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.

 

Go Back   Doom9's Forum > Hardware & Software > Software players

Reply
 
Thread Tools Search this Thread Display Modes
Old 20th October 2011, 14:19   #10241  |  Link
JaylumX
Something Else
 
Join Date: Feb 2006
Location: Somewhere Else
Posts: 76
Quote:
Originally Posted by RyaNJ View Post
I can confirm this.
Same here. Also some of my videos do not have the seek bar
JaylumX is offline   Reply With Quote
Old 20th October 2011, 14:33   #10242  |  Link
Fadeout
Registered User
 
Join Date: Dec 2009
Posts: 150
Yes, two problems: crashes on exist and no seek bar. The second in particular needs fixing.

I also have another problem carrying over from previous versions: when I exit exclusive mode the player window misses the close button and I have to close through the menu.
Fadeout is offline   Reply With Quote
Old 20th October 2011, 14:57   #10243  |  Link
naoan
Registered User
 
Join Date: Oct 2009
Posts: 151
Also crash if you drag another video while mpc-hc still playing.
naoan is offline   Reply With Quote
Old 20th October 2011, 15:07   #10244  |  Link
nand chan
( ≖‿≖)
 
Join Date: Jul 2011
Location: BW, Germany
Posts: 380
Quote:
Originally Posted by madshi View Post
madVR v0.75 released

Code:
* subtitles run through the 3dlut now, too
* added option to scale Luma in linear light, disabled by default
* RGB input with unknown range is now treated by default as full range
* added detailed information about matrix, primaries and levels to debug OSD
Awesome, now all we need is a 64-bit build and 10-bit output and we are good to go again!
__________________
Forget about my old .3dlut stuff, just use mpv if you want accurate color management
nand chan is offline   Reply With Quote
Old 20th October 2011, 15:29   #10245  |  Link
TheShadowRunner
Registered User
 
TheShadowRunner's Avatar
 
Join Date: Feb 2004
Posts: 399
Thanks for the new build!
Report:
On ZP, 0.75 doesn't crash on exit but madVR's own OSD has become gigantic in fullscreen mode and minuscule in windowed mode. (when it indicates windowed/exclusive)
Also ZP's "Display OSD through madVR's API" is now broken = ZP's control bar / OSD don't show up at all anymore in fullscreen when displayed via madVR's API.

As for the (ZP) bug which makes ZP slow/freeze on seek when "Display OSD through madVR's API" is disabled and a frequency switch occurs, it is still there. The madVR crash when "show OSD filename when opening a new file" is disabled is still there too, but the address has changed.
With 0.74, crash in madVR.ax module at address 0x00004770
With 0.75, crash in madVR.ax module at address 0x000047b0

Hope it helps, somehow ^^;;
__________________
XP SP3 / Geforce 8500 / Zoom Player
TheShadowRunner is offline   Reply With Quote
Old 20th October 2011, 15:29   #10246  |  Link
Boltron
Registered User
 
Boltron's Avatar
 
Join Date: May 2011
Posts: 94
Crash for me also when I exit. On Win7 x64, MPC-HC>LAV Filters>madVR.
Boltron is offline   Reply With Quote
Old 20th October 2011, 15:32   #10247  |  Link
THX-UltraII
Registered User
 
Join Date: Aug 2008
Location: the Netherlands
Posts: 851
Quote:
Originally Posted by irishrobbie View Post
i can tell ya that madvr with lav decoder is deffo a good combination and im using an ati 5670 graphics card, very smooth indeed no frames been dropped
I don t think it will make any difference when using a 4550, 5670 or even a higher end ATI card because the GPU of the ATI is not beeing used because madVR and LAV video decoder both do not support ATI hardware decoding. That s why I got the 5450.

atm I m thinking about switching to a NVIDIA card. This way I can use CUVID in LAV Video Decoder and reduce my CPU usage (because I want to use SVP with high settings http://www.svp-team.com/wiki/Main_Page)
THX-UltraII is offline   Reply With Quote
Old 20th October 2011, 15:53   #10248  |  Link
6233638
Registered User
 
Join Date: Apr 2009
Posts: 1,019
Quote:
Originally Posted by madshi View Post
To my best knowledge the VideoEQ Pro is not able to dither. So you're working with *rounded* 8bit or 10bit values. If you work with rounded (or truncated) data, of course every bit more makes a dramatic difference. 10bit really is *much* better than 8bit, when no dithering is used.
Correct, and I agree, the dithering improves the results considerably... but I would still like to be able to tweak the pre-dither (or no-dither) results to perfection, and look forward to when I can get 10-bit sent to the display.

