These CUDA filters are packaged into DGDecodeNV, which is part of DGDecNV.
tormento
Posts: 727 Joined: Mon Sep 20, 2010 2:18 pm
Post
by tormento » Mon Jul 04, 2022 2:37 pm
Rocky wrote: ↑ Mon Jul 04, 2022 8:27 am
Can you please explain the use case and justification for this request?
AFAIK you can recover full chroma from double the required resolution, such as per post
https://forum.doom9.org/showthread.php? ... ost1938846
I was asking myself if all that procedure could be ported inside some of the DGTools libraries.
Bullwinkle
Posts: 309 Joined: Thu Sep 05, 2019 6:37 pm
Post
by Bullwinkle » Mon Jul 04, 2022 6:39 pm
Why do you want to make 1920x1080 4:4:4 material? What players will support it?
tormento
Posts: 727 Joined: Mon Sep 20, 2010 2:18 pm
Post
by tormento » Tue Jul 05, 2022 3:53 am
Bullwinkle wrote: ↑ Mon Jul 04, 2022 6:39 pm
Why do you want to make 1920x1080 4:4:4 material? What players will support it?
Any nVidia based PC, as last 2 generations of cards support HEVC 4:4:4.
Anyway, it's not a priority. Many other things are needed first
Dion
Posts: 25 Joined: Sun Dec 04, 2016 12:30 am
Post
by Dion » Thu Jul 21, 2022 11:39 pm
Did DGHDRtoSDR_1.15 get a different tonemap algorithm vs DGHDRtoSDR_1.14? Results seem very different.
Bullwinkle
Posts: 309 Joined: Thu Sep 05, 2019 6:37 pm
Post
by Bullwinkle » Fri Jul 22, 2022 7:00 am
No change to the tonemapping. All changes are documented in the Binaries Notification thread.
* HDRtoSDR white parameter (PQ) was made nonlinear to better represent the interesting parameter range.
* HDRtoSDR white parameter (PQ) range was widened and the default is now 2000.
Dion
Posts: 25 Joined: Sun Dec 04, 2016 12:30 am
Post
by Dion » Fri Dec 09, 2022 2:06 am
Mega late reply but I found the reason. Seems the "default" values for .14 and .15 are different.. I just had to adjust .15 back to .14's default values and the results are now identical.
tormento
Posts: 727 Joined: Mon Sep 20, 2010 2:18 pm
Post
by tormento » Fri Dec 09, 2022 3:50 am
Dion wrote: ↑ Thu Jul 21, 2022 11:39 pm
Did DGHDRtoSDR_1.15 get a different tonemap algorithm vs DGHDRtoSDR_1.14? Results seem very different.
It's way better to use a proper LUT and DGCube. At least for me and my results.