Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.


Go Back   Doom9's Forum > Hardware & Software > Software players

Thread Tools Search this Thread Display Modes
Old 18th May 2024, 13:43   #10581  |  Link
Registered User
Join Date: Oct 2012
Posts: 7,976
copyback is always slower. if this is measurable is a different story.
d3d11 in madVR is always slower because it is still dx9 it just does 1 or 2 additional interops which just cost performance. madVR is not a DX11 renderer mpcVR is.

and frequency's obviously big issue for such trivial easy tasks.
huhn is offline   Reply With Quote
Old 18th May 2024, 15:26   #10582  |  Link
Registered User
Join Date: Aug 2020
Location: Bilbao, Spain
Posts: 10
Originally Posted by huhn View Post
looking at the code there is nothing that tells the tone mapping the new maxcll and the display brightness and that's bad because of how tonemapping spec is define.

same file as an example:
125 peak brightness that means a knee point around 54.04 nits(this is the old spec i don't know the new one it's but goes higher).
maxcll 4000

current algorithm uses 10000 and peak 125 the high light 54.04-125 will be inefficeintly compressed and only about 100 nits will be used.

at 4000 the algorithm needs to recalculate the knee point or even the part of 0-54 will be to bright which should be the same no matter the maxcll.

i don't see anything even considering do that in the code.
It doesn't make sense to use 10000 when most movies are 1000 max or less. In practice, MPC-BE/MPC shows all HDR dark. You should use preferably MaxCLL or MaxMDL value from file and map from there using different methods, such as ITU-R Report BT.2446, method A:


In 4.1 paragraph, you have Table 2: HDR to SDR conversion reference functions. They assume Lhdr=1000 nits for HDR and Lsdr=100 nits for SDR, but I guess you could choose the exact value from MaxCLL or MaxMDL.

There is also SMPTE ST2094-40 Annex B which is Dynamic HDR10+ tone-mapping:

Mpv, by means of libplacebo library, can choose between one of these algorithms, using metadata from the file when available. When you have proper metadata from a well mastered file, you use Bt.2446, and when you don't, you use ST2094-40 (dynamic).

Here you have the calls from libplacebo library:

Microsoft provides a HDR to SDR tonemapping implementation built-in in D3D11, which uses maxCLL. Maybe you could use this:

It would be nice that you implemented any of such methods, if possible, to improve actual tonemapping. Thanks

Last edited by wyup; 18th May 2024 at 15:42.
wyup is offline   Reply With Quote
Old 18th May 2024, 16:13   #10583  |  Link
Registered User
Join Date: Dec 2008
Posts: 1,995
For what purpose are you writing this?
Show your patch and people will have something to test.
v0lt is offline   Reply With Quote
Old 18th May 2024, 23:03   #10584  |  Link
Registered User
Join Date: Aug 2020
Location: Bilbao, Spain
Posts: 10
I write this as a user to help improve HDR tonemapping in MPC-BE/MPC.
I don't have a patch, I'm not a programmer. I leave some ideas.

Right now MPC has a problem: HDR looks dark on a normal SDR monitor. Its tonemapping is not good.
MPC-BE/MPC is and has been to me the best overall windows media player. MPV is open source and doing things better. Let's learn from it.

Last edited by wyup; 20th May 2024 at 11:29.
wyup is offline   Reply With Quote

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

All times are GMT +1. The time now is 23:57.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.