madVR - high quality video renderer (GPU assisted)

[agustin9]

Originally Posted by madshi

OMG, yes, "Black 1997" gets treated as a valid tag, changing the black level of the video to 1997. Obviously that doesn't make much sense. But any "someTag someValue" combination is accepted as a file name tag. And "black" is a supported tag name. If you put the 1997 in brackets like "Men in Black (1997)", the problem should go away. That's also how IMDB names movies and how I'm storing movies on my server. Well, I'm glad we got this mystery solved.

Now the question is: Should I require a "tag=value" syntax? Currently I support "tag=value", "tag:value" and "tag value". And "black" is a supported tag.

tag=value to avoid confussions i think

[zero9999]

Originally Posted by madshi

Now the question is: Should I require a "tag=value" syntax? Currently I support "tag=value", "tag:value" and "tag value". And "black" is a supported tag.

Storing this kind of information in file names at all is not only error prone, but also very ugly. How about using NTFS alternate data streams instead?

The user would then simply run Add-Content V:\myvideo.mkv -Stream madVr "deint: ivtc" to add tags while the filename is kept clean.

[Razoola]

Personally I think it needs something like this in the filename so there can be no mistakes....

'Men in Black madVR[tags here]'

Basically the need for 'madVR' to be written in the file name.

[romulous]

Originally Posted by zero9999

Storing this kind of information in file names at all is not only error prone, but also very ugly. How about using NTFS alternate data streams instead?

ADS are NTFS only, say goodbye to using tags on non-NTFS drives if madshi implements that. For me, ADS is just an added complication, and shouldn't be done - needing to use another command to enter the tags when a simple rename of the file is all that is required as currently (and with renaming, you can see the tags without having to check the ADS as an added side benefit).

As I too use [ ] in my filenames, I would vote for '='

[madshi]

Ok, tags will require "=" in the next build.

Originally Posted by turbojet

Found a video that crashes with deint=ivtc and potplayer but 'send bug report' opens a blank browser window with mailto: but nothing happens. Is it reporting?

No. For some reason uploading the bug report seems to fail in your case, and mailto seems to fail, too. Your best bet is to not even try sending, but just close the bug report window. You should then get a crash report file on your desktop which you can PM to me.

IVTC currently only officially supports 8bit 4:2:0. Maybe the video you tested with had a different format?

Originally Posted by turbojet

Would it make sense to deband after downsizing to lower gpu load?

Hmmmmm... Good question. I believe all "artifact removal" algorithms should ideally be run on the unscaled source. But then, running debanding after downscaling might work fine, too.

Originally Posted by Werewolfy

In motion I can see the difference between medium and high. The face appears without any details and is very smooth with high.

Ok. Would still like to get feedback from other users, as well?

Originally Posted by Werewolfy

But I have a very weird issue with this new feature. When I'm watching a movie in 24hz, I get frames dropped when I activate a different strength for fade in/out. It happens only in exclusive mode. It doesn't happen in windowed mode or in 60hz mode.

Already been reported. Will be fixed in the next build.

Originally Posted by zero9999

How about using NTFS alternate data streams instead?

Wouldn't work for me. I'm storing my movies on a Linux based server, using ReiserFS. Maybe I'll think of an alternative tagging solution in some future version, but not now.

[DragonQ]

Originally Posted by zero9999

Storing this kind of information in file names at all is not only error prone, but also very ugly. How about using NTFS alternate data streams instead?

The user would then simply run Add-Content V:\myvideo.mkv -Stream madVr "deint: ivtc" to add tags while the filename is kept clean.

Oh yes people will love having to edit videos via the command line.

This is a Windows application, not a Linux one.

[DragonQ]

Originally Posted by Boltron

Noooo, I use [] in my naming convention.

Originally Posted by romulous

As I too use [ ] in my filenames, I would vote for '='

I'm curious as to why you'd use square brackets in file names for anything other than metadata? I don't know of any TV or film titles using square brackets, only normal ones.

Requiring = between tag name and tag value seems sensible.

[romulous]

Originally Posted by DragonQ

I'm curious as to why you'd use square brackets in file names for anything other than metadata? I don't know of any TV or film titles using square brackets, only normal ones.

I often change the filenames of files so that they make more sense to me - when I do, if I want to keep the original filename for some reason (the original filename can be important, so sometimes I do want to keep it, sometimes I don't), I put the original filename in square brackets at the end. I've used this same system for years, so I have built up a large number of files named in this way.

So, the file "foo.bar" (which tells me nothing about what the file actually is) would become "More descriptive name [foo].bar" for example.

[HeadlessCow]

Originally Posted by madshi

Ok, tags will require "=" in the next build.
Wouldn't work for me. I'm storing my movies on a Linux based server, using ReiserFS. Maybe I'll think of an alternative tagging solution in some future version, but not now.

While not every video format supports arbitrary tags, MKV does for sure and MP4 and AVI, I believe, do as well and those formats probably comprise 90% of the media that madvr users have. It would perhaps be useful to define a set of tags and read them directly out of the file rather than the filename.

[madshi]

As I said, I might look into alternative tagging solution in the future, but not now.

[6233638]

Originally Posted by madshi

I believe all "artifact removal" algorithms should ideally be run on the unscaled source. But then, running debanding after downscaling might work fine, too.

I actually did some testing on this, and with low resolution videos, and high strength debanding, it was actually beneficial as far as artifacts are concerned, to upscale externally and then have madVR process the image. That way you have pixel-sized "dither/noise" rather than giant upscaled noise. (I know you said you don't dither, but it's done something like dithering to smooth out those gradients)

However, debanding was considerably less effective after doing this - it needed much higher settings to work, and some things just weren't seen as gradients.

