[Ivan Myles]

I agree with Kris’ comment that the issue is related to I-frames; it looks like the P- and B-Frames are being read directly. Were the clips cut prior to importing to Premiere? Try transcoding a short test clip from the source file into an I-frame only format (preferably lossless), import to Premiere, insert into the timeline, and export to see if the issue is resolved.

[Ivan Myles]

Thank you for the clarification. Did you try removing all the third party codecs/players (vlc, divx, klite, etc)?

[Ivan Myles]

[Navin Goyal] “How do I proceed with editing such a video. Its an avi with something related to mpeg2.”

Please provide more information regarding the file type. Your posted video is formatted as H.264/MOV, which coincides with the A2300 specs. Was the source file transcoded to AVI? Is this the preferred output format? Can you play the source H.264/MOV file in QuickTime?

[Ivan Myles]

Happy go help. Good luck with the project.

[Ivan Myles]

Hopefully the discussion was helpful. We didn’t seem to resolve the root cause, though.

It will be easier to avoid macroblocking if grain is added before the first ProRes export. However, this might impact your colorization. Try a few test samples and decide which workflow produces the best overall result with your footage.

Good luck with the project.

[Ivan Myles]

It depends which codec is selected. QuickTime H.264 seems to be limited to 2000 pixels, but exporting H.264 directly allows 3840×720 at Level 5.1. Alternatively, you can use a different QuickTime codec. I am able to specify 3840×720 with every standard QT codec except the DV’s, DVCPRO’s, and H.264.

[Ivan Myles]

Thanks for the follow-up. How do the FCP and Premiere ProRes exports compare to the source file when viewed with the Vect/YC Wave/RGB Parade scopes? If the Premiere version has lower contrast, you should see the YC and RGB curves compress slightly (brights shift down, darks shift up) compared to the source file. Or is the Premiere version closer to the original, and the FCP file has an expanded color profile where brights shift up and darks shift down?

Here are samples based on your png screen captures. First, the Premiere version:

And here’s the FCP version:

When looking at the RGB profiles, the dark bands are more prominent in the Premiere file (most noticeably in the red channel). It looks like a slight gamma shift and loss of color fidelity. Does this coincide with the scopes you see when comparing the three files in Premiere?

The only comment I could find from Adobe came in this post comparing macroblocking in H.264 exports from Premiere and FCP. An Adobe product manager stated the following:

… our philosophy is to preserve every bit of color, light and detail we can – so we apply no filtering… However we found that if you introduce grain at the front end of the decode chain you end up with a more appealing image in these cases, however, any noise introduced reduces picture quality…

So – there you have it – we give you the pixels in the purest representation as captured – what you do with them from there is up to you. QuickTime? Well you’d have to ask Apple, but it seems that there is some form of image alteration being applied, but we can only judge by the results.

He seems to suggest that the only way to resolve the issue is to introduce grain and/or adjust luma/color. I am not sure this explains why you are seeing a difference when using the same codec in Premiere and FCP.

Based on the settings you already tried, at this point the only other thing might be to select a different option for Gamma Correction under “Codec Settings” in the export dialog box. Also, be sure that “Use Previews” is unchecked. Otherwise, you might just go with ProRes 444 or an alternate codec. It doesn’t explain the observed differences between Premiere and FCP, but at least you can get the desired results.

[Ivan Myles]

I just layered the screenshots from your original post and the two files are not scaled the same. The Premiere Pro frame is about 3.4% bigger than the FCP frame.

The other point of note from your comments is that the files are being compared in QuickTime. Have you tried to evaluate in an NLE? I suggest the following test:

[Ivan Myles]

The following chart shows image accuracy with different levels of compression for a grayscale gradient. 4:2:2 introduces some degradation, and compressed 4:2:0 is significantly worse (H.264, MPEG-2, VC-1).

Have you tried encoding with ProRes 422 (HQ)? Also, please describe your workflow and how you plan to use the ProRes file.

[Ivan Myles]

I just reread your last post and saw the reference to 5D footage. So that’s H.264/MP4 source footage, yes?

Also, were the Premiere and FCP images shown above taken on the same computer/monitor? If not, the issue could be related to the graphics card, monitor, and/or look-up table.