Dolby Vision w/FEL + 3D for PC (AMD/Intel/nVidia)

    Firstly thanks to everyone for the hard work you have put in, I never thought I would see the day that we would be here, and that my N150 would replace my shield!

    Is it possible to to clone the hdmi output so I can connect one directly to the TV and the other to my AVR? I have the handshake issue with my Denon 3800 and it is driving me crazy!

    Quote from LNorden

    Will it has LLDV as good as the UGOOS?

    On LLDV: Hades' build has it (setting off by default, and it applies only when the display offers DV solely as Low-Latency). I tested RC4 on an RX 7600 against a Homatics R 4K Plus, through a VRROOM with an LLDV EDID, and both looked natural. The tone mapping is Hades' own rather than Dolby's, so it isn't identical: moderate colours matched a reference composition, saturated ones came out brighter. CroqueMr's build had no LLDV when I last tested it (RC2).

    I haven't tried an H 255 (780M) or Plex, so I can't say anything about those or about bitrates.

    Quote from Samui67

    I was wondering whether extending the kernel patch to support DP-to-HDMI connections might be technically feasible.

    Thanks for the offer, but I won't be writing this one. From reading the driver, only a passive DP++ dongle could work with a small generic change, because the GPU then sends a real HDMI signal. Active converters are blocked by the driver itself (only frame-alternate 3D is allowed, "bad infoframe translation"), and the 3D infoframe isn't sent on DP at all. A 4K DP-to-HDMI cable is often active, and both adapters I own are active, so I can't test the passive case. If CroqueMr wants to try it, the place to start is the HDMI-only stereo_allowed check.

    Hi CroqueMr,

    Just to clarify: My DisplayPort-to-HDMI cable is passive, not active!

    It is connected directly from the DisplayPort output of my HTPC to the HDMI input of my 3D display, without any additional adapters.

    This means my setup uses DP++ (Dual-Mode DisplayPort), where the HDMI signal is generated directly by the graphics hardware.

    Since you mentioned that you could only test active adapters, my setup might be particularly interesting for testing the passive case.

    Could the HDMI-only stereo_allowed check be preventing 3D frame-packing output through this passive DP++ connection?

    Would it be possible to modify this check to allow 3D frame-packing on passive DP++ connections as well?

    I'd be happy to test a modified build if that helps!

    Thanks again for your work!

    Quote from cinemaONE

    I checked your log: Kodi is on connector DP-3 and its mode list has no 3D modes (no 1920x2205 frame-packing line). On AMD the kernel patch behind 3D on these builds offers those modes only on real HDMI connectors, because DP-to-HDMI dongles rejected the 3D timings in link validation.

    The "3D sbs (32)" is the same plain 1080p mode (32) as your 2D playback, with Kodi drawing both eyes side by side. No 3D signal reaches the projector, which is why you see two images. So the limit is the DP output, not Vega 11, and SBS or TAB files need the same 3D modes. We have only tested 3D on native HDMI.

    Thanks for the precision I thought DP converted was ok for 3D packing but unfortunatly no. I will had this limitation on the repo.

    I thought 3D was not HDMI depend but I was wrong.
    For 3D, I've taken up the existing work of CinemaOne and PopcornMix. DisplayPort to HDMI conversion might be possible via a kernel update for the specific DisplayPort model, but unfortunately, supporting all existing DisplayPort to HDMI converters isn't a priority for me, knowing that even with a kernel update, some won't be compatible with Dolby Vision or other formats.

    If I do support DisplayPort to HDMI conversion for Dolby Vision and 3D more globaly, it will be after the stable release and by specifying a precise adapter for which I'm certain the LSPCON doesn't change from one version to the next.

    Hi CroqueMr,

    Thanks for your detailed explanation!

    There is one technical distinction I'd like to clarify regarding my setup.

    I'm using a passive DisplayPort-to-HDMI cable, not an active converter with an LSPCON chip.

    Assuming the connection operates in DP++ (Dual-Mode DisplayPort), the graphics hardware generates an HDMI-compatible TMDS signal directly, without an external DisplayPort-to-HDMI conversion chip.

    Wouldn't this make my setup different from the active DP-to-HDMI adapters that failed the 3D timing validation?

    I understand that the AMD kernel patch currently enables HDMI 1.4 3D frame-packing modes only on native HDMI connectors.

    However, could the HDMI-only connector check potentially be relaxed specifically for passive DP++ connections, allowing the driver to attempt the required 3D timings?

    Of course, successful link validation would still need to be verified.

    I'm happy to test an experimental kernel patch or LibreELEC build on my hardware if you're interested.

    Thanks again for your excellent work!

    I don't know, I need to check, unfortunatly my priority is to move forward regarding the stable release, not to extend support to DP. Maybe later once stable release finalized

    Hi CroqueMr would you take a PR adding desktop OpenGL support, so Dolby Vision works on a GBM OpenGL build as well as GLES? I’d like to run my AMD box on the OpenGL build without losing DV.

