Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
28th August 2017, 07:30 | #45161 | Link |
Registered User
Join Date: May 2015
Posts: 1
|
I'm facing some weird bug where custom modes for my TV is showing the resolutions for my primary monitor instead of the TV. In the previous version, it would show the correct resolutions, but I had to set my receiver as a TV instead of the receiver option because the custom modes tab would disappear instantly.
I also didn't see this in the Custom Mode Tutorial, but I feel this should be noted there. Apparently you can't have DSR enabled on NVIDIA cards if you want custom resolutions to work. I was really confused when my timings weren't being saved at all, until I checked the NVIDIA Control Panel for custom resolutions and it said DSR had to be disabled for me to use custom resolutions. Last edited by Xterminator; 28th August 2017 at 08:23. |
28th August 2017, 08:36 | #45162 | Link |
Registered User
Join Date: Mar 2002
Posts: 35
|
Hi madshi, thx for you response, here are the Infos.
Sony ZD9 Denon X6300H Windows 10 En 2016 LTSB (1607) Sap Nitro 4GB Rx460@560Bios AMD Cat 17.8.2 Lavfilter Nightly 70.2.66 MadVR 92.2 testet with MPC-hc and Dvbviewer all 32bit https://picload.org/view/rwipgpal/lav.jpg.html https://picload.org/view/rwipgpai/madvr.jpg.html https://picload.org/view/rwipgpaw/osd.jpg.html I testet also a lot of settings RGB 4:4:4 full and reduced, YCbCr 4:4:4,4:2:2 in driver, 10 and 12 bit in driver(madvr set to 10bit and higher), in LAV all 3 settings I try´d windowed and exclusive.... HDR from shield with amazon or netflix is working. If you need more info, just let me know. Last edited by Jasch; 28th August 2017 at 08:39. |
28th August 2017, 10:35 | #45163 | Link | ||||||||||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
Quote:
Quote:
Is this with the new native D3D11 decoder, only? Or does this also happen with other decoders? Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
What does the EDID report say in the madVR settings dialog? Does it report HDR to be supported for your display? |
||||||||||
28th August 2017, 10:38 | #45164 | Link |
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
P.S: @FDisk80 and @Asmodian, can you please update to the very latest LAV nightly build? Older LAV nightlies had a bug which could result in D3D11 copyback to become active instead of D3D11 native. So the latest LAV nightly might already fix the problem you've reported.
|
28th August 2017, 12:29 | #45165 | Link |
Registered User
Join Date: Mar 2007
Posts: 934
|
When using DX11 decoding in LAV Filters, rendering time dropped from ~30 ms to ~10 ms for 1080p videos on my 1080p laptop when I switched from DX9 to DX11 render path. I'm sure I had issues in the past with the DX11 render path though, probably related to interlaced files, but for now I'll leave it like that and see how it goes. If deinterlacing gets implemented and works correctly I think I'll be sticking with it going forward!
__________________
TV Setup: LG OLED55B7V; Onkyo TX-NR515; ODroid N2+; CoreElec 9.2.7 Last edited by DragonQ; 28th August 2017 at 12:32. |
28th August 2017, 12:35 | #45166 | Link | ||
Registered User
Join Date: Jul 2014
Posts: 942
|
Quote:
I reverted to my system image from the 23rd (when I did the HDR Metadata tests on the 3 files with MadVR 0.92.1). Behaviour is the same, so not caused by a change in the code. I was focusing on the HDR data when I tested the file and as it is displayed I might have missed the issue then (it's a brightness pattern so mostly dark). You can download the demo patterns from RM here: http://www.avsforum.com/forum/139-di...-patterns.html They all seem to cause the same issue. I calibrate HDR with the UB900 (my primary source for HDR content) so I haven't noticed this before on the HTPC. Beware, I know that you won't believe that it could be the cause, but testing the pattern yesterday and letting it run a few times longer than I had when I was simply reading the metadata (I was trying to test the various decoders as requested) completely warped my OS partition, possibly due to the RAM overflow condition, hence why I had to reinstall from system image. I therefore highly suggest someone testing this has an up-to-date backup before running this patterns for more than a little while. I am aware it sounds impossible, I am just reporting what happened here. Better safe than sorry Quote:
I also believe that the way you describe Passthrough mode, "the display receives the original HDR content untouched" is misleading if you are manipulating HDR metadata for whatever reason you see fit. In addition, changing the metadata prevents finding issues in the way you handle it, as we have found in the last few days. Again, in my opinion, pass-through should be pass-through if called that way and described that way. I would therefore suggest one of the following options: 1) If you don't care about the possible issues I'm reporting, you keep things the same but stop calling Passthrough mode that way and stop describing it as passing the HDR content untouched, because this is not what MadVR is doing if it changes the metadata for arbitrary reasons. It's an optimized mode that might be better for some displays (those that actually act on the static HDR10 metadata), but not for all. At the very least there should be an indicator in the OSD to warn that the metadata was molested and doesn't reflect the data present in the original content. 