-
BMCC alternate workflow
-
[Walter Soyka] “Interestingly, FCPX’s 16b TIFF export and the original 16b TIFF sources differ slightly on brightness values; I would have expected them to be identical.
I am also seeing noticeable shifts from the originals in AE with 10b Uncompressed 422, ProRes 422, and ProRes 4444. (10b Unc 422 and ProRes 422 look to be identical; perhaps a faulty RGB/YUV transform somewhere? Perhaps I’m doing something wrong?)
ProRes 4444 shows dithering in addition to the shift (but again no truncation). The dithering here strikes me as extremely odd.”
The 16bit stuff is still really dicey in the mac os when going thru different apps.
The color difference being seen between import and export is the difference between “FullRGB” and “SMPTE” output ranges.
NO you are not wrong about faulty transforms with the conversion.
gary adcock
Studio37Post and Production Workflow Consultant
Production and Post Stereographer
Chicago, ILhttps://blogs.creativecow.net/24640
follow me on Twitter
@garyadcock -
Thanks Walter for the comprehensive testing. I have gained a lot of technical knowledge from your posts over the last year.
As Oliver noted in an earlier post, FCPX optimizes all stills to ProRes 422. Do you think this may be influencing the results of your tests. I wouldn’t have thought FCPX was using these files an intermediate codec for rendering deep formats, but who knows. Starting with a 16 bit RGB file, transcoding it to a 10 bit YCbCr intermediate, processing it in 32 Bit float linear RGB then exporting as 16 bit RGB would seem less than optimal.
I noticed when rendering the BMCC dngs on a ProRes444 timeline that ProRes 422 optimized media was simultaneously created in the event folder. I think the same occurred on export except when using compressor. I’m away from my computer so can’t test this with tiffs.
-
[gary adcock] “The 16bit stuff is still really dicey in the mac os when going thru different apps. The color difference being seen between import and export is the difference between “FullRGB” and “SMPTE” output ranges. “
I think there is more to it than that — the shift varies according to format.
When compared to the TIFF originals, FCPX’s 16b TIFF export shows the slightest change; 10b Unc 422 and ProRes 422 both show the same greater change, and ProRes 4444 shows a still greater change.
More on the ProRes 4444 outputs: a correction to the above. ProRes 4444 is not showing dithering as I had assumed; it’s showing very faint, regular horizontal lines — every eighth line is a touch brighter in all three channels (more pronounced in red and blue than in green). This is true in Motion 5 as well as AE CS6.
I will try to make time to collect some stills to share with the group here today so you can see what I’m talking about, but what started out as a quick test is quickly becoming a science project…
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events -
Rafael — reach me direct at walter at keenlive dot com.
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events -
[Rob Mackintosh] “As Oliver noted in an earlier post, FCPX optimizes all stills to ProRes 422. Do you think this may be influencing the results of your tests. I wouldn’t have thought FCPX was using these files an intermediate codec for rendering deep formats, but who knows. Starting with a 16 bit RGB file, transcoding it to a 10 bit YCbCr intermediate, processing it in 32 Bit float linear RGB then exporting as 16 bit RGB would seem less than optimal.”
Rob, I was working with only 1024 frames from black to white (ergo, 1024 discrete shades of gray) in that 16b TIFF original, so it should be able to be expressed in a 10b format. Repeating this experiment with a longer temporal ramp to purposefully violate that 10b barrier (say, 2048 frames/11b or 4096 frames/12b) would show if 10b intermediates are being used for render or not. I’ll put it on the to-do list. More to come later.
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events -
Test patterns are nice, because they are definitive; however, I like to look at real world stuff to see what the practical implications are. So here’s a slightly different test.
I took Shot #1 (close-up face shot of actress) and made a nice, punchy-but-pleasing image in After Effects from the DNG image sequence. Color correction was just tweaking the camera raw settings. I placed this into a 16-bit composition and exported a ProRes4444 file. Then I placed the same in an 8-bit composition and exported an 8-bit uncompressed YUV file.