    What I have in mind:

    • One code path rather than a GL copy. CB1’s GLSL ES 3.x shaders compile unchanged in a desktop GL context on Mesa (ARB_ES3_2_compatibility on radeonsi and iris), and the GL calls are the same apart from a few OES/EXT names.
    • Lift the GLES-only checks in CMake and package.mk, build DVBridgeGLES for both, and hook it into LinuxRendererGL, RendererVAAPIGL and WinSystemGbmGLContext the same way as their GLES versions, sharing the output code between the two GBM contexts.
    • Test on an AMD 780M across TV-Led, Master Reference, HDR10 conversion and FEL.

    Would you take that, and would you rather have it as one PR or split (build and shared code first, then the GL renderer and context)?

    fyi for those doing development or wanting to increase the amount of testing and verification of the final outputted pixels, some devices (at least an i3-1115G4, presumably lots of other devices), are capable of performing a lossless capture on device themselves.

    This capture occurs after all shader/kodi/linux kernel/hardware planes/hardware video processing (upscaling, chroma resampling, hardware plane blending, etc) occurs. It functions by rerouting the final pixel data to file instead of the hdmi module. Example code to perform this is https://github.com/doppingkoala/l…6fd78224e3c71c2, I have checked that the 8-bit RGB capture mode is indeed 100% bitperfect of the data sent of the hdmi signal. Note the linked code is only setup for a single frame, the docs suggest captures of a number of sequential frames should be possible to capture as well.

    Might be useful to someone. Being on device both means no other hardware required and that it is also able to be ran via ssh which opens the possibility of automated testing to test as many frames / test patterns as one could possible want.

    Hello, i'am new to all of this. I installed this build and so far i can say it works fine and i really like it but i have power savings problem or maybe its a hdmi handshake thing. I disabled all power saving options and the screensaver but somehow when i turn my tv off over night or for a longer time, i cannot get a picture from my pc again without rebooting it.

    Also when i try to reboot directly in kodi i get no picture after the reboot. Also the TV has to be on when i start the pc or i wont get a picture out of it. GPU is a RTX 3050, CPU is a 5500K I5 i think and its a Gigabyte Mainboard.

    Can someone maybe point me in the right direction to fix this?

    Quick question, how do i update once a new version comes out ?


    Thank you

    Quote from hadesdevls

    lfemperor Thanks for testing RC7 so thoroughly, and good to hear the FEL test plays in all four formats, seeking included. Transformers 2 dropping to HDR10 is a fault in Hades. ht2tweak saw the same with the Transformers and Tron samples on an N150, so both reports are on Intel graphics. I'll reproduce it with those samples and fix it for the next version.

    Yes please, the kodi.log of a playback that drops would help, ideally with the diagnostic on (rename DIAG-OFF.TXT to DIAG-ON.TXT on the boot partition, play the film until it drops, then zip the hades-diag folder from that partition). When Hades leaves native Dolby Vision, it writes the reason in kodi.log and on the Dolby Vision page of the player information. I'll start with the samples from ht2tweak and ask for a clip only if they don't show it. Thanks also for pointing to RC2 of CroqueMr.

    Hi hadesdevls,

    That makes sense, if ht2tweak is also experiencing this on an N150, it's definitely an Intel i915 pipeline issue.

    I have already enabled the diagnostic mode by renaming DIAG-ON.TXT, played Transformers 2 until it dropped to HDR10, and zipped the hades-diag folder. Please find the attached diagnostic log.

    I also noticed the reason is written in the Player Information Dolby Vision page, which is a great debugging feature! Hopefully, this log helps you nail down the issue quickly. Since you have samples from ht2tweak, I'll hold off on sending the clip for now. Let me know if you still need it!

    Thanks!

    Hi CroqueMr and cinemaONE,

    I have some very interesting news!

    I tested my smaller HP EliteDesk with an optional HP HDMI output module and an AMD Ryzen 5 PRO 2400GE (Vega graphics), using the same LibreELEC build.

    HDMI 3D frame packing works!

    My LG 60-inch 3D plasma TV automatically detects the 3D signal.

    The attached sanitized log confirms that the AMD driver detects the output as HDMI-A-1 and exposes the following modes:

    1920x2205 @ 24.000000 Hz 3D fp

    1920x2205 @ 23.976080 Hz 3D fp

    During playback of an original 3D Blu-ray directly from the optical drive, Kodi selects:

    Display resolution ADJUST : 1920x2205 @ 23.976080 Hz 3D fp (40)

    There are also some switches between frame packing and SBS in the log.

    This is particularly interesting because my other EliteDesk with the Ryzen 2400G and DisplayPort-to-HDMI cable only exposes DP-3, without the required frame-packing modes.

    The HDMI module's internal implementation is still unclear, but the practical result is confirmed by the LG plasma's automatic 3D detection.

    I've attached a sanitized 3D diagnostic excerpt without unrelated Add-on information.

    Could this working configuration help identify why frame packing is unavailable on the DisplayPort-based system?

    I'd still be very happy to test an experimental build if useful.

    Thank you both for your help and your excellent work!

    Quote from Samui67

    Could this working configuration help identify why frame packing is unavailable on the DisplayPort-based system?

