View Single Post
Old 15th May 2016, 21:26   #37966  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by James Freeman View Post
madshi, is there a way to ignore wrong ST.2086 metadata and use peak white was selected by user?
The official test video from Sony has completely useless metadata and madVR simply shows all white picture (0.5 nit peak white)
I don't want to add options for broken files. However, I've PM'ed nevcairiel about the issue to check if we can maybe auto detect files with broken HDR metadata and handle them differently.

Quote:
Originally Posted by aufkrawall View Post
Too bad, your deringing filter seems to work wonders in general.
Especially with super-xbr, the difference in the cartoon example without SuperRes is enormous.
Try the next build, maybe it has improved slightly?

Quote:
Originally Posted by YxP View Post
What a remarkable addition this deringing is. Ringing is my number one annoyance, much more so than mild banding or aliasing. Whatever leftovers you guys are talking about, they definitely are not as bad as ringing itself :P Great work!


Quote:
Originally Posted by omarank View Post
While we are at it, I will share some thoughts:

1. 3D DCP video is a frame sequential 48fps stream with alternate left eye and right eye frames for 24 fps content. If LAV video decoder just flags and sends the stereo frames in the same way as it does for MVC content, no special handling of 3D would be required in madVR (I believe).

2. Currently, the black bar detection algorithm doesn’t work for RGB content (and also YUV content, IIRC), so it wouldn’t work for XYZ as well without extra work on your side. However, you can use the “Y” of XYZ for black bar detection as you do for YCbCr content because “Y” in XYZ and YCbCr are very similar as both have lightness information.
One step at a time.

Quote:
Originally Posted by SamuriHL View Post
Another positive vote for the new deringing. Very impressive results I must say! I only had limited time to check it out, but, I tried it on some Family Guy episodes and I'm quite impressed. Nicely done!
Glad you like it!

Quote:
Originally Posted by ryrynz View Post
Yeah I'm using it on Naruto Shippuden ATM, and I'm happy with the results. Dehalo Alpha was one filter I used a little quite some years back, in the end I think it destroyed the image too much so I stopped using it.


Quote:
Originally Posted by cxrn View Post
I'm trying to find your post, but I can't remember where I read it unfortunately.

Anyway, you said (if I remember correctly) something about adding a feature which added/rendered black bars into the video. This could be a misinterpretation, but me and several others noticed it and discussed it after the fact.

Is that a possible, viable feature to implement?
For which purpose?

Quote:
Originally Posted by iSunrise View Post
OMG, is this for real? The difference is unbelievable!

All the details still intact, while it only removes the ringing COMPLETELY.

Wow, I am speechless. This one feature alone is probably the best proof that madVR is so much better than anything else out in the market for video playback rendering by far.

Great work madshi!


Quote:
Originally Posted by 70MM View Post
Guys Ive just checked with my installer and he says we are installing madvr correctly in conjunction with JRiver. He said the madVR install he did was using the Program Files/madVR folder, not the one in the user folder so he updated the folder under users and will see how it goes....
JRiver MC uses the madVR version in the user folder. Ideally you should upgrade both versions. In theory you don't need 2 installations. You could remove the one from the Program Files folder. madVR would only work with JRiver MC then, though, and not in any other media player.

Quote:
Originally Posted by Stereodude View Post
Can I ask for further improvements in the 6:4 film mode used for p60 video? I don't know if you use it all that often which is why it's in the state it's in or if it's extremely hard to detect the pattern accurately, but it's not very polished from my use of it. It's probably the least polished feature I've found in madVR, by far. It sometimes takes several seconds to detect a pattern change resulting in the slow-mo playback look, it sometimes seems to displays frames out of order, or it flashes a random frame it shouldn't after a pattern change. It feels like an experimental / alpha feature. The 3:2 mode for i60 content works much, much, much, much better.
Samples are always welcome. I can't promise when I'll find time for this, but if I get samples, at least it gets on my (long) to do list.

Quote:
Originally Posted by QBhd View Post
It seems mixed upscaling (target 1024x768) is broken again... First pic is latest madVR (no settings changed), second pic is madVR v0.90.17
Try again with the next build. If it's still broken, please let me know.

Quote:
Originally Posted by fhoech View Post
Not only that, it's the only correct way - a linear 16-bit ramp with the goal to not alter the input in any way has to start at 0 and end at 65535, not 65280, so the CalibrationTester is simply wrong :-)
At least it seems more logical that way. However, and here comes the weird part: If you read the BT.709, BT.2020 etc specs, they explicitly say that limited range 8bit video has peak white at 235, while limited range 10bit video has peak white at 940. Using this logic, it could be 65280 instead of 65535.

In any case, when analyzing those 16bit ramps, madVR understands both types of LUTs to be "linear". But when creating a linear 16bit ramp, madVR uses 257.

From what I remember, in some OSs and with some GPU drivers, 256 was used and sometimes 257.

Quote:
Originally Posted by leeperry View Post
Turns out it seems to work stunningly well with SuperRes LL@2 + AB@100 so please don't remove AB from SR anytime soon, still playing around but I certainly like what I see in combination with vanilla 100% LL SSIM2D & vanilla sxbr(whose main problem was amplifying existing ringing as I understand it).

Outrageous results really, hats off
Glad it works well for you!

