madVR - high quality video renderer (GPU assisted)

[madshi]

Originally Posted by ryrynz

Any chances of a deband on/off toggle?

What for? Do you really plan to disable/enable debanding all the time? Isn't a toggle to run through the various options good enough? Normally I'd guess that you just set it to a default value for all movies and just increase the strength for bad files. The current toggle should work fine for that purpose, no?

Originally Posted by ryrynz

Just want to run this one by you, one other filter I've often used in Avisynth is Psharpen, IMO it does a fantastic job of sharpening in a natural fashion.
Having come from sharpeners like LSFmod, Seesaw and Finesharp amongst many others over the years I've settled on Psharpen, would you take a look and see if you feel it's worthy for entry into MadVR?

Sorry, no feature requests.

Originally Posted by huhn

i'm going to get an colorimeter but it is very hard to get information for that at the moment i like to go with the X-Rite EODIS3 i1Display Pro. my aim is to create 3d lut for madvr but is this even the right device or i'm on the wrong path with this ?

There are different names for the meter I'd recommend. The name I remember is "i1 Display 3 Pro". Not sure if that's the one you have in mind.

Originally Posted by romulous

I notice that the default Zoom Player key mappings conflict with the way you set custom deband values. The arrow keys (all four of them) and F2 are all bound to particular Zoom Player functions, and when you try and set a custom value in madVR, the keys activate Zoom Player functions as soon as you press them. You can of course de-map all of those keys, set the custom values and then re-map those keys back to the original functions, but that is cumbersome. So, I was wondering if you could offer a way for Zoom Player to interact with the custom deband setting without having to de-map the hotkeys first?

Weird. In MPC-HC madVR's keyboard functionality actually "overwrites" the MPC-HC keyboard mapping. So if madVR is in a situation where it accepts arrow key presses, MPC-HC doesn't even see them. Not sure why it's different with ZP. Probably ZP uses a different method to get key presses.

In any case, if Blight wants to make the debanding customization accessible inside of ZP, he can directly edit the registry values. Changes there will only affect the next madVR instance, though, not the current. I don't think it's worth it, though. 99% of all users will probably just use the default presets, and for those arrow keys are not needed. You can also use Ctrl+ArrowKey btw, madVR will still accept that. Maybe that's a solution for you? Not sure if ZP has events registered for that, too?

Originally Posted by leeperry

BTW, I've been meaning to ask you for a while but would you consider keeping all the sub-sections in mVR's config panel permanently open?

Not sure if that's a good idea. Yes, currently all settings fit on one screen, but that might change sooner or later. At least when I implement profiles, users with many profiles will have a long list on the left side and it will not fit on the screen, anymore.

Remembering which sections are open would be a nice feature, but that's not something I can implement in a couple of minutes work, so I consider that a feature request and I don't accept feature requests at the moment (at least not any which cost me more than a couple of minutes of programming work).

Originally Posted by 6233638

I haven't had a chance to try out the presets in these latest builds yet, though the removal of NonGradientPenalty is somewhat concerning, as I found it to do a good job preserving details near black without artifacting as much as LocalContrast, or having as much of a performance hit.

This is going to take a few sentences to explain, but I'll try: The original intention of the NonGradientPenalty was to separate gradient pixels from non-gradient pixels, and to apply a stronger debanding to the gradient pixels. However, the implementation I chose was not really good for testing, because it practically lowered the "midDif" (formerly called "gradient") threshold for almost all pixels in the video, and only left it at the chosen value for very few pixels. So basically if you chose a NonGradientPenalty higher than neutral 1.0, this effectively reduced the midDif/gradient threshold for almost everything. In order to test the real benefit of NonGradientPenalty you had to dial midDif/gradient up the same amount as NonGradientPenalty, to counter this effect.

So recently I decided to change the implemention of NonGradientPenalty to no longer punish the majority of the pixels (namely the non-gradient pixels) and to instead boost the minority of the pixels (namely the gradient pixels). Practically I modified the NonGradientPenalty into a GradientBoost. After I did this I played with high GradientBoost values to see which effect it would have. I even went as high as a 40.0 boost (= 40x factor), but I couldn't really see much (if any) difference. So my final judgement was that the only positive effect of the original NonGradientPenalty was that it turned the midDif/gradient threshold knob down again. Or in other words: Instead of increasing NonGradientPenalty you could also have simply decreased midDif/gradient. So that's what I've done now. Instead of the original "Low" preset of gradient=1.4,nonGradientPenalty=1.2 I'm now simply using gradient=1.2, which produces pretty much the same results.

