Posts by Katana

    Great guide and YouTube channel @morphinapg 😉 as I also always aim for the best quality for my content I want to start with uploading 4KHDR in 2021.
    I had a lot of problems with limited and full (I care if my blacks are crushed even if there are people out there who likes it) when I started uploading regularly in 2013 😂 Now it feels like 2013 all over again 😂 even worse... So before I break my head again for fully understanding everything I go with the try and error approach.

    As I play on a PC Monitor and don't have any HDR display (no really good monitor available yet) I cannot verify my files which makes it even harder.

    So I got a PS5 (yeah they really exist and not only bots, influencers and scalpers got one 😂)
    So I did some testing for my upcoming walkthrough.

    I'm using Magix Vegas but I think for the Output it doesn't matter that much.

    So lets start (videos are still encoding on YT so 4K may or may not be available at some point, however YouTube decides to encode them or not)
    First off I just wanted to see the quality of the internal PS5 4KHDR capture of the PS5.

    Unedit File PS5 Internal File ~40Mbps:

    Looks "limited"?

    Unedit external 100Mbps capture:

    I think this looks also ok but maybe a little bit too dark? Maybe I have to force the PS5 to limited. Or other Youtubes increased their Gamma 🤔

    Magix Vegas specific:
    So I went with my external capture and just tried to let the output file look identical to the unedited input file. When I choose HDR10 in Vegas I cannot use Voukoder which isn't an option in my Opinion. So I didn't choose HDR10 but instead switched the project to 32-bit floatingpoint (video levels), with the other option which is applied with HDR10 (32-bit floatingpoint (full range)) the video in the preview looks too dark.

    I used the setting of the starting post. I couldn't get the x265 encode to work. It started but never rendered anything?!
    So I went with NVENC (low 40Mbps just for testing purpose) and now I have a different problem (or lack of understanding).
    What about the "Limited" Option? I know that correct should be limited (however my external record was full?! It seems I cannot really control that, maybe I should use a different recording software)
    I tested both and even MediaInfo tells me full/limited the files on YouTube looks identical? I think YT converts them correctly no matter what I choose. I think now both are limited, didnt' redownload and test.
    Why is with HDR there still limited/full? Thought this was only because of 0-255/16-235 and this doesn't apply to HDR which is 0-1024



    And additional I think both files now look like the input. Again I can only see the SDR result and cannot really verify.

    So my first question is would this workflow be ok (either lmited or full encode)? I cannot choose MaxCLL and MaxFALL as it isn't available with NVENC.

    Second why is my footage a little bit too dark? Because of being a "full" source file, or are others set their game brighter. 🤔

    Why does the internal capture of the PS5 look so "washed out"?

    Any help how to verify my workflow and the output files would be appreciated.

    As the title suggests, I would like to aim for a certain maximum file size while maintaining the best possible image quality.

    Normally, Constant Quantizer is used, but since the video is very long, this is not an option.

    For testing purposes I took 1min (=60sec or 3600frames) video and aim for 10MByte (= 10240KByte = 81920kbit).

    The easiest approach for a certain file size would be CBR

    Audio 256 kbit/s => 256*60 = 15360 kbit

    Video (81920 - 15360)/60 = 1109 kbit/s (encode a little less for being on the safe side, so 1100)

    HVENC (NVIDIA NVENC) then gives me a file with 10.199.040 bytes (9960 KByte) at the end. So far so good.

    But since the bitrate is constant it will not give the best picture quality for 10MByte as a result.

    If I try the whole thing (Video 1100 Audio 256) with VBR I get "something" at the end (23 887 872 bytes)

    I used to use MeGUI, here I can specify a desired file size when encoding and the program will do a 2-pass encoding with VBR in the background. I don't have to specify anything except the file size and the encoder.

    Since I would like to encode on the GPU for speed reasons I try to reproduce the same in Voukoder and fail miserably so far. I am aware that encoding on the GPU gives worse results anyway but then I want to get the best possible results.

    So what would be the best setup to get a specific maximum file size? (I'm working with Magix Vegas Pro 16 if this is of any interest)