> On 29 Dec 2025, at 02:25, Richard Bartczak via ffmpeg-user > <[email protected]> wrote: > > Hello, > > once again a new try with additional information. > > I did try to compress h.262 videos to h.265 and would expect, if I compress > it to half of the size a comparison e.g. with vivictpp would give a result, > that nearby no difference will be seen. > > (Refer also to > https://netflixtechblog.com/vmaf-the-journey-continues-44b51ee9ed12 . In this > context "bad" for me does mean a vmaf score <= 85.) > > > But its not. The following command I use (crf 20 only as example) : > > Code: > > #!/bin/sh > for f in *.ts; do ffmpeg -y -i "$f" -vcodec libx265 -crf 20 -preset slow -vf > nlmeans="1.6:7:5:3:3" -c:a copy "$(basename "$f" .ts).mkv";done > > ffmpeg info of an example video : > > Stream #0:0[0x100]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), > yuv420p(tv, progressive), 720x576 [SAR 64:45 DAR 16:9], 25 fps, 25 tbr, 90k > tbn, start 0.124000 > The videos all ways have a non square pixel format. > > Are there any special vf parameter to handle this pixel format ? > > Any ideas to improve it ? > > (Beneath, its not a question for people who live in a lossless video world, > also to say h.265 is a lossy codec is wrong, crf 0 leads to a lossless result)
Bullshit. H.265 IS LOSSY. Not sure where it states that a value of 0 is lossless, but at least all other values are lossy. (H 264 says that the best quality is ‘visually lossless’, that means not technically lossless.) I’ve told you before to omit the noise reduction and the de-interlacing. So you do listen and have omitted the de-interlacing, but if you add noise reduction, not even a lossless codec will give you an ‘identical’ file. How could it if you re-arrange pixels? So test without the NR, and do not only trust testing software, but look/trust your eyes. No matter what Netflix says. (I know Netflix, they have a gazillion quality standards and they sometimes sell horrible stuff that would never have made it outside my studio walls.) For the aspect, there is a filter to set the pixel format, or scale to a non-square pixel format. Now I’m going to make it way more complex. If the origin is SD video, you also need to crop as SD was an over scanned format, so the edges were never intended to be seen on the screen. (Hence you often can see a mic dipping in the image. This is NOT a mistake of the sound guy / boom operator, it shows his quality that he gets as close as possible without being in the active image.) Bouke > > Thanks > _______________________________________________ ffmpeg-user mailing list -- [email protected] To unsubscribe send an email to [email protected]