Of course there's a chance I did something wrong in my testing and nonGradientPenalty actually worked better than I think it did. But my tests with the reversed nonGradientPenalty / gradientBoost were pretty conclusive...

Originally Posted by 6233638

It's not specific to this build, but I've noticed that there are often "outlines" around objects when switching from fullscreen to windowed mode: [...]

This is with debanding disabled, but I'm guessing it's a bug introduced by AngleBoost. They disappear after a few seconds of playback.

I'm not sure what I'm supposed to see in that image?

Originally Posted by 6233638

Something unrelated to that however, is that I was wondering about seeking performance.
I was having to seek through a couple of videos quite a lot the other day, which got me to change my settings to try and improve it.
To give you an idea of what was happening, it was seeking 10s at a time, with the button repeating every 0.1s

The way I had madVR set up before, seeking was very slow and uneven.
Switching to CPU decoding rather than GPU accelerated decoding helped, and of course I unchecked the "disable playback start until render queue is full" option.
After making those changes, it would seek smoothly, but then ever 6-8 seeks, it would freeze up for a second.

The only way I was able to get completely smooth seeking without it pausing was to use Fullscreen Exclusive mode, Windowed Overlay did not seem to help at all.

Is this simply due to the fact that you get better performance in FSE mode, or could it be something else to do with how frames are presented?
I've been avoiding FSE mode ever since installing Windows 8 last year, as madVR keeps switching my display to the wrong refresh rates (23/59 rather than 24/60) but maybe I'll just have to stick with it until that is resolved.

---

And now that I'm using FSE mode again, I'm getting serious problems switching between FSE and Windowed mode when there is a refresh rate change involved. With the maximum queue sizes, I get about 1fps displayed for the first 20s when switching to FSE mode. Reducing the queue sizes to 4 brings that down to about 5 seconds.

This only affects Fullscreen Exclusive and not Fullscreen Windowed mode, and only when there's a refresh rate change involved.

I don't know what to say about this. These are problems I do not experience on my PC. Do many other users have the same problems? I'd have to be able to reproduce this on my PC to be able to do something about it.

Originally Posted by kasper93

I've noticed that it doesn't happen with all files. Try with this one https://dl.dropboxusercontent.com/u/16282309/madVR/sample_002.mkv Doesn't matter which MPC-HC version you use.

It crashes always at the same point. Watch from the beginning its around 0:01

And crash is somewhere in "fading" code. When I disable debanding or set settings to (medium/medium) there is no crash. (medium/high) crash always... I think madVR crash internally and then when it is poked by mpc-hc everything crashes.

The sample doesn't crash for me. But the hint with the "fading" code was very helpful - thanks! I've found a bug in that code and hope that the crashes are fixed in build 14 (see below)?

Originally Posted by dansrfe

It basically stopped working after replacing with the latest deband build. But since then I've tried replacing it back with the latest official build, resetting to default settings, updating to the latest MPC-HC, updating to the latest stable nvidia drivers, and it's still the same problem. I'm not exactly sure what caused the problem which is why I don't know where to begin.

According to your log switching to exclusive mode worked just fine. Can you clarify "stopped working"? Some more details might help. And please quote my post, that way I can directly click/link through our old communication. Otherwise I have to search through the thread to find your previous posts about this topic.

[madshi]

Test build 14, hopefully fixing the last crashes?

http://madshi.net/madVRdeband14.rar

[kasper93]

Originally Posted by madshi

The sample doesn't crash for me. But the hint with the "fading" code was very helpful - thanks! I've found a bug in that code and hope that the crashes are fixed in build 14 (see below)?

Good work! Everything is fine now. Thanks.

[Demonik]

Hello, i've found a problem with brightness (too high) while playing a file, problem which i traced back to build 0.85.4, 0.85.3 seems fine. attached a sample here: http://www.sendspace.com/file/dfuqnc
Thank you!

[romulous]

Originally Posted by madshi

Weird. In MPC-HC madVR's keyboard functionality actually "overwrites" the MPC-HC keyboard mapping. So if madVR is in a situation where it accepts arrow key presses, MPC-HC doesn't even see them. Not sure why it's different with ZP. Probably ZP uses a different method to get key presses.

In any case, if Blight wants to make the debanding customization accessible inside of ZP, he can directly edit the registry values. Changes there will only affect the next madVR instance, though, not the current. I don't think it's worth it, though. 99% of all users will probably just use the default presets, and for those arrow keys are not needed. You can also use Ctrl+ArrowKey btw, madVR will still accept that. Maybe that's a solution for you? Not sure if ZP has events registered for that, too?

