View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000280 | madVR | bug | public | 2015-04-03 07:51 | 2015-04-04 09:00 |
| Reporter | altered | Assigned To | madshi | ||
| Priority | high | Severity | major | Reproducibility | always |
| Status | closed | Resolution | fixed | ||
| Summary | 0000280: Errant skin tone colour modification | ||||
| Description | With some videos madVR seems to alter skin tones, even with all colour-management options disabled. (I have disabled all external video or colour processing options in the GPU and in the various player' options) This only happens if madVR is the renderer, and only with some videos and not others. I've tested 5 players with various decoders and renderers. Every combination of player/decoder and renderer shows skin tones the same way *except* if the renderer is madVR. If madVR is the renderer, all players and decoder combinations show the errant skin tone colour shift. I tried every option in madVR to fix this but was unable to find one. Tried various decoders also - the only way to get this colour shift I could find was with madVR. I've also formatted and re-installed Windows, to no avail. The file attachment feature here was not working so please find screenshots and an example video here: https://mega.co.nz/#!RYpFwbSA!5gTWVHhSUWWO4VwnczLn5ki46DWNsqeo6sxSMKmtVD8 Please see the attached screenshot examples. These are all captured in windowed mode, but the issue remains in fullscreen non-exclusive and exclusive mode. This does not affect all videos - I've also attached an unconverted (made using AviDemux copy process) crop of a video which is affected. | ||||
| Steps To Reproduce | 1. Use any video player with switchable renderers to play the attached example video with any renderer OTHER than madVR (e.g., EVR, overlay, etc). 2. Observe the skin tone of the T-1000's face. 3. Switch the renderer to madVR with all colour management options disabled. 4. Observe that the skin tone colour on the T-1000 has changed. | ||||
| Tags | No tags attached. | ||||
| madVR Version | 0.87.14 | ||||
| Media Player (with version info) | Various tested | ||||
| Splitter (with version info) | Various tested | ||||
| Decoder (with version info) | Various tested | ||||
| Decoding | Software | ||||
| Deinterlacing | none (progressive) | ||||
| DXVA2 Scaling Active | no | ||||
| Aero / Desktop Composition | Off | ||||
| Problem occurs with mode | all modes | ||||
| GPU Manufacturer | NVidia | ||||
| GPU Model | GTX680 | ||||
| GPU Driver Version | 347.25 | ||||
|
|
|
|
|
Apologies for the multiple submissions. This bug submission app is apparently very buggy. |
|
|
Don't worry about the multiple submissions. Not your fault, happens all the time. |
|
|
What you are seeing is madVR using BT.709 matrix while other players are using BT.601. Since your sample is 1280x544, I'd consider madVR to be guessing correctly. XySubFilter for example guesses matrix by resolution for subtitles based on the video being either >1024 width OR >600 height. I believe madVR does something similar when video bitstream flags are not present. mpv also matches madVR on this sample, even though they use a slightly different check (>=1280 width OR >576 height). Haali renderer also gets it correct. Players and GPUs which get this wrong, are likely only taking video height into account. |
|
|
I'm checking "width > 1050" or "height > 576". So it's somewhere between XySubFilter and mpv. @altered, can you please double check whether what cyberbeing is correct? You can press "Ctrl+Alt+Shift+M" to toggle decoding matrixes. Set it to BT.601. If that matches what you expect to see, then that would proof cyberbeing correct, and that would mean that madVR renders the image correctly and the other media players you've tested are doing it incorrectly. |
|
|
It is indeed what cyberbeing said. I had actually tried checking for that myself using a switch in the MPC Video Decoder options but apparently that setting had no effect. Using the Ctrl-Alt-Shift-M keys did, however. Thank you for your help. Also thanks * 1 million, madshi, for this amazing piece of software. |
|
|
Great, then I close this bug report. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2015-04-03 07:51 | altered | New Issue | |
| 2015-04-03 07:51 | altered | File Added: disregard.zip | |
| 2015-04-03 07:54 | altered | Note Added: 0000915 | |
| 2015-04-03 09:05 | madshi | Note Added: 0000916 | |
| 2015-04-03 14:38 | cyberbeing | Note Added: 0000918 | |
| 2015-04-03 14:41 | cyberbeing | Note Edited: 0000918 | |
| 2015-04-03 14:53 | cyberbeing | Note Edited: 0000918 | |
| 2015-04-03 18:34 | madshi | Note Added: 0000919 | |
| 2015-04-03 18:34 | madshi | Assigned To | => madshi |
| 2015-04-03 18:34 | madshi | Status | new => feedback |
| 2015-04-04 01:42 | altered | Note Added: 0000920 | |
| 2015-04-04 01:42 | altered | Status | feedback => assigned |
| 2015-04-04 09:00 | madshi | Note Added: 0000925 | |
| 2015-04-04 09:00 | madshi | Status | assigned => closed |
| 2015-04-04 09:00 | madshi | Resolution | open => fixed |