This is one of the reasons I got rid of the VideoEQ though—that and the completely broken CMS.

I've also had a Lumagen Radiance XE (overpriced junk) and iScan Duo, neither of which has stayed in my system too long either. I think the iScan lasted the longest, as it had some unique features that were quite useful back when I was still using stand-alone devices for media playback.

Quote:
Originally Posted by madshi View Post
The problem is that you don't have an 256x256x256 3dlut with the VideoEQ box, AFAIK. How many control points do you have?
You had 10-bit control for greyscale/gamma (three 1024-point 1D LUTs) and matrix controls for the CMS from what I remember.

I'm not trying to compare it to what madVR is doing though, just explain why it was much easier to get the results I wanted with it, and why I look forward to 10-bit output.

Quote:
Originally Posted by madshi View Post
Hand tuning a 3dlut can only work if you have a limited number of data points. How many data points do you have with the VideoEQ box? With madVR, if you wanted to hand tune the 3dlut, you'd have to play with 256x256x256 data points!
I'd probably work with 10, 20 or 100 points, time permitting, or possibly 10 points plus touch-ups where necessary. (fix any obvious errors) I realise it sounds like this goes against "hand-tuning" the LUT if you are letting the rest of the points be interpolated, but I'd prefer some sort of control rather than tell madVR what my display currently measures and hope it all turns out OK in the end.

It's not just greyscale though, I'd love to be able to hand-tune 25/50/75/100% saturation for RGBCMY to make sure that the response is linear—I'm not sure that my display is when put into the "wide gamut" mode. (necessary to cover BT.709 gamut as it's a white LED display) Haven't had a chance to actually measure it and check for linearity yet.

Quote:
Originally Posted by madshi View Post
You could play with the measurement data you feed into yCMS, though.
Right, but then I feel like I'm basically guessing—yCMS only accepts XYZ or xyY data, right?

I suppose the simplest way to put it, is I just want to give it:
100%: 254,234,245
90%: 230,225,240
etc. and have it fill in the rest. (ideally this would be 10-bit though...)


Quote:
Originally Posted by madshi View Post
Did any of those boxes meet madVR's specs, which are:

- 256x256x256 data points
- 16bit output
- dither
No, this is the reason they didn't stay, and why madVR has replaced all stand-alone devices connected to my TV.

Quote:
Originally Posted by madshi View Post
You said it yourself: With an "empty" 3dlut, you see no banding and no discoloration. But still, with an empty 3dlut, madVR runs the full processing with the final dithering down to 8bit output. If either the 3dlut or the downdithering to 8bit is a problem in itself, why do you not see any problems with an empty 3dlut? That makes no sense. So the problem is neither the 3dlut processing in itself, nor the dithering down to 8bit. The problem you're seeing with madVR is probably caused by either bad measurements and by yCMS doing something wrong. At least that's my current understanding.
The main part of the problem isn't even madVR itself. (though I'm still not entirely convinced about 8>16>8 avoiding possible colour shifts near black when making gamma changes)

The problem is that you cannot disable a modern display's own internal LUTs/Processing, and have it display exactly what you send it. To get the best results, you're basically fighting against the display.

I feel that if you had used one of these external LUT boxes, you would immediately understand the problem, and I'm really not doing a good job of explaining it.

When tweaking with the VideoEQ, it was not uncommon to see things such as this:
Code:
Pass|Output Value|Relative R|  G   |  B
  1 | 64, 64, 64 |   105%   | 100% | 98%
  2 | 62, 64, 66 |   104%   | 100% | 99%
  3 | 60, 64, 67 |   103%   | 100% | 99.2%
  4 | 58, 64, 68 |   103%   | 100% | 99.4%
  5 | 56, 64, 70 |   101%   | 100% | 102%
  6 | 55, 64, 69 |   099%   | 100% | 101%
10-bit precision helped dial in the results better, and the reason you have things jumping several steps at once sometimes is because you're essentially fighting against the display's internal processing to get the result you want from it.