Thanks, I've let bLight know. I also did some quick checking with Ctrl+Arrows, and that does seem to work...the first time. Something seems to be going wrong somewhere. This is what I did.

1. Mapped hotkey to deband custom settings toggle.
2. Hit that hotkey, Low displays on madVR OSD. Hit it again - this time Low displays, but also then lists the values you can change after it.
3. Use the Ctrl+Arrows to move along, change some values. Then hit F2 to save. All seemed to work fine.
4. Hit custom hotkey again, used arrows to move along, change some values - then when I hit F2, Zoom changed to audio mode.

I do note that only the first time I try and change some values, the 'F2' in the madVR OSD displays. Second and subsequent times, it does not (which is why Zoom enters audio mode, madVR is not attempting to intercept that F2 key). I don't know why madVR does not show the F2 prompt the second and subsequent times.

One other thing I noticed - this may be because it is a beta build, but when you select madVR from Zoom's filter properties, a box usually pops up (MadVR Video Renderer Properties). It gives the text madVR, and then the version number - the version number text is totally cut off with this version of madVR (it is fine with the last stable 0.86.11).

[ryrynz]

Originally Posted by madshi

What for? Do you really plan to disable/enable debanding all the time? Isn't a toggle to run through the various options good enough? Normally I'd guess that you just set it to a default value for all movies and just increase the strength for bad files. The current toggle should work fine for that purpose, no?

Just useful for quick comparisons. It's a lot easier spotting changes when you can enable and disable rather than running through all presets.

Originally Posted by madshi

Sorry, no feature requests.

All good.

[Ap3 n1nja]

Originally Posted by kasper93

I can't reproduce issue with your sample. Are you sure you are using http://madshi.net/madVRdeband13.rar ?

I got crashes for 0.86.10, 0.86.11, deband13, as well as deband14.

Originally Posted by kasper93

Anyway if madVR catch the crash just send report to madshi and he will fix it.

Okay, I sent a report. Can't confirm if it went through or not, however.

Originally Posted by vivan

Ap3 n1nja,
it's probably this issue http://code.google.com/p/xy-vsfilter.../detail?id=153

Originally Posted by sneaker_ger

There's a test build up for XySubFilter which should include such old fixes:
https://dl.dropboxusercontent.com/u/...XySubFilter.7z

Unfortunately, the crashes still happen with version 3.1.0.640...

Thanks for the help though, guys.

[rack04]

Does debanding only work when using the internal decoding? Also, I noticed in Test build 14 the version field in the filter properties isn't viewable.

[kasper93]

@madshi: There is one minor problem with subtitles in MPC-HC related to this "fading" feature. Basically subtitle blinks. I'm not sure this can be fixed easily fixed. Probably it would need changes in mpc. Anyway I wanted to say that in case you don't know it. Disabling ISR queue fix this glitch.

Originally Posted by rack04

Does debanding only work when using the internal decoding?

No, it works with every decoder. Anyone is using internal filters? I don't think they are particularity useful.

Originally Posted by rack04

Also, I noticed in Test build 14 the version field in the filter properties isn't viewable.

Version number is too long probably... But well it's test release, stable will have "normal" number ;p

Originally Posted by Ap3 n1nja

Okay, I sent a report. Can't confirm if it went through or not, however.

Well, if you send report you need to wait. I'm sure madshi will look at it in the right time

[Razoola]

I was wondering, is there some way to disable the 'EXCLUSIVE' message that appears in the top left corner then a video starts? This never used to happen but I recently updated my GFX card drivers and MPC-HC and after that the issue started happening (I don't know which update caused it to start). I have madVR set up to open video in full screen exclusive mode instantly and I'm not delaying the change by three seconds.

[ryrynz]

Originally Posted by Razoola

I was wondering, is there some way to disable the 'EXCLUSIVE' message

Not yet.

[Incriminated]

I really like to have a "toggle exlusive fullscreen mode on/off" in the madVR-trayicon right-click-context-menu, because most of the times I would not like having it enabled for multi-monitor video processing capability, but sometimes i like to have it on for some high-bitrate h264-content to avoid frame-dropps. Actually from this point on i have to click on "edit madvr settings", "rendering", "general settings", switch the "enable automatic fullscreen exclusive mode", click "apply", click "ok". Same for undo/reverting. Thats pretty annoying task, after you did this about 50 times xD.

With a context-menu shortcut it would be possible to reduce it to 1-click (2 click overall). Can you please integrate such function, madshi?

[michkrol]

Originally Posted by Incriminated

With a context-menu shortcut it would be possible to reduce it to 1-click (2 click overall). Can you please integrate such function, madshi?

