View Full Version : Handbrake settings
Braciata
15th July 2023, 18:52
Hello.
I wonder what you think about these settings handbrake put in the presets:
strong-intra-smoothing=0:rect=0:aq-mode=1:rd=4:psy-rd=0.75:psy-rdoq=4.0:rdoq-level=1:rskip=2
Thank you
microchip8
15th July 2023, 19:38
I would enable rect, change aq-mode to 3 and increase psy-rd to 4 and psy-rdoq to 15 (the latter two highly reduce banding, even in 10-bit mode)
Braciata
16th July 2023, 08:33
I would enable rect, change aq-mode to 3 and increase psy-rd to 4 and psy-rdoq to 15 (the latter two highly reduce banding, even in 10-bit mode)
Thank you.
I am trying it on 1080p. I'd give it a try
benwaggoner
18th July 2023, 17:43
Thank you.
I am trying it on 1080p. I'd give it a try
Note that --aq-mode 3 can increase bitrate quite a bit, if that's a concern. It's a useful tool, but pretty heavy-handed in implementation.
Braciata
19th July 2023, 12:06
Note that --aq-mode 3 can increase bitrate quite a bit, if that's a concern. It's a useful tool, but pretty heavy-handed in implementation.
Yeah, I noticed it.
I am fighting with grain.. I am not sure if I want to keep it or retain it!
I think that x264 retains better. Grain with x265 is strange, different from the original movie. I have to make a choice.
Thank you for your answer.
microchip8
19th July 2023, 15:12
Yeah, I noticed it.
I am fighting with grain.. I am not sure if I want to keep it or retain it!
I think that x264 retains better. Grain with x265 is strange, different from the original movie. I have to make a choice.
Thank you for your answer.
Lower deblock to something like -3,-3 and disable all SAO optiomizations. Should be better preserved that way.
A dirty way to compensate for AQ3 is to increase CRF by a few to counter the increased bitrate. AQ3 basically shifts bits around and allocates when needed due to its bias. I personally encode with AQ3 1080p content at a CRF of 24 and am pretty pleased, both visually and with file size.
That said, I find myself going back more and more to x264 for a few reasons. First, as you mentioned, x264 better retains details while x265 needs a lot of tweaking to roughly get the same level of detail retention, and it is still a bit lower than x264. Second, on my 12c/24t CPU, I can encode 1080p content with 1.5 and sometimes 2x realtime, so an encode is done in less than the runtime of the movie. (Yes, I'm a bit impatient and certianly don't want to wait many, many hours for a single encode). Using x265 with my settings, at most I can get an encoding speed of ~13 fps (Vs. 35-40 fps for x264)
Braciata
19th July 2023, 22:37
Lower deblock to something like -3,-3 and disable all SAO optiomizations. Should be better preserved that way.
A dirty way to compensate for AQ3 is to increase CRF by a few to counter the increased bitrate. AQ3 basically shifts bits around and allocates when needed due to its bias. I personally encode with AQ3 1080p content at a CRF of 24 and am pretty pleased, both visually and with file size.
That said, I find myself going back more and more to x264 for a few reasons. First, as you mentioned, x264 better retains details while x265 needs a lot of tweaking to roughly get the same level of detail retention, and it is still a bit lower than x264. Second, on my 12c/24t CPU, I can encode 1080p content with 1.5 and sometimes 2x realtime, so an encode is done in less than the runtime of the movie. (Yes, I'm a bit impatient and certianly don't want to wait many, many hours for a single encode). Using x265 with my settings, at most I can get an encoding speed of ~13 fps (Vs. 35-40 fps for x264)
That's what I meant. x265 is more consuming and even though files are not big, x264 seems better to my eyes. If I want a good result I have to go with x265 at CRF=18. The result is a big file. So maybe for 1080p x264 is still better and quicker.