Sometimes changing one of the output values from the LUT box would not make a change on the display, and sometimes making a single digit change would have it jump 2-3%.

This is why I don't believe that you can get the best results from a calculated LUT compared to one where you are telling it at least a portion of what things need to be to help get the best out of your display and tweaking the results where problems occur the most. (you might do 20-100% in 10% steps, 5-20% in 5% steps and manually tweak 1-5% individually for example)

Quote:
Originally Posted by madshi View Post
added option to scale Luma in linear light
Is there a reason why this is only for Luma? I look forward to trying it out.

This will hopefully be seeing a lot of use from me shortly, as I'll soon be downsampling everything to 720p once Sony launches the HMZ-T1.

I can confirm the MPC-HC crashes other people are reporting.
6233638 is offline   Reply With Quote
Old 20th October 2011, 15:57   #10249  |  Link
Gaius
Registered User
 
Join Date: Aug 2011
Posts: 28
I don't have a crash on exit, but I do have exclusive mode turned off, and I don't use a separate device and I have all the "trade quality for performance options" checked.

Last edited by Gaius; 20th October 2011 at 16:49.
Gaius is offline   Reply With Quote
Old 20th October 2011, 16:30   #10250  |  Link
Boltron
Registered User
 
Boltron's Avatar
 
Join Date: May 2011
Posts: 94
Crash happens with and without exclusive mode for me.

Problem signature:
Problem Event Name: APPCRASH
Application Name: mpc-hc.exe
Application Version: 1.5.3.3757
Application Timestamp: 4e94a9c2
Fault Module Name: StackHash_6c74
Fault Module Version: 6.1.7601.17514
Fault Module Timestamp: 4ce7ba58
Exception Code: c0000374
Exception Offset: 000ce653
OS Version: 6.1.7601.2.1.0.256.1
Locale ID: 1033
Additional Information 1: 6c74
Additional Information 2: 6c74644f2a776d4eebd8a21e66ac538c
Additional Information 3: a883
Additional Information 4: a883ebc5ad391b943d97848a87e3b702
Boltron is offline   Reply With Quote
Old 20th October 2011, 17:23   #10251  |  Link
nand chan
( ≖‿≖)
 
Join Date: Jul 2011
Location: BW, Germany
Posts: 380
Quote:
Originally Posted by 6233638 View Post
Correct, and I agree, the dithering improves the results considerably... but I would still like to be able to tweak the pre-dither (or no-dither) results to perfection, and look forward to when I can get 10-bit sent to the display.
That's like saying you want to master audio in the next room. Simply silly/stupid, no offense intended. You *NEVER* want to tweak the no-dither results, since those results will be wrong. Tweaking those until they're right will end up with a LUT that's completely off. You're shooting yourself in the foot. And if you calibrate against these faulty procedures, of course madVR is going to give you undesirable results.

Quote:
This is one of the reasons I got rid of the VideoEQ though—that and the completely broken CMS.

I've also had a Lumagen Radiance XE (overpriced junk) and iScan Duo, neither of which has stayed in my system too long either. I think the iScan lasted the longest, as it had some unique features that were quite useful back when I was still using stand-alone devices for media playback.

You had 10-bit control for greyscale/gamma (three 1024-point 1D LUTs) and matrix controls for the CMS from what I remember.
No, his gamma controls don't use a LUT - that would be absurdly silly. It's a simple exponent calculation..

Quote:
I'm not trying to compare it to what madVR is doing though, just explain why it was much easier to get the results I wanted with it, and why I look forward to 10-bit output.
Your reasoning is “I did it wrong, therefore I'm waiting for methods to make it less possible for me to do it wrong”? Can you prove that with proper dithering, 10 bit vs 8 bit makes a noticeable difference? If not, it's your personal problem - not a problem of 8 bit.

