Repository navigation
Replies: 4 comments 4 replies
|
@TomNaber , try this build it has this adjustment. Just a note that the logic does not limit to 1000 nits but to the video file source max nits (MDL) which can be different then 1000 nits. In any case, when I provided (PR) the code for this feature to cpm, he made some improvements but it had this side effect, specifically for 1000 nits. On my build attached it goes to the 1000 nits mark. After you update to this build, make sure you go to Settings/System/CoreELEC at the bottom of the screen click on "Reset above settings to default" and then confirm by clicking "yes". If you want to go back to A14, after updating to A14 do the same reset. |
|
@TomNaber , glad to help. Test it out to make sure it is working and, if so, please mark this as answered so it can be closed. |
|
@oourieloo , what sink (display) are you using - brand and model? |
|
@oourieloo , It seems that some Samsung have a problem with the HDMI info which I was capturing to automate this feature. Below is a build that has the option to select the colour space. After you update to this new build, make sure you go to Settings/System/CoreELEC at the bottom of the screen click on "Reset above settings to default" and then confirm by clicking "yes". After the reset you will be able to see and change the colour space. Let me know if by changing the colour space it solved your over saturation problem. |
Uh oh!
There was an error while loading. Please reload this page.
I'm very happy that a workaround for limit max VSVDB luminance has been implemented to avoid tonemapping up for displays exceeding 1000 nits that use player-led DV (i.e., Samsungs). However, yesterday an AVSforum user going by the name of Thoth did a digital capture of the LLDV output with CPM A14 and found that limit max luminance in fact takes the next lower luminance value that can be encoded in a V2 VSVDB data block, which is 896 nits. This results in the content being tonemapped down to 896 nits. The luminance accuracy is then considerably worse (average dEITP is 4.30, std dev 2.25, max 11.56) than when using 1037 nits in the VSVDB (1.60, std dev 1.09, max 7.69).
Could you adjust the logic to use a 1037 nit max device luminance in the VSVDB for 1000 nit MDL content if the device has a peak exceeding 1000 nits?
Thoths posts: https://www.avsforum.com/posts/64055798/, https://www.avsforum.com/posts/64055798/
All reactions