Try the "automatic fullscreen exclusive mode - enable (disable)" keyboard shortcuts? Every other user would like something in the context menu and it would get cluttered quite fast with options

[Incriminated]

I wasn't awre of that, but thanks for you noting it. Sadly there is only a shortcut for "on" and one for "off". This way it at least works with two keys. Thanks.

[6233638]

Originally Posted by madshi

In order to test the real benefit of NonGradientPenalty you had to dial midDif/gradient up the same amount as NonGradientPenalty, to counter this effect.

Ah, I see.

Originally Posted by madshi

I'm not sure what I'm supposed to see in that image?

How it should look:


What often happens when switching between fullscreen and windowed mode:


Probably not the exact frame as I just hit pause, but you get the idea.
It doesn't happen every time, but does happen often, and disappears after a few seconds. (I guess once the queues are full again?)

Originally Posted by madshi

I don't know what to say about this. These are problems I do not experience on my PC. Do many other users have the same problems? I'd have to be able to reproduce this on my PC to be able to do something about it.

Well that's concerning, it's 100% reproducible herre. I'm not sure when this change happened either, as I've not used FSE mode for more than a year at this point - it used to work fine for me.

I've tried MPC-HC as well as JRiver now, and it's happening there too.



Is there any way to customize the debanding settings in the newer builds, without editing the registry? I've had a couple of videos now that needed high/high debanding settings to produce smooth gradients, but they were losing far too much detail.

[stasjok]

Originally Posted by madshi

The way the new subtitle interface works, madVR needs the GPU to perform subtitle alpha blending - and in 16bit. Unfortunately some older GPUs only support alpha blending in 8bit. For such GPUs you're likely to see an opaque black box instead of smoothly blended subtitles. I'm sorry, but at this point I don't see how I could fix it. It's a hardware limitation of your GPU and the only solution will probably be to upgrade your GPU to a newer model. I believe all newer GPUs can do alpha blending in 16bit without any problems. At least my HD4000 can.

Originally Posted by madshi

madVR v0.86.10 released

* fixed: #115: XySubFilter: opaque black box on older GPUs

I'm sorry. I didn't get it. I still get this "opaque black box" when smooth motion is enabled. Is it possible to fix this? Or is this "alpha blending in 16bit" required for smooth motion and there is nothing you can do? My GPU is GeForce 9600GT.

Smooth motion disabled

Smooth motion enabled

[Ver Greeneyes]

Originally Posted by 6233638

What often happens when switching between fullscreen and windowed mode:

I've seen this effect as well, but usually after leaving a video paused and going to bed instead. I don't use FSE mode that much because I'm used to multitasking. My GPU is a GeForce GTX 580 and I'm using the 331.65 drivers.

[iSunrise]

Originally Posted by 6233638

Ah, I see.

How it should look:


What often happens when switching between fullscreen and windowed mode:


Probably not the exact frame as I just hit pause, but you get the idea.
It doesn't happen every time, but does happen often, and disappears after a few seconds. (I guess once the queues are full again?)

Well that's concerning, it's 100% reproducible herre. I'm not sure when this change happened either, as I've not used FSE mode for more than a year at this point - it used to work fine for me.

I've tried MPC-HC as well as JRiver now, and it's happening there too.

Do you have a small sample for that so we can verify? I also am using an Nvidia GPU. Because I have never encountered such a thing and I am using fullscreen exclusive almost exclusively.

I am using a GTX 580 and 331.65 WHQL drivers.

[huhn]

i can still reproduce this bug: http://abload.de/img/resizingbug0fsot.png

and i completely reinstalled the pc since the last time i posted about this.

it only happens with an nvdia card (an 760) with lanczos 8 ar the hd 4000 is not affected

[madshi]

Originally Posted by kasper93

Good work! Everything is fine now. Thanks.

Thanks for your help in locating the bug!

Originally Posted by Demonik

Hello, i've found a problem with brightness (too high) while playing a file, problem which i traced back to build 0.85.4, 0.85.3 seems fine. attached a sample here: http://www.sendspace.com/file/dfuqnc
Thank you!

Thanks for figuring out which version introduced the problem. That might be helpful, but first I need more information. On a quick check the sample seems to play fine here. Are you using software or hardware decoding? If you're using hardware decoding, please check whether the same problem also occurs with software decoding. Can you please post a screenshot of the "too high brightness" with the sample you've uploaded?

Originally Posted by romulous

4. Hit custom hotkey again, used arrows to move along, change some values - then when I hit F2, Zoom changed to audio mode.