    Nice result, and it answers your question: the driver sees that module as HDMI-A-1, so the 3D modes are offered. On the other PC it sees DP-3, so they aren't. I was wrong to assume the module wouldn't help. HP lists the HDMI Flex IO card (3TK75AA) for the 705 G4 SFF too, so try moving your module across, if it's that part. If the log then shows HDMI-A, you have 3D. I can't promise it will, as I haven't seen this on the SFF.

    If it doesn't work out, I'm happy to reconsider what I said earlier and make you a test build for the passive cable. It would be untested on my side, since I only have active adapters, so you'd be the test. Let me know how the module goes.

    Hi cinemaONE,

    Thanks again for your support and your willingness to help!

    I've ordered an additional HP HDMI Flex IO module (3TK75AA), as I'd like to equip all of my HP EliteDesk systems with the same HDMI hardware.

    Please bear with me until the module arrives. Once installed, I'll test 3D frame packing on the 705 G4 SFF and report back with the results.

    This should give us a useful comparison between the systems.

    Thanks again for your patience and help!

    Quote from cinemaONE

    Nice result, and it answers your question: the driver sees that module as HDMI-A-1, so the 3D modes are offered. On the other PC it sees DP-3, so they aren't. I was wrong to assume the module wouldn't help. HP lists the HDMI Flex IO card (3TK75AA) for the 705 G4 SFF too, so try moving your module across, if it's that part. If the log then shows HDMI-A, you have 3D. I can't promise it will, as I haven't seen this on the SFF.

    If it doesn't work out, I'm happy to reconsider what I said earlier and make you a test build for the passive cable. It would be untested on my side, since I only have active adapters, so you'd be the test. Let me know how the module goes.

    he needs to make sure he has the right module for his sff, I did some research before
    https://forum.kodi.tv/showthread.php?tid=385600&pid=3315610#pid3315610

    HP only sell v2, which I don't think it works on his machine?
    This is version 1
    https://www.ebay.com/itm/395447887372?_skw=HDMI+port+small+board+card+port+906318-002+for+HP+400+600+800+G3+G4+G5+DM+SFF&itmmeta=01M5XBAYERZX89Y9SM3QX2VQJV&hash=item5c1287e60c:g:EWIAAOSwLe5mXuRP&itmprp=enc%3AAQALAAABMCHNgUiiE%2FjQhlDTaMiAde%2BhxNEpexp9dNbu2eZcweJmmx%2BWFKbzZ%2BhxfDtSX2%2F3i6c%2Bhicm1046xhblc3qGlElNdKbIANhIDl0SxXpQMlo%2FO2NGtLKWViDJvP9Q9rB%2FbM1H3k2HiiFlbHlsN3ud1jheqe2PUJniHuU104zfA%2FnhStlj9vl1wDVzf2UVFLNoMHWMPYNpD9se0G0NjLJXl5my7t2Bqe1S90hYhGp4F4GwV5lCr47%2B34g4s444G5%2FuBYqYvafEnhIG2rux2y7jfOSULrWwJINggPGd9q1x1bGrVxGtOlvIXnB1W9NtDjJGSCV%2BoW8Onx9iI%2B8xreb02n%2BOpANLBlGzEuW%2FMlU9R5wrCZ43hRBt88%2Bqu6mgNa1t5J%2Bpy0upbiY5GpxfhYmnty0%3D%7Ctkp%3ABk9SR8jnq6uvaA

    Samui67 check

    Hi cinemaONE and everyone,

    I've found the official HP Maintenance and Service Guide for my HP EliteDesk 705 G4 Small Form Factor Business PC (HP document L23140-001, June 2018).

    On page 5, under Illustrated parts catalog → Option boards, HP explicitly lists:

    Quote

    HDMI option board
    DisplayPort option board
    VGA option board
    USB Type-C option board

    The manual also shows the optional rear port on page 3 and documents the option port board removal procedure on page 46.

    So HP officially confirms that an HDMI option board was available for this exact SFF model.

    However, the manual doesn't specify the exact HDMI module part number or distinguish between version 1 and version 2.

    I've already ordered an additional HDMI module and will check the part numbers and compatibility before testing it in the SFF.

    I'll report back once I've tested it.

    Thanks again for your help and research!HP_EliteDesk_705_G4_SFF_HDMI_Auszug.pdf

    Quote from ashlar

    Try the current RC 4 preview build and, under system, force 8 bit color depth output. I am currently using a i3-1115G4 (Tiger Lake as well) and I discovered that, otherwise, even SDR gets outputted as 10bit. And that messes up everything on our machines.

    If the above setting solves things for you, you have my same exact situation.

    I tried this on the new RC4 and set the maximum bit depth to 8. That fixed the issue for 4k SDR and 4K HDR movies. But as expected all the 10 bit and now being downsampled to 8 bit. Also still have massive frame drops and stuttering when a movie is playing and you trigger the OSD using the O button on the keyboard. Again this only happens with Tigger Lake. I ran another using vm on proxmox with AMD Ryzen 5 5600G and it had no such issues.