Posts by CroqueMr

    Thanks for the feedbacks, I am working on a more precise conversion a real LLDV over HDR10, it will replace HDR expert conversion at target.

    Overkill is my middle name. This is literally my 6th external player in my home theater cabinet. Also why I introduced myself as parachuting in from AVSforum where overkill is a way of life. ;)

    I don't want any test failures with the GUI on (see the results  CroqueMr posted). The more horsepower I have that can just brute force that, the better. FTR - and I think this point seems to be lost on folks used to inexpensive Amlogic devices - DV7 FEL is not computationally easy. Doing it in software is so taxing even a 3rd Gen Apple TV 4K can't do it and Atmos height channels simultaneously:

    I want to be able to do 4K60 (e.g. Gemini Man) with the GUI and all audio channels.

    FEL is not a big issue, and we are not doing the maths in software/CPU, we use the shader of your iGPU to upscale the FEL and process all the picture. The big bottleneck is the 60fps. I will add the 7430U to my benchmark to have a better lock on requested performance to have no drops even with GUI

    FYI : From now on, I'll gradually add fixes and features to the next build. Once enough changes have accumulated, I'll clean up the code, optimize everything, and push a new release.

    The RC3 preview is already available here https://github.com/CroqueMr/libre…0.0-RC3-Preview

    Feel free to check it out and let me know if it meets your expectations, especially the CM 2.9 to 4.0 information, HDR10+ settings, stability fixes, etc.

    Changes since R1.0.0-RC2

    Dolby Vision

    • Added GPU bilinear FEL upscaling, replacing the nearest-neighbor fallback when VAAPI scaling is unavailable.
    • Added the Reference GPU FEL filter as a selectable option for other conversion modes. Master Reference retains its reference filter.

    Playback and stability

    • Fixed cases where stopping playback during a resolution change could leave HDR signaling active instead of returning to SDR.
    • Fixed stale HDR10 session reuse after a rejected output activation, ensuring subsequent playback prepares the required output surface.

    Settings

    • Separate remembered rendering preferences for Dolby Vision, HDR10 and HDR10+ inputs.
    • Automatic fallback no longer replaces the preferred rendering mode.
    • Fixed rendering options incorrectly remaining unavailable after stopping a different source format.

    HDR10+

    • Fixed source-specific preference selection during continued HDR10+ playback, preventing accidental use of the plain-HDR10 preference.

    Player information

    • Corrected misleading "Unused blocks" reporting: metadata forwarded to the TV is no longer marked unused.
    • Displays whether CM2.9-to-CM4 conversion is actually active, including live updates for Auto and Auto 2.

    Blu-ray menus

    • Updated BD-J Java add-on packaging with matching libbluray 1.5.0 support files.
    • Prevented stale or incomplete Java support files from being packaged. Requires the updated JRE add-on attached to this release.

    No worries, HDR10+ is not yet perfect and some of your issues as already been fixed for RC3. Normaly last settings is saved and will be applied to your next playback. And if the rendering output is not correct you can switch the mode on the fly during your play back, press "O" twice to have the quick pannel settings

    For SDR and limited video the fix will be simple.


    For the support of your NUC please open an issue on github it’s much easier for me to keep track of everything this way. If necessary, I just finished my build with remote access that allows me to make on-the-fly fixes and investigate any hardware issues directly on your minipc. This build is private, of course, and includes all the required consent disclosures. We can go that route if needed and if you agree.

    I need to check, maybe it will be a fork not supported once the stable release is finished

    Thanks for the kind words,

    Yes 60fps is difficult to render, but the two most important circuits are perfectly stable on the two small CPU : Master Reference & DV to HDR10 Expert. I’ve spent the most time optimizing them, if I have a bit less support work to do, I could push the optimization further for the rest, but it will not be a game changer

    Found a little bug but it's related to using plex4kodi like I do, when I start my HDR10+ video from the beginning and ask for hdr10+ to dv the video does not start. If I resume it works and if I play from beginning or resume directly in kodi it works.