[Boltron]

Originally Posted by DragonQ

I'm curious as to why you'd use square brackets in file names for anything other than metadata? I don't know of any TV or film titles using square brackets, only normal ones.

Requiring = between tag name and tag value seems sensible.

I do use them for metadata. I use it to identify the video and audio streams for my own purposes. For example [BlurayRip - DTSMA 7.1] or [1080p@12.3 - DTS].

[DarkSpace]

Originally Posted by 6233638

I actually did some testing on this, and with low resolution videos, and high strength debanding, it was actually beneficial as far as artifacts are concerned, to upscale externally and then have madVR process the image. That way you have pixel-sized "dither/noise" rather than giant upscaled noise. (I know you said you don't dither, but it's done something like dithering to smooth out those gradients)

However, debanding was considerably less effective after doing this - it needed much higher settings to work, and some things just weren't seen as gradients.

How about just "scaling" the debanding values from video size to output size and applying debanding after scaling?
Can the debanding range (what pixels are looked at) be scaled for x and y dimensions separately? Can the AngleBoost algorithm be adapted to this? How about the banding/detail decision?

Does it perhaps even make sense to specify the debanding options for a fixed size and always scale from that size to the video (or output) size? (I know, that idea might just as well be incredibly bad.)

[turbojet]

Originally Posted by madshi

No. For some reason uploading the bug report seems to fail in your case, and mailto seems to fail, too. Your best bet is to not even try sending, but just close the bug report window. You should then get a crash report file on your desktop which you can PM to me.

IVTC currently only officially supports 8bit 4:2:0. Maybe the video you tested with had a different format?

I updated the bug tracker with the crash report. Switching madvr to debug was the only way to get the log. http://bugs.madshi.net/view.php?id=125

Originally Posted by madshi

Hmmmmm... Good question. I believe all "artifact removal" algorithms should ideally be run on the unscaled source. But then, running debanding after downscaling might work fine, too.

I remember when I messing with avisynth resizing and f3kdb there wasn't a noticeable difference whether it was pre or post resize. Maybe worth checking out to lessen load?

Originally Posted by madshi

Ok. Would still like to get feedback from other users, as well?

It doesn't appear noticeably faster for me.

[Megalith]

This might be a silly question, but does adding a 3D LUT to madVR override your Windows color profile completely, or are its settings combined with the color profile that you have set in Windows color management?

[dansrfe]

Originally Posted by Megalith

This might be a silly question, but does adding a 3D LUT to madVR override your Windows color profile completely, or are its settings combined with the color profile that you have set in Windows color management?

Ticking "disable GPU gamma ramps" under 'Calibration' means that Windows color profile will be disabled. If you've selected a 3DLUT then that will be applied after the Windows color profile if not disabled. It's better to calibrate without the Windows color profile in the middle and disable it upon playback if you're using a 3DLUT.

[Gagorian]

@madshi

Nice work on the debanding algorithm. The only issue I've found (like someone else already mentioned) is that subtitles blink occasionally, I normally use xysubfilter but if I download subtitles through the MPC-HC subtitle download interface it's always loaded into the ISR Could the backrender feature be activated/deactivated based on subtitle renderer?

Also unfortunately my iGPU on the HTPC (A10-5700) isn't powerful enough to run Jinc3-AR combined debanding in some scenes (without debanding it was fine). I'm considering getting the R9 270 GPU, does anyone have any other suggestions?

P.S.
Awaiting the convergence adjustment feature like a kid for christmas

[Ver Greeneyes]

Originally Posted by Megalith

This might be a silly question, but does adding a 3D LUT to madVR override your Windows color profile completely, or are its settings combined with the color profile that you have set in Windows color management?

Windows does not automatically apply installed color profiles to applications (though GDI can opt into this I think) - so the only thing that affects madVR is the videoLUT that tends to get loaded along with the profile.

madshi: How does "disable GPU gamma ramps" interact with the -a and -H options of collink?
1) For a 3DLUT created with -a, does madVR load the appended linear gamma ramp if "disable GPU gamma ramps" is not checked?
2) For a 3DLUT created with -H, does madVR load the appended gamma ramps if "disable GPU gamma ramps" is checked?

In other words: is "disable GPU gamma ramps" an accurate description of what the setting does (in the light of -a and -H), or would it be better named "override GPU gamma ramps" or something else?

(personally I don't run into this problem because I don't use -a or -H for my 3DLUT, since I want the videoLUT active while I'm multitasking, but it seems confusing)

[kasper93]

Originally Posted by Gagorian

Nice work on the debanding algorithm. The only issue I've found (like someone else already mentioned) is that subtitles blink occasionally, I normally use xysubfilter but if I download subtitles through the MPC-HC subtitle download interface it's always loaded into the ISR Could the backrender feature be activated/deactivated based on subtitle renderer?

It's know issue, and its ISR bug. Moreover I can say that one day it will be fixed on MPC-HC side. For now if you want to use ISR you have two options either disable subtitle queue in MPC-HC settings or disable madVR "fading" feature by setting both debanding options to the same preset.

If you want to use other subtitle renderer, you need to download subs and load them by XySubFilter there is no other option at the time.

[huhn]

Originally Posted by Ver Greeneyes

Windows does not automatically apply installed color profiles to applications (though GDI can opt into this I think) - so the only thing that affects madVR is the videoLUT that tends to get loaded along with the profile.

madvr outrput is changed by colorprofiles and disable gpu ramps disabled this. in window mode the hole desktop is srgb then.

and nearly all games are affected too