Quote:
I'd probably work with 10, 20 or 100 points, time permitting, or possibly 10 points plus touch-ups where necessary. (fix any obvious errors) I realise it sounds like this goes against "hand-tuning" the LUT if you are letting the rest of the points be interpolated, but I'd prefer some sort of control rather than tell madVR what my display currently measures and hope it all turns out OK in the end.
10, 20 points? ArgyllCMS which you've previously denounced can calibrate using up to 2400 points. And yes, these are better calibrated than your rounded “hand-crafted” LUT.

Quote:
It's not just greyscale though, I'd love to be able to hand-tune 25/50/75/100% saturation for RGBCMY to make sure that the response is linear—I'm not sure that my display is when put into the "wide gamut" mode. (necessary to cover BT.709 gamut as it's a white LED display) Haven't had a chance to actually measure it and check for linearity yet.
Why RGBCMY? It's not related to video playback. And you don't need to hand-tune a LUT for this.

Quote:
Right, but then I feel like I'm basically guessing—yCMS only accepts XYZ or xyY data, right?
As opposed to what? XYZ and Yxy are raw measurement data. What else do you want?

Quote:
I suppose the simplest way to put it, is I just want to give it:
100%: 254,234,245
90%: 230,225,240
etc. and have it fill in the rest. (ideally this would be 10-bit though...)

No, this is the reason they didn't stay, and why madVR has replaced all stand-alone devices connected to my TV.

The main part of the problem isn't even madVR itself. (though I'm still not entirely convinced about 8>16>8 avoiding possible colour shifts near black when making gamma changes)

The problem is that you cannot disable a modern display's own internal LUTs/Processing, and have it display exactly what you send it. To get the best results, you're basically fighting against the display.

I feel that if you had used one of these external LUT boxes, you would immediately understand the problem, and I'm really not doing a good job of explaining it.
There's no difference between an external LUT box and an internal LUT box. Unless your video card is not outputting faithful signals. If so, that's a fault of your video card and should be easy to fix.

Quote:
When tweaking with the VideoEQ, it was not uncommon to see things such as this:
Code:
Pass|Output Value|Relative R|  G   |  B
  1 | 64, 64, 64 |   105%   | 100% | 98%
  2 | 62, 64, 66 |   104%   | 100% | 99%
  3 | 60, 64, 67 |   103%   | 100% | 99.2%
  4 | 58, 64, 68 |   103%   | 100% | 99.4%
  5 | 56, 64, 70 |   101%   | 100% | 102%
  6 | 55, 64, 69 |   099%   | 100% | 101%
10-bit precision helped dial in the results better, and the reason you have things jumping several steps at once sometimes is because you're essentially fighting against the display's internal processing to get the result you want from it.
You shouldn't need to fight against the display's internal processing. The challenge is not to calibrate your display to make it perfect, the challenge is to characterize your display the way it is, then simply use that knowledge to display the correct signal. This is how ICC profiles work. You aren't supposed to modify the way your display behaves at all.

Quote:
Sometimes changing one of the output values from the LUT box would not make a change on the display, and sometimes making a single digit change would have it jump 2-3%.
That's why you dither. As has been mentioned often enough, with proper dithering, 8 bit output vs 10 bit output becomes far less of an issue.
__________________
Forget about my old .3dlut stuff, just use mpv if you want accurate color management

Last edited by nand chan; 20th October 2011 at 17:46.
nand chan is offline   Reply With Quote
Old 20th October 2011, 17:38   #10252  |  Link
Nicola
Registered User
 
Join Date: Aug 2004
Posts: 29
subtitles have blackness instead of transparency in mpc-hc too.
Nicola is offline   Reply With Quote
Old 20th October 2011, 17:47   #10253  |  Link
Skwelcha
<(ovO)>
 
Join Date: Jun 2011
Location: Bremen, Germany
Posts: 50
When i switch to another videofile during playback MPC-HC crashes without any warnings, it just closes.
__________________
MadVR 0.92.9 x64, MPC-HC, LAV 0.70.2, GV-N1060G1 GAMING-6GD @ 385.41, LG OLED65B7D, Denon AVR-1912, Windows 10 Pro 1709 x64, i7-6700k
Skwelcha is offline   Reply With Quote
Old 20th October 2011, 17:55   #10254  |  Link
XRyche
Registered User
 