    Also, when the credits started because I let the movie play while doing something in the background, at some point I got a message that HDR10+ not available or something and it went back to standard HDR10.

    Finally, in the player process infos, when you ask for hdr10+ to dv, even on HDR10 normal I get messages that says it could not do DV Reference AI. The playback seemed ok in normal HDR10.

    Can you send me the log with all your issues inside? I will check what I can do.

    Here is a short benchmark to give you an idea of how two small CPUs perform. Take a look at the specific circuits you're interested in. I’ve included a table covering the GUI, it’s not a huge deal for me personally, but apparently some people really hate experiencing frame drops when a Kodi graphical element is on screen. (these aren't the subtitles, it is Kodi overlay)


    R1.0.0 DV RC3 Preview: Intel N100 and AMD 3200U without GUI
    R1.0.0 DV RC3 Preview: Intel N100 and AMD 3200U with GUI

    Glad you're happy with the addition.
    For HDR10+ conversion to DV, this is a first approach, I need to refine my algorithm, It doesn't meet my standard, we lose some informations (It’s normal, we’ll always lose some, but I’m sure we can do better). I think, I will train a dedicated AI, like I did for HDR to DV conversion and add the model in the build.

    I will work on HDR10+ deeply next week.

    And also a complet rebuild of the player info panel is planned, this was just some small adjustment

    hadesdevls  CroqueMr

    ok so now both forks are getting stuck, Hades installs itself and after the reboot the same window with the LibreELEC version gets stuck.

    The only thing that changed(since the last time I successfully started Hades with zero problems), is that I installed the newest hotfix NVIDIA drivers and I removed the overclock on my GPU.(so one Windows thing and something that should increase stability, not create more problems...)

    Could it be the driver update that causes these problems?

    The only .log that was created with Hades fork is the "fs-resize.log", I'm gonna assume that's just for the install start and not the real first run?

    No it's not linked, I used some code from Hades in the RC2, this is why we have the same behaviour, I will check with my 5090 this week what is happening, it will be fixed for RC3.

    Quick question: Does the YCbCr vs IPT only change the format sent to the TV and not the PQ? Is the PQ completely identical? DMDReviews seemed to say the standard way is to sent IPT so I switched it to that but not sure if it changes anything in the end?

    The selector YCbCr / IPT is only available for Master Reference output, Standard output (now called Reference) use YCbCr

    To be simple, if you select (except for Profile 5 that is always send in IPT because the file is in IPT) :

    - IPT as prefered output, in master reference the conversion YCbCr -> IPT will be performed by your PC,

    - YCbCr as prefered output, in master reference the conversion YCbCr -> IPT will be performed by your TV


    Using YCbCr means one less conversion step on the PC side, shifting that task to the TV instead, which helps small iGPUs. Your TV should handle the conversion perfectly, so you shouldn't notice any difference, however, if your PC is capable enough, you can opt for IPT, which ensures the conversion is performed with high precision right on your PC.


    In the re-release of RC2 PGS subtitles work correctly again, thank you.

    I have a request regarding them. This is already in a test build of avdvplus R11:

    Would it be possible to allow PGS subtitles to be moved up and down on-the-fly when manual positioning is enabled? This already works for text-based subtitles while watching a video by binding remote keys to the "subtitleshiftdown" and "subtitleshiftup" actions.

    I can check that

    Installed and working well here. Some feedback on the new features of RC2:

    1. Could the HDR10+ option be separate from the HDR10 options, as you can't currently config HDR10 to DV and HDR10+ to DV without changing it during playback. Also the menu entries in settings remain flagged as (unavailable) based on the previous playing file which is confusing.

    2. Is it possible to show if the CMv4 convertion is active in the PPI? Preferably updating live if one of the auto modes is in use.

    Thanks

    Yeah np I will add that for the RC3


    What is your GPU ?