FORUMS: list search recent posts

Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files

COW Forums : DaVinci Resolve

<< PREVIOUS   •   VIEW ALL   •   PRINT   •   NEXT >>
Scott Clements
Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 25, 2013 at 2:20:39 pm

Hi, Everyone.

I posted this in the FCP forum, but this question is also related to a grade that will most likely be done in Resolve. My question is about how information is lost through re-rendering of Apple ProRes 422 files. I have a feature where the original files were Apple ProRes 422. The finished film will have a 2.35 crop and some titles for the opening credits. If the ProRes files are graded in a dedicated system, such as Resolve and then delivered to the director as ProRes 422, can he then bring the movie back into FCP and do an online, one in which he puts black bars and titles over top? Can he do this without degrading the parts of the image unaffected by the bars and text? I am not fully aware how rendering works. If black bars are cropping, they're of course not affecting the parts of the image we care about, but does just the simple act of re-rendereding the video degrade the image as a whole? My initial thought was that he absolutely had to put the bars on during the grade and I was also concerned about putting titles on once the film had gone back to ProRes. Is there anyone out there who understands the exact mechanics of rendering who can shed some light on this matter? Many thanks.

Sincerely,

Scott

Film Editor, London UK
http://www.scottclementseditor.com


Return to posts index

Juan Salvo
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 25, 2013 at 2:46:39 pm

Resolve can easily add blanking to any desired aspect ratio, and also do repositions as needed. I'd recommend doing this in the color session, as Resolves scaling is far superior to that of FCP, plus you'd do your scaling, cropping and color in one codec cycle.

Colorist | Online Editor | Post Super | VFX Artist | BD Author

http://JuanSalvo.com


Return to posts index

Scott Clements
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 25, 2013 at 2:51:50 pm

Thanks, Juan. So a codec cycle does degrade the frame as a whole, not just the part of the frame that has been manipulated? I'm assuming this is the case, because Apple ProRes 422 is a lossy codec.

Film Editor, London UK
http://www.scottclementseditor.com


Return to posts index


Juan Salvo
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 25, 2013 at 3:17:40 pm

It's a very marginal loss. You'd have to go probably dozens of generations before you could see any degradation in most viewing conditions. But even uncompressed codec cycles are lossy. Anytime you're resampling an image, or doing any processing to it, something is lost. That shouldn't be alarming though. Going 2-3 gen on ProRes 422HQ is not a big deal. A bigger issue is the crappy quality of FCPs scaling.

Colorist | Online Editor | Post Super | VFX Artist | BD Author

http://JuanSalvo.com


Return to posts index

Jef Huey
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 25, 2013 at 3:46:28 pm

Juan,

I think you need to be more specific about your comment that even an uncompressed codec cycle is lossy. You did qualify that by mentioning resampling. But I don't think you went deep enough with this.

In the case of the OPs question, if the image is not re-sized or repositioned at all and the codec type was not changed (same frame size, frame rate and codec type) - then there will be no "loss" because there is no resampling of unaffected pixels. Adding black at the top and bottom of the frame will not change the pixels in the inside of the black mask. That is the nature of an uncompressed codec. Each pixel is handled individually.

In a compressed codec, pixels are handled as a group. With any codec using any compression type, to change any material, the whole frame must be decompressed, the change made and the whole image recompressed. In a situation such as was described, there should be minimal change in the image when adding a black mask if the rules of no other picture change are followed.

But I would start to be worried about color accuracy issues seeing as Quicktime will be involved (ProRes) within multiple devices. Not saying it will occur, but many variables can get in the way of a perfect codec cycle with Quicktime.

Good luck,
Jef


Return to posts index

Scott Clements
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 25, 2013 at 3:54:18 pm

Thanks guys for the amazing responses! Very helpful!

-Scott

Film Editor, London UK
http://www.scottclementseditor.com


Return to posts index


Juan Salvo
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 25, 2013 at 4:56:31 pm

[Jef Huey] "In the case of the OPs question, if the image is not re-sized or repositioned at all and the codec type was not changed (same frame size, frame rate and codec type) - then there will be no "loss" because there is no resampling of unaffected pixels. Adding black at the top and bottom of the frame will not change the pixels in the inside of the black mask. That is the nature of an uncompressed codec. Each pixel is handled individually. "

Not so I'm afraid. DaVinci and most other apps work in RGB, ProRes 422 and Uncompressed 4:2:2 codecs store information in YUV. Doing any processing at all, or in some cases just running it through the app, will result in a resampling cycle. Now, like I said, the degradation of said cycle is minimal. But it DOES occur.

One of the few exceptions is a situation like an FCP project where the source content, sequence setting and export codec all match. And there is no processing needed on the entire frame. In which case FCP can just pass through the frame data. If any processing is done at all (including a crop the image) then the entire frame is sent through the processing path.

Colorist | Online Editor | Post Super | VFX Artist | BD Author

http://JuanSalvo.com


Return to posts index

Jef Huey
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 26, 2013 at 12:57:20 am

Hi Juan,

Of course you are correct about the YUV to RGB to YUV conversions.

But the OP asked about the director doing the work INSIDE FCP. At that point given that he was talking about presenting the director with 4:2:2 files from color grade to do the final work there. Working inside FCP or any other NLE if you stay in the same uncompressed codec (as I noted in my response) and no transforms, a crop should not incur any change to pixels. Adding a title will only change the pixels where the title is added. Which is the goal.

Jef


Return to posts index

Juan Salvo
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 26, 2013 at 4:05:44 am

Hi Jef,

I'm afraid you're mistaken. Any frame with a title or a crop or anything else that needs to be rendered in FCP, has the entire frame go through the entire processing path, not just the pixels under the title. The entire frame. This is the case regardless of codec used.

