FFAStrans forum

Hello.

I have been testing trying to use FFAStrans to replace an encoder I have been using for many years to make HD (and 4K) MPEG 2 files.

The encode quality and performance I am finding here is very good... but there seems to be a color depth issue. All files seem to be 'not as vibrant' as what I am used to. After further testing I am wondering if perhaps the 4:2:2 switch is not working. If I do encode the same exact file with all the same settings but vary this pix format command like so:
-pix_fmt yuv422p and -pix_fmt yuv420p I get 2 exactly same files that are exactly the same size.

Am I missing a switch somewhere that will use Set Chroma to 4:2:2?

Thanks much,

Keith

Are you using the "Custom FFmpeg"-encoder? If so, what are your swithes and what does the rest of your workflow look like?

-steipal

Workflow is simple. 1 Folder Input (I am manually submitting files for processing) --> Custom FFmpeg (settings below) --> Folder output

WorkFlow
https://www.dropbox.com/s/knrrqi63drdg6 ... 0.JPG?dl=0

Custom FFmpeg
https://www.dropbox.com/s/qavtyz81s862p ... 1.JPG?dl=0

Options line in encoder:
-threads auto -pix_fmt yuv422p -bf 0 -b:v 25000k -minrate 25000k -maxrate 25000k -g 15 -bf 0 -an
(only difference between the 2 processors show here is the bottom one changes to fmt_yuv420p)

Thanks for checking into this.

Keith

Eh, when you're forcing 25k bitrate for the video then you will get 25k bitrate, no matter what your pixel format look like. It's a feature, not a bug;-)

-steipal

OK - agreed, the file size should be the same, but shouldn't the colors look different in 422 vs. 420?
I am comparing and they are exactly alike.

thx.

No, the colors should be exactly the same.
The only difference between 420 and 422 is that with 422 the chrominance is sampled every 2 pixels and with 420 every 4 pixels, so you have larger "chroma blocks".
You can notice the difference in diagonal hard edges between areas of different colors.

What kind of "vibrant color" codec and software are you used to?
Are you shure not to have a 0-255 vs 16-235 issue with your playback software or graphic board settings?

Interesting - and yes, I now concur. After your last message yesterday I did a bunch more testing. Other encoders also have no difference between 420 and 422 when encoding this file either. I have been using Canopus ProCoder for many years, and have been comparing all encodes to that as my baseline. I remember a drastic difference when switching to 422 many years ago - and for some reason I was sure that was the difference here. No so, and with your description, I now remember the difference was all along the edges of elements.

Looking deeper it seems that ProCoder is actually adding MORE vibrance (when compared to the original). The FF output (while still a tiny bit less vibrant than the original) is actually quite a bit closer to the original than the ProCoder output.

So - First, thanks for the lesson. Second, please accept my apology for posing in the 'bug' section. I'm sorry

-kt

No problem:-) It's a nice thing when users learn something. I'm currently on vacation so I have not been very active on the forum so thanks to screenshaper for helping out.

-steipal