Reconforming to original H.264 after editing 5D footage in ProRes 422?

[Michael Sacci]

You cannot edit or color correct H264 footage. It has to be transcode before you start editing.

[Zak Stoltz]

Yes, I’m aware of this. If you had read my post more closely, you would see that we are currently editing in ProRes 422 and color correcting Uncompressed HD footage. If you would like to address my original question, I encourage you to go back and read the original post.

[Michael Sacci]

So you are asking if you should go from H264 to ProRes to H264 to Uncompressed. I have never seen anyone recommend the step back to H264 and you should never do that. That is adding a layer of major compression again to the process.

[Zak Stoltz]

I’m afraid you’re still not getting it. The native recording format for the 5D is H.264. First we take the original H.264 footage and transcode it to ProRes 422 to edit. Then, once we are done editing, we replacing the transcoded footage with the original footage before exporting it as uncompressed HD. It’s like an online/offline edit. Get it?

So the question then becomes whether or not it makes sense to go back to the original footage as opposed to staying in ProRes before going uncompressed.

It’s the difference between…

[Original H.264] –> [Uncompressed HD]

and

[Original H.264] –> [ProRes 422] –> [Uncompressed HD]

Will the second method listed result in noticeably lesser video quality than the first method?

[Michael Sacci]

I guess I was reading to much into the post. If you want to do it, why not, most people will not even go to the uncompressed codec at all. But maybe that is what your finishing people want. But do some test, if you think you are gaining something do it. I would not even bother, I would keep it ProRes to the end.

[Uli Plank]

The initial H.264 compression is so destructive, you won’t see any difference. Just wasting space!

Director of the Institute of Media Research (IMF) at Braunschweig University of Arts

[Zak Stoltz]

Thanks for the input, guys. This is what I’ve always thought, but the higher-ups at my company don’t seem to share that view…

Does anyone know of a definitive test or example that proves reconforming to H.264 is unnecessary? I’m trying to convince the decision makers here that, as Dave said, it’s a waste of time. 🙂

Also, if anyone disagrees, and thinks it WOULD be beneficial to go back to the original H.264 footage, I’d like to hear from them as well.

[Chris Tompkins]

Why do they want to re-compress the footage further degrading it when it’s edited and finished?
To use as a deliverable?

Chris Tompkins
Video Atlanta

[Zak Stoltz]

Haha, thanks for saving me that explanation again, Dave. Chris is the second person to misunderstand what I’m talking about. I wonder, is “reconform” the wrong word to use when talking of replacing the ProRes with the original H.264?

In my quest to convince the powers that be to cut out the inefficiency in their workflow, I performed a couple of tests and got some very, VERY interesting results:

Using the Canon EOS plugin for FCP, I transcoded a clip from H.264 to ProRes 422 and compared the gamma, color, and sharpness by “stacking” the original footage and the ProRes one on top of the other and switching back and forth between them at various zoom levels. In terms of gamma and color, they’re IDENTICAL.

Also, I’m not exactly sure how it happened, but it looks like sharpness is actually INCREASED when transcoding from H.264 to ProRes, and that increased sharpness carries over into the final uncompressed 10 bit exports!

After performing the same tests with 10 bit uncompressed exports (one coming directly from the source footage and the other coming from the ProRes transcode), the one coming from ProRes was still the clear winner.

Thanks for the input everyone. Looks like this case is closed.

[Doug Beal]

is “reconform” the wrong word to use when talking of replacing the ProRes with the original H.264?
Yes the operation you describe would be called a conform. Usually used for an offline/online situation where an offline codec is used to edit, then an xml, EDL or reconnection of media using the highest quality available source footage takes place based on the TC/reel number from the edit.

reconform would assume a conform had taken place already and you were doing it again.
you have merely transcoded then edited in your case.

Glad you were able to convince your people to skip the return to h264. Totally unnecessary.

Doug Beal
Editor / Engineer
Rock Creative Images
Nashville TN