Major quality difference between FCP X and Premiere Pro with regards to compression

[Shawn Miller]

[Noam Kroll] “Hey Guys –

Just thought I would share this article with all of you. I’ve noticed that the quality of my Premiere Pro/Media Encoder outputs are significantly lower than the outputs coming from FCP X/Compressor when using identical H.264 settings.

Not sure if anyone else has noticed this, but I posted some frame grabs on my website here: https://noamkroll.com/there-is-massive-quality-difference-between-fcp-x-prem...”

Which profile and level are you using, is it the same for both programs?

Shawn

[Walter Soyka]

What version of AME did you use for your test?

There was a serious bug affecting H.264 in AME CC 2014 8.0, where “complex video assets produced [a] blurry image every 1 second.” This was fixed in AME 8.0.1, released a few weeks ago.

https://blogs.adobe.com/aftereffects/2014/07/adobe-media-encoder-cc-2014-0-1-update.html

Walter Soyka
Designer & Mad Scientist at Keen Live [link]
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
@keenlive   |   RenderBreak [blog]   |   Profile [LinkedIn]

[Dennis Radeke]

Hi Noam,

I’m sure your end results are correct but without posting an almost exhaustive list of your settings on both programs and your methodology for testing, it’s just too subjective. I’d welcome the opportunity to see this article with all of that information presented. If you did that, I’d be happy to forward on to AME engineering as we are always striving to improve our quality when we can. Quantifiable data is always the best way to do that.

Dennis – Adobe guy

[Craig Seeman]

As a compression geek there’s very little useful technical information to make any judgement at all.

Profile, Entropy, keyframe rate, number of passes, VBR or CBR, GOP structure are all factors.

Generally (with some exceptions) Apple’s Compressor now does best with High Profile, CABAC, One Pass (two pass not an option with these settings). That seems to come reasonably close to x264 quality.

For the longest time MainConcept, used by Adobe, has been far superior to Apple’s H.264 at comparable settings but Apple has been working on things and Compressor 4.1.x H.264 is not bad.

[Tangier Clarke]

Maybe take a look at this, just released:

https://www.larryjordan.biz/video-compression-speed-test/?utm_source=Larry%27s+Newsletter+-+140818&utm_campaign=8%2F18+Newsletter&utm_medium=email

Tangier

[Dave Jenkins]

Larrys test where done on a 2012 iMac which I believe uses the
NVIDIA GeForce GTX 660M graphics processor with 512MB of GDDR5 memory
or
NVIDIA GeForce GTX 675MX graphics processor with 1GB of GDDR5 memory
Configurable to NVIDIA GeForce GTX 680MX with 2GB of GDDR5 memory.

It would be interesting to try this on a computer with an ATI graphics card.

Dajen Productions, Santa Barbara, CA
Mac Pro 3.5MHz 6-Core Late 2013
FCP X

[Tangier Clarke]

Yeah, I was curious to know how this test would fair on my Mac Pro (the new one). Before his test came out I did a very quick test of my own coincidentally between AME and Compressor , but I was using CS6. I am Compressor came out about 37% faster getting ProRes 422 to H.264 ready for vimeo.

I am going to run some tests similar to Larry on my Mac Pro to see what results I get. I favor compressor, but am very interested in AME if it’s truly and consistently this fast.

Tangier

[Tangier Clarke]

Hey folks, just a quick question or rather wanted some insight. Being that compressor can create a custom preset by dropping in a quicktime. Would it be a fair test to create transcodes first using AME, then dragging those resulting clips into compressor to create comparable settings?

There are some other distinct things go consider as well:

AME allows choosing GPU or software encoding.
Compressor you can choose the number of instances.

I am not certain I can get 1:1 parameter settings situated such as compressor’s entropy mode.

Your thoughts…

Tangier

[Tangier Clarke]

By the way, my first sample of that type of test looks like this:

Computer:
Mac Pro Late 2013/ 3.5 GHz 6-Core Intel Xeon E5/ 32 GB Memory/Dual AMD FirePro D700

Source File:
Apple ProRes 422 Quicktime/11m11s/12.26 GB/1920×1080 @ 29.97 fps/Linear PCM 24-Bit 48k Hz Audio

AME CC 2014 build 8.0.1.48 Transcode
Using default Youtube 1080p HD setting and GPU Renderer
Result = H.264 .mp4 Quicktime/ took 9 minutes to encode/1.36 GB file

Compressor 4.1.3 Transcode
Dragged above AME result file to Compressor so it would assume the settings of the file as a custom preset
Result = H.264 .mp4 Quicktime/ took 7 minutes to encode/1.15 GB file

AME file has slightly more contrast
Compressor result file looks more like my original file.

Other things I am not sure about I’d like to find out:
Are either of the apps 64-bit?
Does compressor use GPU at all?

Tangier

[Tangier Clarke]

Addendum:

I am using Mac OS 10.9.4 (13E28)

Compressor 4.1.3 is not 64-bit

Adobe Media Encoder CC 2014 8.0.1.48is 64-bit

Compressor is not GPU aware.

Tangier