Then I took these into FCP X and stacked them on a timeline setting the top layer’s blend mode to “difference”. Black screen, so no visible differences.
I did the same in After Effects in both 8-bit and 16-bit compositions. Again, black screen throughout.
This is with rendered QT files. When I get a chance, I’ll test QT refs, but the bottom line for me is that 8-bit vs. 10-bit vs. 16-bit doesn’t seem to make much of a difference with these images.
– Oliver
Oliver Peters Post Production Services, LLC
Orlando, FL
http://www.oliverpeters.com -
PS: One unknown in this discussion is whether the CinemaDNG raw importers are actually passing a 12-bit signal or truncating to 8bpc in the conversion to RGB.
– Oliver
Oliver Peters Post Production Services, LLC
Orlando, FL
http://www.oliverpeters.com -
[Oliver Peters] “Test patterns are nice, because they are definitive; however, I like to look at real world stuff to see what the practical implications are.”
The test was just designed to see if depth was preserved from 16b TIFF ref movies. It may or may not be a dealbreaker, depending on your workflow needs, but it’s certainly worth being aware of when you select a workflow.
[Oliver Peters] “I took Shot #1 (close-up face shot of actress) and made a nice, punchy-but-pleasing image in After Effects from the DNG image sequence. Color correction was just tweaking the camera raw settings. I placed this into a 16-bit composition and exported a ProRes4444 file. Then I placed the same in an 8-bit composition and exported an 8-bit uncompressed YUV file.”
I think that Adobe Camera Raw might be 8-bit only — the color picker in it is on a 0-255 scale, not 0-32767. Does anyone know for sure?
Also, did you change the ProRes 4444 output module’s depth to Trillions of Colors? Ae won’t do this by default, even if you have deep sources, and even if you’re working at 16 bpc, so it’s very easy to accidentally dither down to 8bpc on output.
[Oliver Peters] “Then I took these into FCP X and stacked them on a timeline setting the top layer’s blend mode to “difference”. Black screen, so no visible differences. I did the same in After Effects in both 8-bit and 16-bit compositions. Again, black screen throughout.”
If you push the viewer’s exposure up in AE, you’ll see the differences. Some may be dithering from the depth reduction, some may be compression, and some may be chroma sub-sampling on the 422 stuff, but they are there.
They will not be visually significant. Since you will more than likely end up with an 8-bit output or an 8-bit display, you’re right that it’s likely not important to preserve the extra depth for straight cuts. It will get smushed down on display eventually anyway.
Deep color is certainly useful for effects and grading, though, so if the depth is one of the reasons someone is considering using this camera, preserving it may be an important consideration when choosing a workflow.
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events -
[Walter Soyka] “The test was just designed to see if depth was preserved from 16b TIFF ref movies. “
Didn’t mean to sound like I was downplaying it. Certainly a very worthwhile and illuminating test.
[Walter Soyka] “Also, did you change the ProRes 4444 output module’s depth to Trillions of Colors?”
That’s a great question. I’ll have to recheck. You may be right and I may have exported it incorrectly.
[Walter Soyka] “If you push the viewer’s exposure up in AE, you’ll see the differences.”
Yes, I understand.
[Walter Soyka] “Deep color is certainly useful for effects and gradin”
I will say that it looks like you can push the grading pretty hard. I don’t see the range of raw tweaks as you would get from RED in REDCINE-X PRO, however. I also see completely different raw interpretation (such as color temp) in Adobe versus Aperture versus Resolve.
– Oliver
Oliver Peters Post Production Services, LLC
Orlando, FL
http://www.oliverpeters.com -
[Walter Soyka] “I think that Adobe Camera Raw might be 8-bit only — the color picker in it is on a 0-255 scale, not 0-32767. Does anyone know for sure?”
John Brawley’s blog and Blackmagic’s website both report that it’s 12 bit.
Shawn
Reply to this Discussion! Login or Sign Up