benwaggoner
20th July 2023, 02:11
To improve grain retention/quality with x265, I recommend
using some --nr-inter. 200 could be a good starting point
Increase --psy-rd and --psy-rdoq. 2 and 3 are a good starting points
Try --rdoq-level 1
Try --rd 4 instead of --rd 6
Use --rskip 2, with --rskip-threshold 3 to start with
Try lower --ipratio and --pbratio (1.2 and 1.1 are good to start)
Braciata
20th July 2023, 12:46
To improve grain retention/quality with x265, I recommend
using some --nr-inter. 200 could be a good starting point
Increase --psy-rd and --psy-rdoq. 2 and 3 are a good starting points
Try --rdoq-level 1
Try --rd 4 instead of --rd 6
Use --rskip 2, with --rskip-threshold 3 to start with
Try lower --ipratio and --pbratio (1.2 and 1.1 are good to start)
Thank you. I will give these settings a try.
I hate rdoq-level=2. I love how rdoq-level=0 'cleans' the movie but then it's strange on TV! I will try 1
A1
30th August 2023, 16:02
To improve grain retention/quality with x265, I recommend
using some --nr-inter. 200 could be a good starting point
Increase --psy-rd and --psy-rdoq. 2 and 3 are a good starting points
Try --rdoq-level 1
Try --rd 4 instead of --rd 6
Use --rskip 2, with --rskip-threshold 3 to start with
Try lower --ipratio and --pbratio (1.2 and 1.1 are good to start)
x265 needs to have a universal quality of attention to the universal standard, even if this standard under the same quality volume is only 10% less than x264, I believe there will be people use.
When you find out that you've spent more time researching x265 than running it on your computer, x265 is frowned upon.
benwaggoner
30th August 2023, 16:32
x265 needs to have a universal quality of attention to the universal standard, even if this standard under the same quality volume is only 10% less than x264, I believe there will be people use.
When you find out that you've spent more time researching x265 than running it on your computer, x265 is frowned upon.
No argument that --tune grain needs a full refractor to reflect current x265 features and best practices!
And the basic --preset options themselves are due updates as well, as it has been years and lots of newer features that should be defaults aren't defaults, like --selective-sao 2.
A1
31st August 2023, 18:32
No argument that --tune grain needs a full refractor to reflect current x265 features and best practices!
And the basic --preset options themselves are due updates as well, as it has been years and lots of newer features that should be defaults aren't defaults, like --selective-sao 2.
I feel that in order to greatly lead x264 in terms of compression efficiency, x265 is at the cost of quality stability and user learning costs, but compression efficiency should not be the only indicator considered.
Most home users of x265 don't need a huge lead in compression efficiency, they only need quality stability and don't need to keep debugging for the type of video.
Only streaming media with a high volume of playback prefers compression efficiency. When the video volume is large and the network fee increases a lot, it needs to be paid attention to. This is the work of professionals in large companies, not users who store videos at home.
benwaggoner
31st August 2023, 23:07
I feel that in order to greatly lead x264 in terms of compression efficiency, x265 is at the cost of quality stability and user learning costs, but compression efficiency should not be the only indicator considered.
Most home users of x265 don't need a huge lead in compression efficiency, they only need quality stability and don't need to keep debugging for the type of video.
Yeah. And the built-in --preset and especially --tune of x265 aren't really fire-and-forget like x264's --preset veryslow --tune film.
A tool like Handbreak certainly could ship with better-tuned settings better suited for its core scenarios.
Only streaming media with a high volume of playback prefers compression efficiency. When the video volume is large and the network fee increases a lot, it needs to be paid attention to. This is the work of professionals in large companies, not users who store videos at home.
Having broad support for 10-bit and HDR encoding is another big reason people use HEVC. Also the efficiency gains are a lot more at 4K resolutions.
Ritsuka
1st September 2023, 08:28
Patches to improve HandBrake x265 would be very welcomed, unfortunately there are only a few people working on HandBrake, so there isn't much time to spend on x265 presets.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.