Quote:
Originally Posted by CarlosCaco View Post
I know we already talk about this but in would like to ask you to rethink about and if its possible implement on madvr
The resolution of my tv is 1366x768 (Windows says that native is 1360x768 but in tv datasheets says 1366x768, i use the two resolutions depending on the movie fps)
That implies aways upscaling or downscaling the movies
I almost watch 720p content and on this movie resolution madvr don't allow me to use upscaling refinements only image refinements because is small scaling factor near the movies native resolution
I like to use some sharpen because its not a big screen and the sharpen plus enhance details give me for depht
But seens that using some of the sharpen algos and upscaling after produces a little distortion on image, more bloated, and ant bloating filter poduces some extra blur and great loss of depht
I really like sharpen edge but seens that to get more depht with ab before scaling i have to use bigger strenghts but ant bloat sometimes produces some strange distortions and much blur when use before scalling
Adaptive sharpen plus ab is great but is unusable in this scenario because using both give me more distortion and dont reduces the oil paint apparence../
In some past version (can t remember the version that this has changed) madvr allowed to use upscalling refinements on such smallers scaling factors even 1919 to 1920 on my other laptop that is full hd and the effect was pretty good much less artificial... And for last this limitation don t let me use super resolution wich produces good natural sharpen results even with this small scalling factor... Same for supersampling that don t allow using upscalling refinements with this resolution, but in the past if i checked aways when upscaling is required the refinements were used... I don t really have much interest in supersampling but really want to use upscaling refinements when scaling 720 to 768 especialy the antibloat
Im on my phone now but this weekend i can make some screenshots examples of
Ant bloat loss of quality if used pre resizing, i made some tests scaling on avysinth... To see the effect after scalling...
Can you allow use upscalling refinements to smal scaling factors?
It's difficult to make this configurable. Some users may share your opinion, others might not. If I added an option for this, it would be hard to understand for most users what the option means exactly.

Maybe you could activate supersampling (see image doubling section)? If you do that, upscaling refinement should work as expected. Of course this solution costs quite a bit of extra performance.

Quote:
Originally Posted by XMonarchY View Post
Thanks. Madshi posted comparison picture and the one for Dark Halo Removal was VERY blurry.
I think the VERY blurry image you mean was not from madVR, but from the AviSynth script DeHalo_alpha.

Quote:
Originally Posted by Werewolfy View Post
Your dering algorithm is very promising, it derings very well when maintaining details. It's way above DeHalo and other deringing filters I've seen so far. But sometimes it adds some artefacts as you can see here :

http://postimg.org/image/efe86s6f5/full/

Also note that the color of the door is slighlty different and the lines on the bottom of the wall just next to the door are narrower.

Here I found it too agressive on 2 spots :

http://postimg.org/image/durdyqd8h/

Could you improve a little bit the situation? Personnaly, I would prefer to have a lower deringing strength with fewer artefacts and better details preservation.
It's not that easy. Edges are either detected as being ringing or not. If they are detected as being ringing then I try to remove them. Different "strengths" don't really make too much sense here. Well, ok, there are some thresholds I'm using. But if I allow you to lower those, you'll get less ringing removed and maybe fewer artifacts in some places, but then you'll get more ringing left in the image, and some other artifacts instead. I don't think it would really improve the situation much.

But please try the next build, it has some (smallish) improvements. Maybe it works better for you? It's not perfect, though. And it probably never will.

Quote:
Originally Posted by Warner306 View Post
Does anti-ringing do anything for compression artifacts:

"The signal is bandwidth-limited, discarding too much information for high frequencies."
Anti-ringing or Deringing?

Quote:
Originally Posted by BluesFanUK View Post
madshi, is there a way to resize the MadVR settings window or collapse all the profiiles upon opening the window?

I've finally managed to set up a number of profiles for different resolutions under both processing and scaling algorithms, but it's a pain having to scoll to find each one.
Not at this point. Cosmetical issues are not a priority at this stage of development.

Quote:
Originally Posted by Sunspark View Post
I think processing image enchancements is before scaling, and scaling algorithims upscaling refinement is after.. so it is possible to sharpen edges (random example) twice if you toggle it in both image enchancements and upscaling refinement panels as they can be set independently. The checkbox is not either/or exclusive.
The image enhancements controls are meant to be used if you play a source which is too soft, regardless of whether you upscale or downscale it.

The upscaling refinement controls are meant to repair softness that most upscaling algorithms introduce.

So these 2 sets of options really have a different purpose, and it may make sense to use both at the same time.

That said, this is all a work in progress. I absolutely plan to make the whole settings mess less complicated at some time in the future.

Quote:
Originally Posted by Sunspark View Post
Possible bug, "disable calibration controls for this display" still allows use of the enable gamma processing function so it is unclear what the difference would be between disable and "already calibrated".
Hmmmm... Yes, I suppose enabling gamma processing probably doesn't make much sense if "disable calibration controls" is set. On the other hand, I don't consider it a crucial problem. Things like this should be fixed before version 1.0. But atm other things are more important.

Quote:
Originally Posted by phm View Post
Perhaps this is not the first time being discussed but I am curious as to whether I am doing something wrong.

I have downloaded a video with hdr, this one in particular: http://demo-uhd3d.com/fiche.php?cat=uhd&id=144

When I use enhanced video renderer it looks fine but when I use madvr it looks wrong: http://imgur.com/XPLQ320

Is there a setting I must change or is this a bug or compatibility issue?
As you already found out, the metadata in that file is broken. But I'll see if I can find a workaround to play this file with decent quality.

Quote:
Originally Posted by oddball View Post
Can someone tell me why I can no longer use MadVR in Windows 10 unless I enable DX11?
Issues like this are hard to diagnose from afar. Try reinstalling/updating DirectX. Newer OSs sometimes don't ship with full DX9 support. But the key question is: If this used to work, what has changed to break this? It's almost impossible for me to say with the limited information I have.
madshi is offline   Reply With Quote