I do note that only the first time I try and change some values, the 'F2' in the madVR OSD displays. Second and subsequent times, it does not (which is why Zoom enters audio mode, madVR is not attempting to intercept that F2 key). I don't know why madVR does not show the F2 prompt the second and subsequent times.

The F2 key only works (and only has any purpose) if madVR displays it in the OSD. If madVR doesn't display it, it is also not accepted. The F2 is only accepted if you changed something which is worth saving. The custom debanding settings work differently than everything else. They're always automatically saved, and to a different place than all other settings. So F2 will not be shown if you change custom debanding settings. Really, the custom debanding settings are a hack I implemented because several people asked for it, but I think it will not be used by the majority of users in the future, that's why I didn't invest the time to implement a "nicely", with F2 etc.[/QUOTE]

Originally Posted by ryrynz

It's a lot easier spotting changes when you can enable and disable rather than running through all presets.

That may be true, but adding a further toggle would make the settings dialog more complicated, and I don't like to do that, unless there's a very good reason for it. Turning debanding on/off all the time to try spotting changes doesn't sound like something a typical user would do.

Originally Posted by Ap3 n1nja

I got crashes for 0.86.10, 0.86.11, deband13, as well as deband14.

Okay, I sent a report. Can't confirm if it went through or not, however.

Is your report the one with the text "opening mp4 file, does this every time but other files seems to work ok"? If so, the crash is somehow related to DXVA. Probably turning DXVA off will fix the crash. Not sure why the crash occurs, though. Maybe updating or reinstalling the drivers, or reinstalling D3D9 would help?

Originally Posted by rack04

Does debanding only work when using the internal decoding?

There is no madVR feature that is limited to only work when using the internal decoders. Except of course decoding.

Originally Posted by kasper93

@madshi: There is one minor problem with subtitles in MPC-HC related to this "fading" feature. Basically subtitle blinks.

Oh yeah, I understand the problem. If a fade in/out is detected, madVR has to go back 5 frames and rerender them, which also means the ISR has to go back in time 5 frames in rerender them, too. The ISR doesn't like that, it usually stutters for a bit when doing such a thing. This problem does not occur with any other subtitle renderer. I don't think there's anything I can do about it, except add an option to not go back those 5 frames. In that case the higher debanding strength would affect the whole fade in/out - with the exception of the first 5 frames. But I'm somewhat reluctant to add an option for a misbehaviour of the ISR.

Originally Posted by 6233638

How it should look:


What often happens when switching between fullscreen and windowed mode:


Probably not the exact frame as I just hit pause, but you get the idea.
It doesn't happen every time, but does happen often, and disappears after a few seconds. (I guess once the queues are full again?)

Is this a new problem only occuring with the latest test builds? Or does this occur with the current official release, too? I've never seen anything like that on my PC. I somehow doubt it's my fault, but what do I know. The problem is: Without being able to reproduce the problem, there's probably not much I can do about it.

Originally Posted by 6233638

Well that's concerning, it's 100% reproducible herre. I'm not sure when this change happened either, as I've not used FSE mode for more than a year at this point - it used to work fine for me.

I've tried MPC-HC as well as JRiver now, and it's happening there too.

Do you have another PC you could double check this on? Maybe I didn't do exactly what you did. Maybe if I knew the exact steps you're taking and the exact settings you're using, maybe I could reproduce it, too? If you think there's a good chance of that, please feel free to create a bug entry in the madVR bug tracker.

Originally Posted by 6233638

Is there any way to customize the debanding settings in the newer builds, without editing the registry?

Yes, define a custom keyboard shortcut for "debanding custom settings", press it, then use the arrow keys to modify the parameters.

Originally Posted by stasjok

I'm sorry. I didn't get it. I still get this "opaque black box" when smooth motion is enabled. Is it possible to fix this? Or is this "alpha blending in 16bit" required for smooth motion and there is nothing you can do? My GPU is GeForce 9600GT.

If this only happens with smooth motion enabled then it might be a bug which I could maybe fix. Or not. I don't know. Please report it to the madVR bug tracker.

Originally Posted by Ver Greeneyes

I've seen this effect as well, but usually after leaving a video paused and going to bed instead. I don't use FSE mode that much because I'm used to multitasking. My GPU is a GeForce GTX 580 and I'm using the 331.65 drivers.

Did this also happen with older drivers?

Originally Posted by huhn

i can still reproduce this bug: http://abload.de/img/resizingbug0fsot.png

And I've not seen anyone else reporting this. Still not sure if it might be a hardware issue. Or a driver issue. In any case, Lanczos 8 is not recommended, anyway. I'll probably remove non-recommended scaling algorithm in the final v1.0 build. That's still quite some time away, though.