anyone try the quick sync handbrake beta?

[deadrats]

here i sit, bored, not having an hvec encoder to play around with so i decided to try the qs capable handbrake beta and i'm quite surprised by the results. i've used every qs enabled app: espresso, video mastering works, sorenson squeeze, gom encoder, media coder but i had never tried the handbrake implementation, maybe it's my eyes but these guys seem to get better quality than all the others (sorenson gets the worst, by far, and that holds true of all the included encoders).

i know there's a few guys here that have SB/IB chips and use the integrated gpu's, if you guys could try out the handbrake beta and report if you find the quality likewise surprising, that would be great.

http://sourceforge.net/projects/handbrake/files/Betas/QSV/

[jagabo]

Yes, I played around with it a bit. On my i5 2500K, x264 at the veryfast preset took about twice as long to encode as QS. And, as usuaul, the QS encode was lower quality at similar bitrates.

It has some bugs -- the autocrop feature wasn't working properly.

[johns0]

On my 2600k on best quality setting on a 720p video was around 250fps encoding.Quality was almost as good as 2 pass.

[deadrats]

Yes, I played around with it a bit. On my i5 2500K, x264 at the veryfast preset took about twice as long to encode as QS. And, as usuaul, the QS encode was lower quality at similar bitrates.

It has some bugs -- the autocrop feature wasn't working properly.

did you try setting the same crf value for both encoders and then comparing size as well as quality.

i'm running such a test now, it'll be interesting to see if 1080p at crf 18 results in the same size.

[jagabo]

did you try setting the same crf value for both encoders and then comparing size as well as quality.

I did but I didn't keep notes on that. The QS encoder delivered much larger files at the same CRF value. Quality was close if I remember correctly. I wanted to compare bitrate-for-bitrate. Here are some notes I took:

x264 slow      crf18:  30 fps, 2.50 GB
x264 veryfast  crf18:  70 fps, 2.45 GB
x264 superfast crf18:  80 fps, 4.8  GB
qs, best, crf20:      164 fps, 2.69 GB
qs, best, crf21:      164 fps, 2.41 GB

I was using a 1080p h.264 Blu-ray rip, about 25 Mb/s, downscaled to 720p. The QS encodes lost lots of small, low contrast detail. I thought I kept the resulting files but I can't seem to find them.

[deadrats]

x264 slow crf18: 30 fps, 2.50 GB x264 veryfast crf18: 70 fps, 2.45 GB x264 superfast crf18: 80 fps, 4.8 GB qs, best, crf20: 164 fps, 2.69 GB qs, best, crf21: 164 fps, 2.41 GB

are you sure the superfast results are not a typo? maybe 2.8 GB instead of 4.8?

[jagabo]

I'm pretty sure the numbers I posted are correct. It's always been my experience that superfast delivers 2x to 3x large files at CRF=18. I just encoded a random 10000 frame sample of the same video at veryfast and superfast and got approximately the same ratio. This was with live action footage, not animation.

[deadrats]

then my question becomes what were the files sizes of the QS encode if you used crf 18 instead of 20?

[jagabo]

then my question becomes what were the files sizes of the QS encode if you used crf 18 instead of 20?

I didn't keep track of that and I don't remember. As indicated, I was evaluating image quality at similar bitrates -- since RF values in different programs don't compare. I encoded CRF=18 in x264 first, then ran several QS encodes looking for CRF values that would give similar bitrates. Then I compared the quality of those files.