One thing for the OP to keep in mind is that the issue of generational loss has been with us from the very beginning of film production, all the way through video to the present day and on into the future. And if anything the present standard for both compressed and uncompressed codecs are the best we've ever had for the purpose they serve. You are, for example, far better off with 10 generation old ProRes 4444, in my opinion, than a ten generation old film print.

Here is one of my favorite things on the Internet. Though now slightly out of date, Marco Solario's codec test page is a great resource for understanding what multiple generations, even within the same codec will do to an image.

http://codecs.onerivermedia.com/

One thing Marco hasn't tested is the ProRes flavors, which I suspect would hold up remarkably well.

Colorist | Online Editor | Post Super | VFX Artist | BD Author

http://JuanSalvo.com


Return to posts index


Jef Huey
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 26, 2013 at 1:25:34 pm

Hi Juan,

I want to be sure we are discussing the same thing here as I feel we may be missing some things.

I am talking about working inside an NLE with a source file of uncompressed YUV material where any rendering will be YUV based. FCP seems to fall into that category assuming projects setting are set so.

Also, one note from Marco's codec test. "Please also note that these are RGB to YUV render tests and NOT a native YUV render test." So any comments he makes are not pertinent to this discussion.

Ok, all that said, in the case where a YUV file is brought into FCP and a black mask is added and NO image interpolation is performed, you say that the entire frame must go through processing path. No argument from me. Where I do not understand you is that it seems as if you are saying that pixels in the middle of the frame - far away from the mask area - have been changed merely by the processing path. Is that correct? If so, then either FCP is doing something that most other NLEs do NOT do or I disagree with you.

I have talked with developers of several NLEs (though not FCP) about this issue in the past. They assured me that in an uncompressed environment only the pixels that needed to be changed by a process WERE changed. All others were left unaffected. If you have documentation to prove otherwise I would love to read it.

Thanks,
Jef


Return to posts index

Joseph Owens
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 25, 2013 at 5:22:20 pm

[Jef Huey] "I think you need to be more specific about your comment that even an uncompressed codec cycle is lossy. "

If there is any interpolation. The only time there is no quality loss in a video process is when you simply clone the file.

Might be time to go back to 2009:
http://images.apple.com/support/finalcutpro/docs/Apple-ProRes-White-Paper-J...

Page 13 (of22) is the meat of the matter. At the end of the day, ProRes422(HQ) and its fancy big sibling 4:4:4:4 were designed to be "visually lossless", while speeding up processing time and reducing storage and throughput demands. Usually that factor is somewhere around 10% of "Uncompressed", while retaining much better than 99% -- more like almost 100% of the multigenerational performance of what Apple calls "10-bit Uncompressed".

As the White Paper refers to, most of the degradation artifacts that we see in final masters are the fruits of what we invented the term "GIGO" to describe in the 70s. You cannot expect 4:2:0 and 4:1:1 sourced Long-GOP to match RGB intra 4:4:4 space without a significant amount of intervention, and if your source discipline is less than perfect, matching scene levels with imaging capabilities (SNR, MTF, DOF, and so on) then either you require an actual magician or a ton of careful value placement. We all have those tools, but they do come with a price.

jPo

"I always pass on free advice -- its never of any use to me" Oscar Wilde.


Return to posts index

Pepijn Klijs
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 26, 2013 at 8:24:26 am

This was an interesting read for sure, and I'm not going to enter this discussion as it is way beyond my level of knowledge!

However, one tip to the original poster: if you are concerned about making generations, try to finish in Resolve. You can do the blanking, scaling and add your final audio track. You can also render out titles with sn alpha channel from something like after effects or motion and add them to your resolve timeline.

Editor/Colorist, Amsterdam, The Netherlands
http://www.pepijnklijs.nl


Return to posts index


MichaelMaier
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 28, 2013 at 10:36:14 am

[Pepijn Klijs] "However, one tip to the original poster: if you are concerned about making generations, try to finish in Resolve. You can do the blanking, scaling and add your final audio track. You can also render out titles with sn alpha channel from something like after effects or motion and add them to your resolve timeline."

Yes. I 100% agree with this. I think it would be the best option. Definitely better than going back to FCP.



Return to posts index

Jef Huey
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 26, 2013 at 12:57:40 pm

Hi Joe,

I understand the issues of compressed codecs. My comments have been about the use of an uncompressed codec. Sorry if I was not clear enough about that.

Cheers

Jef


Return to posts index

MichaelMaier
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 28, 2013 at 10:37:10 am

[Juan Salvo] "Resolve can easily add blanking to any desired aspect ratio, and also do repositions as needed. I'd recommend doing this in the color session, as Resolves scaling is far superior to that of FCP, plus you'd do your scaling, cropping and color in one codec cycle."

Sure. But if he would still need to go back to FCP to master it he would still be recompressing to the final output.

I think a better suggestion would be to master and finish it from Resolve, like Pepijn Klijs said.



Return to posts index


MichaelMaier
Re: Question about Rendering Degradation in Grading/Onlining Apple ProRes 422 Files
on Mar 28, 2013 at 10:40:52 am

In my opinion you should always avoid any recompression like the plague. The less the better. So I would try to master and finish it in Resolve, like Pepijn Klijs said. That way you would only have one recompression. To the final delivery format. That is if you would be working with the original files in Resolve through an EDL/AAF/XML import.



Return to posts index

<< PREVIOUS   •   VIEW ALL   •   PRINT   •   NEXT >>
© 2017 CreativeCOW.net All Rights Reserved
[TOP]