Log in

View Full Version : "Film" Tune for x265 - Please Help!


Pages : 1 [2]

GeoffreyA
26th March 2026, 10:31
https://slow.pics/s/r5QqFtq9

From what I can tell, SAO is adaptive. The bigger the bitrate, or lower the CRF, the smaller effect it has.

microchip8
26th March 2026, 11:19
https://slow.pics/s/r5QqFtq9

From what I can tell, SAO is adaptive. The bigger the bitrate, or lower the CRF, the smaller effect it has.

hence why it's called Sample Adaptive Offset

GeoffreyA
26th March 2026, 15:08
hence why it's called Sample Adaptive Offset

Sometimes, we forget that it isn't Smooth All Objects :)

benwaggoner
27th March 2026, 00:00
I use a strength of 1.1

I've never been able to get AQ3 do its thing in what it claims - improve dark scenes with low energy. Not even in x264 did I notice a big benefit. It just wastes bitrate for not much improvement, if at all
And is really useful only with 8-bit SDR in any case. 10-bit reduces the utility a bunch, and it's actively counterproductive with HDR because PQ is perceptually uniform and doesn't need any dark bias.

benwaggoner
27th March 2026, 00:03
for me, the goal is to get the look/feel of x264's film tune, but with the efficiency of x265, so isn't that a good thing?
Yes. But x265 and HEVC itself are better tuned for filmic content out of the box. For example 0,0 for in-loop deblocking isn't as softening as the H.264 equivalent, which had been tuned more for low bitrate low resolution streaming scenarios. Gary Sullivan told me once that the default alpha/beta being too strong was one of his bigger H.264 regrets (a testament to how amazing it was for its era).

Z2697
27th March 2026, 04:49
And is really useful only with 8-bit SDR in any case. 10-bit reduces the utility a bunch, and it's actively counterproductive with HDR because PQ is perceptually uniform and doesn't need any dark bias.

With or without --hdr10-opt?
As you can see the hdr10-opt is a fixed QP offset per block based on average luma level.

benwaggoner
27th March 2026, 20:55
With or without --hdr10-opt?
As you can see the hdr10-opt is a fixed QP offset per block based on average luma level.
aq-mode 3 is bad for hdr10 in all cases. --hdr10-opt is spiritually similar in effect, and should be used with hdr10 encoding (not HLG or Dolby Vision Profile 5, though).

It's crazy to look back on the limited set of tools I had available when Prime Video launched HDR in Q2 2015. The readthedocs.io release notes don't even go back far. Version 1.7 or 1.6 IIRC. --hdr10-opt (originally --hdr-opt) didn't arrive for two more years.

FoodOcean
31st March 2026, 15:25
And is really useful only with 8-bit SDR in any case. 10-bit reduces the utility a bunch, and it's actively counterproductive with HDR because PQ is perceptually uniform and doesn't need any dark bias.

Are you saying that AQ3 is useful when working with 8-bit source content, but that it is counterproductive when encoding into 265 10-bit?

(ex: h.264 8-bit film -> x265 10-bit = no need for AQ3?)

FoodOcean
31st March 2026, 15:30
Yes. But x265 and HEVC itself are better tuned for filmic content out of the box. For example 0,0 for in-loop deblocking isn't as softening as the H.264 equivalent, which had been tuned more for low bitrate low resolution streaming scenarios.

I have heard that, but if that is true, why does it decidedly look smoother/not has textured? I know their algorithims are different and x265 is designed to get the most out of the least amount of bits, but if it was designed to be more "film" like than base x264, why does it loose that feel? Or am I not fully understanding what the "film" tune is intended for?