> The tone-mapping algorithm place 100% load on a single CPU core.<p>In 4.1, a OpenCL variant filter was added for GPU processing.<p>In 4.2 threading support was added to the tonemap CPU filter.<p>In 4.3, a VAAPI variant filter was added for GPU processing on linux.
HDR/SDR Tone mapping has long been the bane of hardware transcoding for Plex. Even when it "kinda works" which is Nvidia on Linux only it kills my GTX 1650 in RAM and compute required. Normally NVENC can handle 3 4k Streams or many many 1080p streams, only constrained by RAM.
Does anyone know how well this works?<p>I have a real-world use case: I have a full Atmos audio setup (7.1.4) and a projector for video. Now projectors don't do HDR very well[1], and to be honest I'm more interested in immersive audio than shiny highlights in the picture.<p>Unfortunately, a good few film disc releases only put the Atmos audio on the UHD disk along with HDR and running those disks through my rig and forcing SDR yields a picture that is too dark compared to the original HD SDR copy (and letting it try HDR isn't much better).<p>So I'm wondering if this will enable me to rip and convert to a good SDR UHD copy of a film with the Atmos audio ?<p>--<p>[1] ...and 100 inch OLED TVs are stupidly expensive.
ffmpeg would be a great candidate for use with natural language processing, a la<p><a href="https://www.autoregex.xyz/" rel="nofollow">https://www.autoregex.xyz/</a>
does anyone know if DV oriented tonemapping is in development? (i.e. DV to HDR10 or DV to SDR). so DV files can play on non DV screens and look correct.