2) You offer a "native metadata" mode that does what passthrough should be doing (passthrough the HDR content untouched, including the metadata). Of course that doesn't include things we might want MadVR to do and can select manually, for example, I'm very happy to ask MadVR to do the chroma upscaling in that mode (it's doing a great job with NGU anti-alias). But if I ask MadVR to do nothing, it should do nothing in that mode and certainly not change the original metadata, especially when not displaying it in the OSD. 3) Display all the native metadata in the OSD so that we know what was in the source before you send the "corrected" data to the display. It's your opinion which parameters matter and which do not. When calibrating, the reported mastering black level can matter as much as what the max brightness is, depending on the curve/settings we want to use and the options we want to have. HDR10 is a big mess, it doesn't work in "auto" mode with most displays, and behaving as if it did is not really helpful. Again, I'm not talking about my own needs. I can get all this unmolested data from the UB900. I currently play all my HDR content from my UB900. I simply hope that at some point, I'll be able to use MadVR to play my UHD Blurays and it would be a shame to have to keep the UB900 just to be able to know what the actual metadata is so that we can select the appropriate calibration, should we wish to optimize it depending on the content. Finally, this is purely theoretical because MadVR in the current build doesn't seem to be replacing the metadata even when it's set to zero. For example, the Exodus trailer reports 0,0 for MaxCLL/FALL, and MadVR correctly reports this. In any case, it's your and Nevcairiel's decision what you want to do with metadata you deem unfit to use. I think it would be good to have an option to get the native metadata, unmolested, if we wish to. If you decide against it, it's not the end of the word, we'll still enjoy the product of your amazing collaboration I wish we were in a world where all displays would act on metadata and not simply switch to a fixed mode whenever they detect it, in which case your decision would make complete sense, but in the real world as it is today, where many if not most displays simply discard the metadata, MadVR/LAV would behave in a worse way than a standalone player (from a calibration optimization point of view), at least for most current HDR projectors, and that's not something I'm used to with this great combo
__________________
Win11 Pro x64 b23H2 Ryzen 5950X@4.5Ghz 32Gb@3600 Zotac 3090 24Gb 551.33 madVR/LAV/jRiver/MyMovies/CMC Denon X8500HA>HD Fury VRRoom>TCL 55C805K Last edited by Manni; 28th August 2017 at 15:31. |
||
28th August 2017, 12:40 | #45167 | Link |
Registered User
Join Date: Jul 2014
Posts: 942
|
Forgot to say that the latest nVidia drivers (385.41) prevent from selecting 12bits in a custom mode in the nVidia control panel. Only 8bits is available. I reverted to 385.28 for this reason.
__________________
Win11 Pro x64 b23H2 Ryzen 5950X@4.5Ghz 32Gb@3600 Zotac 3090 24Gb 551.33 madVR/LAV/jRiver/MyMovies/CMC Denon X8500HA>HD Fury VRRoom>TCL 55C805K |
28th August 2017, 14:23 | #45168 | Link |
Registered User
Join Date: Nov 2016
Posts: 4
|
https://www.youtube.com/watch?v=hILLamChO8I
OK unchecked private video now. I hope you will look into it. Thanks |
28th August 2017, 14:43 | #45169 | Link |
Registered User
Join Date: Dec 2011
Posts: 1,812
|
Is bilinear downscaling working correctly currently? It gives me quite higher GPU usage and rendertimes than DXVA2 downscaling.
I have the same rendertimes, GPU clock and GPU usage with bilinear for chroma and luma like with bicubic when downscaling 4k 60fps to ~1440p. I've already resetted the madVR settings to check if it makes a difference (doesn't). I really don't find it. |
28th August 2017, 15:31 | #45170 | Link | |
Registered User
Join Date: Jun 2017
Posts: 71
|
Quote:
|
|
28th August 2017, 16:42 | #45172 | Link | |
Registered User
Join Date: Mar 2002
Posts: 35
|
Quote:
HDR is supported. https://picload.org/view/rwipwcwa/edid.jpg.html |
|
28th August 2017, 17:51 | #45174 | Link | |||||||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Oh, I think I know why. I wanted to answer your question of whether I can give an ETA with "No.", but doom9 refused to post a 3 character reply, so it got lost. Quote:
|
|||||||
28th August 2017, 18:28 | #45175 | Link | |||
Registered User
Join Date: Jul 2014
Posts: 942
|
Quote:
Quote:
Quote:
