madVR - high quality video renderer (GPU assisted)

[leeperry]

I get a black screen with .14, .13 is fine. Something to do with that pesky W7SP1 platform update I can't apply again I guess?

I'll provide a log if obie's doesn't help.

[resides]

Heads up!

Black screen here too.

[har3inger]

Crispen edges (and the old FineSharp) suffer a lot in enhancing source aliasing.
Posterized (cel shading) images get dramatically changed by thin edges set at high values, but this is probably by design.

So far so good. Enjoying the new algos.

[Warner306]

Is it possible to improve the performance of Sharpen Edges for fair testing? It is far too expensive on my system.

[ryrynz]

Originally Posted by leeperry

I get a black screen with .14, .13 is fine. Something to do with that pesky W7SP1 platform update I can't apply again I guess?

I'll provide a log if obie's doesn't help.

Getting that here too. Found the issue.. DXVA upscaling/downscaling.

[cvrkuth]

Originally Posted by obieobieobie

I get a black screen in madvr after I updated to 0.89.14 from 0.89.13. I did not change any parameters, I just unpacked the new zip over the old install.

+1
Same here

[ryrynz]

Originally Posted by cvrkuth

+1
Same here

Is it DXVA upscaling/downscaling related for you too?

[cvrkuth]

Originally Posted by ryrynz

Is it DXVA upscaling/downscaling related for you too?

No, I use default settings for upscaling/downscaling, not DXVA ...

[pjred]

Me too, black screen 0.89.14 even with default settings.

[Magik Mark]

Madshi,

All features are running perfectly. Only draw back is it added addition 10ms on rendering time from my default of around 12ms. I'm now getting around 22 to 23ms

[wanezhiling]

The black screen issue of version 0.89.14 seems to be related to the video decoder, at least it happens with mpc-be and PotPlayer built-in software decoder here.

[FreeFall]

madshi,

Here is a new sample, the same original DVD sample I uploaded but has a re-cap of events in the first movie that plays before starting. I cut it to reduce the size of the original but is seems to be the cause of the problem.

https://www.dropbox.com/s/6ifrmre5xl...tdisc.zip?dl=0

Zoom control settings used.

move subtitles, to bottom of the screen / window.
automatically detect hard coded black bars.
notify media player about about cropped black bars, no more than once every 2 sec.
keep black bars visible if they contain subtitles, forever.

The black bars are being cropped, it seems to be working as intended in ver 0.89.9. I tried disabling the move subtitles option but it had no effect.

Thanks.

[krille]

Would it be possible to ad gpu and maybe cpu load to the madvr stats window?

[nlnl]

Quote:
Originally Posted by nlnl View Post
The problem: When source is video (29,97) and "force video mode' option is on, display does not switch to 59 rate, but swtches to 23.
If I set to 'video mode' using keyboard shortcut display switches.

Hmmmm... Which decoder? Software decoding? DXVA copyback? Native DXVA? Does this problem occur with all videos or just some?

LAV 0.66.0.27, DXVA copyback, the problem occurs with all videos (29.97 fr), not films.
Some times it switches when playing video first time but does not switch when playing the second.

[madshi]

Originally Posted by har3inger

Crispen edges (and the old FineSharp) suffer a lot in enhancing source aliasing.

Do you have a sample where this is especially obvious? I doubt there's much I can do about it, but I'd at least like to have a look at it.

Originally Posted by Warner306

Is it possible to improve the performance of Sharpen Edges for fair testing? It is far too expensive on my system.

As I said a couple posts before, performance of the new algos is not fully optimized yet. Performance optimizations costs time and a lot of thought. So I wanted to first get feedback about the quality (not performance) of these new algos. Maybe some algos are bad and should be deleted instead of optimizing them? Want to get confirmation that the algos produce nice quality and are worth spending the time to optimize them on.

Originally Posted by Magik Mark

All features are running perfectly. Only draw back is it added addition 10ms on rendering time from my default of around 12ms. I'm now getting around 22 to 23ms

I hope the slowdown only happens when activating the new sharpening algos? madVR itself didn't get slower, right?

Originally Posted by FreeFall

Here is a new sample, the same original DVD sample I uploaded but has a re-cap of events in the first movie that plays before starting. I cut it to reduce the size of the original but is seems to be the cause of the problem.

https://www.dropbox.com/s/6ifrmre5xl...tdisc.zip?dl=0

Zoom control settings used.

move subtitles, to bottom of the screen / window.
automatically detect hard coded black bars.
notify media player about about cropped black bars, no more than once every 2 sec.
keep black bars visible if they contain subtitles, forever.

The black bars are being cropped, it seems to be working as intended in ver 0.89.9. I tried disabling the move subtitles option but it had no effect.

Are you using the current official LAV version, or a newer nightly build? Newer nightly builds now send the subtitles to madVR instead of burning them into the decoded image. This has the advantage of madVR getting full control over the subtitle drawing and positioning.

When using a newer nightly LAV build, madVR should be able to move the subtitles either into the black bars or into the active video area. However, these are bitmap subtitles, which means that for optimal quality madVR has to draw them onto the decoded video image before doing any scaling, because only this way the subtitles will get the full quality benefit of the scaling algorithm you've selected. Having to draw the subs before scaling means that if you tell madVR to actually crop the black bars away, there's no black bars left for madVR to draw the subtitles into. So if you want the subs to be drawn into the black bars, you have to deactivate the option "crop black bars" in the zoom control settings. (Please note that this only applies to bitmap subtitles.)

Originally Posted by krille

Would it be possible to ad gpu and maybe cpu load to the madvr stats window?

Many things are possible, but at this point I don't consider that important enough to spend my time on. Too many other things on my to do list which I consider more important.

Originally Posted by nlnl

LAV 0.66.0.27, DXVA copyback, the problem occurs with all videos (29.97 fr), not films.
Some times it switches when playing video first time but does not switch when playing the second.

Can you please try if anything changes if you use software decoding? Also, can I have a debug log from a failure situation (I don't need one from a success situation)? Thanks.

[madshi]

madVR v0.89.15 released

http://madshi.net/madVR.zip

* fixed: black screen with some decoders (introduced in v0.89.14)
* fixed: black screen with DXVA processing in some situations
* added "superResRadius" file name tag for testing purposes

[KoD]

Bug report: AR video is broken in 0.89.14 and 0.89.15. Works ok with older versions. Tested this on 16:9 video with ass subs, which is for whatever reason displayed as 4:3 video (and the subs are offset badly). Sample here.

[madshi]

Originally Posted by KoD

Bug report: AR video is broken in 0.89.14 and 0.89.15. Works ok with older versions. Tested this on 16:9 video with ass subs, which is for whatever reason displayed as 4:3 video (and the subs are offset badly). Sample here.

Are you using a newer MPC-HC nightly build? And do you have "keep aspect ratio" enabled? If so, please try whether unchecking "keep aspect ratio" fixes the problem. Updating to the latest MPC-HC nightly build should also fix the problem - if it is what I think it is.

[KoD]

It happens with both latest two MPC-HC nightly builds (190 and 202). Having "keep aspect ratio" checked or not does not matter. The sample I provided is only 500Kb, so it's easy to download it and test.

Edit: looks like madVR ignores the container set AR (which is 16:9) and instead uses the video stream AR (which is 4:3).

[nlnl]

Can you please try if anything changes if you use software decoding? Also, can I have a debug log from a failure situation (I don't need one from a success situation)? Thanks.

Can not reproduce using software decoding.
Debug log:
https://drive.google.com/file/d/0B7a...w?usp=sharing]