH plug-in anyone?

[Mark Pugh]

Maybe 3k could be an achievable compromise here - at 27MB/s

[Justin Anderson]

Does 2k use the 27 MB/s data rate? If so, a test between 4k and 2k would be worthwhile.


I assume 2K is 1/4 the datarate since it's 1/4 the image data. Correct me if I'm wrong.

[Justin Anderson]

Would it help to reduce blue and red noise by using filters to correct the white balance before it hits the sensor? (Like CTBs and stuff like that)

[Mitch Deoudes]

Would it help to reduce blue and red noise by using filters to correct the white balance before it hits the sensor? (Like CTBs and stuff like that)

Check out the thread I linked above. This was indeed one of the suggested workarounds. It has drawbacks, of course.

[David Mullen ASC]

It's not very practical for typical scenarios where you'd use tungsten lighting (night interiors and night exteriors), except for bluescreen shots.

The general consensus is that 320 ASA is the native speed of the sensor, so an 80A blue filter would knock this down to nearly 80 ASA, not very practical for your typical night interior scene with tungsten lamps. And of course using a higher ASA rating to compensate sort of defeats the point if you're trying to reduce noise.

I'd say that you're going to have to light night interiors scenes like you'd normally do on film -- with smaller tungsten lamps to match tungsten practicals in the scene, at a reasonable ASA level (320 to 500 ASA) so the light levels are not unrealistic nor overpower the natural ambience -- and deal with any noise level in the blue channel in post somehow if you find it to be a problem. Blue correction filters or blue gels is not really a good idea, not if you're trying to light a night interior scene naturalistically.

[jbeale]

Re: Redcode data rates: I don't know what the current firmware does, but based on Jim's two dragster R3D files posted a few weeks ago, I measured:

A001_C006_071102_001.R3D "Drag2" 2k 16:9 72 fps (526 frames)
Data rate: about 263 kB/frame, or 6 MB/sec at 24 fps

A003_C007_071105_001.R3D "Drag1" 4k 2:1, 23.98 fps (236 frames)
Data rate: about 1047 kB/frame, or 24 MB/sec at 24 fps

This is based on my analysis of the R3D files using a hex editor. The size of image frames within those R3D files is an integer multiple of 512 bytes (padded out with 0x00 bytes). The frame size is variable, but not by much: for example, with the 2k res file, most frames are exactly 263 kB, and the smallest frame is 245 kB.

When you think about it, a 4k 2:1 image is 8 Mpixels with 12 bits per pixel. When the output file is only 1 MB in size, that's quite a bit of compression (12:1).

[Oliver Koeppel]

I think, the problem we´re talking about only gets weight when using full 4k for mastering. If you downconvert 4k in 2k you will not see anything of these problems.
In my mind right now we have only two solutions:
downconvert 4k to 2k makes sense nearly most of times,
if you really need 4k, denoising blue is the way to fix it.
It´s only a workaround for now, maybe red will get it done someday.

btw: ok, have been late with the scars. Got the message :greedy:

[Stacey Spears]

I assume 2K is 1/4 the datarate since it's 1/4 the image data.

Yeah, I suppose it would be difficult to overcrank if you were running at the same rate. :) If you are not overcranking though, being able to up the bitrate could be a good thing.

[Brian D. Goff]

The left half of the image is a HDCam 730 frame, the right half is Jims dragster footage. RED footage matches pixelsize of HD (RED footage zoomed in). This is the blue channel - Makes me feel good to know that I will soon never shoot HD again:w00t:

[Brian D. Goff]

The left half of the image is a HDCam 730 frame, the right half is Jims dragster footage. RED footage matches pixelsize of HD (RED footage zoomed in). This is the blue channel - Makes me feel good to know that I will soon never shoot HD again:w00t:

[Stuart English]

Re: based on Jim's two dragster R3D files I measured:

A001_C006_071102_001.R3D "Drag2" 2k 16:9 72 fps (526 frames)
Data rate: about 263 kB/frame, or 6 MB/sec at 24 fps

A003_C007_071105_001.R3D "Drag1" 4k 2:1, 23.98 fps (236 frames)
Data rate: about 1047 kB/frame, or 24 MB/sec at 24 fps

Just a comment, the target data rates have been increased since this footage was shot..

[Álex Montoya]

RED takes amazing pictures, but what you have posted can't be a pixel per pixel comparison. I believe you are comparing 1080 HDCAM with 4K RED downrezzed to 2K.

Though it is true that for a projection this would be the way to compare it.

[Warren Kommers]

From my observations it just seems like too heavy on the compression. Is 40 Mbps REDCODE impossible? It would possibly be the end of CF cards but nonetheless it's about low light images that are ready for a 50' screen more than data effeciency.

[Álex Montoya]

I wouldn't like to drop the CF option. In any case, I guess they would create two presets for transfer rates.

[Steve Sanacore]

I am sure in a few years as CF storage gets much larger and faster, going an uncompressed route will be possible as long as RED has that option available in the camera. I think I read somewhere about 64 and 128GIG cards on the horizon already.

[Luis de la Cerda]

Hello. I've been lurking around for quite some time and I guess now's a good time to start posting. I don't have any firsthand experience working with red footage except for the clips that jim posted. However, I' ve had to deal with some compression artifacts in my own aerial footage originated on hdv 60i (ouch). I recommend trying out an Aftereffects plugin called neatvideo. It does a remarkable job at denoising/deartifacting footage since with it's adaptive noise detection engine, and can be configured to denoise with specific aggressiveness settings for each channel. There's always a tradeoff when doing this sort of thing, and in this case it'll most likely be losing some detail. But depending on your requirements it might be worth it. Furthermore, some luminance masking might do the trick of only denoising the dark areasofyour footage. You really have to see it in action to believe it. I've managed to salvage 16db + footage shot in hdv with it. It's light years ahead of AE's native denoising engine. And if you select an artifact plagued area of your footage to analyze noise characteristics it also does a very decent job at getting rid of said artifacts. Ymmv. Hope someone finds this info helpful.
PS. It's also quite inexpensive.

[jbeale]

I gather NeatVideo is from the same folks that brought us Neat Image (denoising for still photos). I know there must be loss of detail, but not as much as I would think. The examples here are pretty impressive. http://www.neatvideo.com/examples.html

http://www.bealecorner.com/D30/misc/NeatVideo1.jpg

[Gavin Greenwalt]

Neat video works! Miracles! I've used it for everything from denoising to cleaning up sample flickering in renders with GI. It's the bee's knees. That being said. The problem with REDCode isn't that it's too noisy. It's that it's not noisy enough. High noise + Heavy Wavelet compression = De-noised image and unfortunately it's not temporally denoised (like neat will do) it's intra frame de-noised and fine details such as... distant trees in aerial plates... are assumed to be noise and just scrubbed away leaving a smeary mess (which is a lot better than a boxy DCT mess but still problematic).