Point is, you can do whatever you want to provide the best user experience in any of the HDR modes controlled by MadVR, but IMHO you should also provide a proper passthrough mode (possibly different from the current passthrough mode, or offering an option within this mode) where the data is unmolested. It is not up to MadVR to decide which calibration I want to use. For example, when playing bluray, where there is no standard for gamma, you're not forcing which gamma curve you prefer. We are free to use BT1886 or power gamma 2.2 or 2.4 depending on our taste, setup, and what we believe is most correct for each title. There is no metadata indicated which gamma curve was used for bluray, but if there was one and displays were not able to act on it automatically I wouldn't want MadVR to decide what to do if it was incorrect. I'd want to know it is incorrect so I can make the best decision. At the moment, we are in a similar situation with HDR10. There is no standard for the consumer playback side, it's not implemented the same way by each manufacturer, so it's important to have all the info in order to make the right decision. I am not talking about tweaking anything in MadVR. I'm talking about having a mode that passes through the metadata unmolested, a manual mode if you wish. Think of it as a "source direct" mode in an AVR, where we have the option to ask the AVR to not touch the content so that an external VP or the display can upscale/process it instead. If you're dumbing down MadVR so that it can be used by people who don't know what they are doing, great, I'm sure they'll appreciate the simplicity, but that doesn't take away that some of us will prefer to get the data unmolested and make the decision ourselves about which calibration to apply. Again I think this is because of the discrepancy between your claim for HDR Passthrough, "the HDR content is untouched", and the actual process which can result in some data being changed without us knowing which part of the data or in which way it was changed. I think your aim to automatize/optimize things and try to make the metadata as correct as possible for displays that interpret it is great. It doesn't take away that in other situations, especially for displays that don't give a monkey about metadata such as most HDR projectors on the market now, knowing what the metadata is exactly is far more useful than getting default metadata without knowing which part is original and which part is processed or fixed. I agree that projectors are a special case due to their limited max brightness, but that only makes it more important to get all the metadata so we can make the best decision according to the content. In which way? Does it mean that you will always report them as they are even if they are set to zero, or does it mean that for now we get the original data but in the future you will replace the zero values with default ones?
__________________
Win11 Pro x64 b23H2 Ryzen 5950X@4.5Ghz 32Gb@3600 Zotac 3090 24Gb 551.33 madVR/LAV/jRiver/MyMovies/CMC Denon X8500HA>HD Fury VRRoom>TCL 55C805K Last edited by Manni; 28th August 2017 at 18:42. |
|||
28th August 2017, 18:43 | #45176 | Link | |
Registered User
Join Date: Dec 2011
Posts: 1,812
|
Quote:
I tested again a few times, and there is a difference, but a very small one. DXVA2 scaling is definitely way faster than bilinear. Is that supposed to be? DXVA2 scaling: 2.1ms bilinear: 5ms bicubic: 5.5ms (GPU set to maximum performance power profile, which results in ~1.6GHz for all three scalers) Meh. Well, still better than no answer at least. Last edited by aufkrawall; 28th August 2017 at 19:00. |
|
28th August 2017, 18:50 | #45178 | Link |
Registered User
Join Date: Jan 2002
Posts: 1,264
|
OK just in case you (Madshi) can get HDR working properly on my LG OLED perhaps this might help?
Code:
00 FF FF FF FF FF FF 00 1E 6D 01 00 01 01 01 01 01 1A 01 03 80 A0 5A 78 0A EE 91 A3 54 4C 99 26 0F 50 54 A1 08 00 31 40 45 40 61 40 71 40 81 80 01 01 01 01 01 01 08 E8 00 30 F2 70 5A 80 B0 58 8A 00 40 84 63 00 00 1E 02 3A 80 18 71 38 2D 40 58 2C 45 00 40 84 63 00 00 1E 00 00 00 FD 00 3A 3E 1E 88 3C 00 0A 20 20 20 20 20 20 00 00 00 FC 00 4C 47 20 54 56 0A 20 20 20 20 20 20 20 01 9F 02 03 5D F1 58 61 60 10 1F 04 13 05 14 03 02 12 20 21 22 15 01 5D 5E 5F 65 66 62 63 64 29 3D 06 C0 15 07 50 09 57 07 6E 03 0C 00 30 00 B8 3C 20 00 80 01 02 03 04 67 D8 5D C4 01 78 80 03 E2 00 CF E3 05 C0 00 E3 06 05 01 E4 0F 03 00 18 EE 01 46 D0 00 24 18 09 00 AD 52 44 A9 23 0C 66 21 50 B0 51 00 1B 30 40 70 36 00 40 84 63 00 00 1E 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 E8 |
28th August 2017, 19:09 | #45179 | Link | |
Registered User
Join Date: Feb 2002
Location: San Jose, California
Posts: 4,406
|
Quote:
Instead of picking display calibrations based on the metadata in dodgy HDR files just use your eyes, it sounds like you have a very good grasp of how you want HDR to look and your eyes are MUCH more reliable than known bad metadata. Don't get me wrong, I would love a super verbose mode for the OSD with multiple columns of everything I can think of (original HDR metadata included). A little graph of the rendering times for the last 500 or so frames would also be great, it could even have a horizontal line at the video's frame time. There are a lot of interesting things to know but they aren't critical for display decisions.
__________________
madVR options explained Last edited by Asmodian; 28th August 2017 at 19:14. |
|
Tags |
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling |
Thread Tools | Search this Thread |
Display Modes | |
|
|