Join Date: May 2008
Posts: 211
Quote:
Originally Posted by Nicola View Post
subtitles have blackness instead of transparency in mpc-hc too.
confirmed
XRyche is offline   Reply With Quote
Old 20th October 2011, 17:57   #10255  |  Link
gbates
Registered User
 
Join Date: Feb 2004
Posts: 12
Quote:
When i switch to another videofile during playback MPC-HC crashes without any warnings, it just closes.
Quote:
subtitles have blackness instead of transparency in mpc-hc too.
Unfortunately +1
gbates is offline   Reply With Quote
Old 20th October 2011, 18:25   #10256  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
(1) Missing seek bar and OSD: Problem found and fixed.

(2) No transparency with subtitles: Can I have a screenshot that shows this problem, please? Does activating the "use 10bit luma texture" in the "trade quality for performance" make a difference?

(3) Crash on exit: I can't reproduce that here. Can anybody help me reproducing this crash, please?
madshi is offline   Reply With Quote
Old 20th October 2011, 18:38   #10257  |  Link
Skwelcha
<(ovO)>
 
Join Date: Jun 2011
Location: Bremen, Germany
Posts: 50
(4) crashes when switching to another videofile during playback

for me it's with every h.264 videofile, on the other hand it doenst crash on exit.
__________________
MadVR 0.92.9 x64, MPC-HC, LAV 0.70.2, GV-N1060G1 GAMING-6GD @ 385.41, LG OLED65B7D, Denon AVR-1912, Windows 10 Pro 1709 x64, i7-6700k
Skwelcha is offline   Reply With Quote
Old 20th October 2011, 18:42   #10258  |  Link
6233638
Registered User
 
Join Date: Apr 2009
Posts: 1,019
For anyone wondering about the benefits of linear light scaling, here's an example from Tron Legacy when downscaling.

Original/Normal

Original/Linear


These need to be viewed at 100% in the browser, and it's best to open them in separate tabs/windows and switch back/forth.


What you should notice is that the linear version maintains the original brightness of the source image, whereas the normal image is darker than intended.


Depending on the image being displayed, and the scaling algorithm used, the results may be more or less dramatic.

It definitely seems to introduce ringing with some of the scaling algorithms though, so look out for that.

Haven't done much testing with upscaling yet to produce some example shots, but I'm sure the results are similar.
6233638 is offline   Reply With Quote
Old 20th October 2011, 18:49   #10259  |  Link
noee
Registered User
 
Join Date: Jan 2007
Posts: 530
Quote:
Originally Posted by madshi View Post
(3) Crash on exit: I can't reproduce that here. Can anybody help me reproducing this crash, please?
It's happening on file/close too. Using a later version of MPC-HC, start a video, then exit the app or choose file/close (Ctrl+c). MPC-HC will then crash.

I get with LAVVideo and madVR video decoding. Any MKV with h.264 video.
noee is offline   Reply With Quote
Old 20th October 2011, 18:50   #10260  |  Link
nand chan
( ≖‿≖)
 
Join Date: Jul 2011
Location: BW, Germany
Posts: 380
Quote:
Originally Posted by 6233638 View Post
Depending on the image being displayed, and the scaling algorithm used, the results may be more or less dramatic.
Yeah, this is pretty much the best case scenario for linear light scaling: Lots of fine bright/dark patterns with very high contrast. I have (more or less unsurprisingly) not managed to see any worthwhile differences when viewing anime samples, but I do generally prefer linear light scaling myself.

Quote:
It definitely seems to introduce ringing with some of the scaling algorithms though, so look out for that.
Would you mind testing multiple scaling algorithms to see how much each of them ring when in linear light? I'm guessing Lanczos might be too extreme in this mode. What about Catmull-Rom and friends?
__________________
Forget about my old .3dlut stuff, just use mpv if you want accurate color management
nand chan is offline   Reply With Quote
Reply

Tags
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 14:11.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.