madVR - high quality video renderer (GPU assisted)

[Stereodude]

feel free to step through each frame IVTC isn't needed for this file.
That all depends on if the decoder honors the soft pulldown flag or not and then what it does after that (deinterlace / pass as interlaced / etc).

the output is already 23p the "only" issue is that madVR isn't switching to 23p and stays at 60 hz because source filter says it is 29p.
Again, it depends on what the decoder does. The various HW decoders all do slightly different things. For example, the Quicksync HW decoder will output that clip as progressive 29.97Hz video without a repeated frame and madVR will only find a 2:2 pattern. avcodec (software decoding) seems to be needed for proper handling of this and many other DVDs as well as HD 1080i MPEG-2 content.

don't get me wrong madVR IVTC shouldn't fail in this case but this file is not a typical DVD.
From what I've seen it is a typical NTSC DVD. Most DVDs I've tested behave the same way, and I'm not talking about anime or small studio releases of non-theatrical content, but major studio releases of recent theatrical motion pictures. A few will indicate to the processing chain that they are interlaced, but not many.

i guess the muxer did something to it but this is just a guess.
The .m2v extracted straight from the DVD does the same thing, as does playing the DVD. Putting it in a .mkv container didn't change the playback behavior of the MPEG-2 content at all.

[Aktan]

The .m2v extracted straight from the DVD does the same thing, as does playing the DVD. Putting it in a .mkv container didn't change the playback behavior of the MPEG-2 content at all.

It doesn't matter if you change containers if the same demuxer is being used. Which demuxer are you using?

[huhn]

i tested now quicksync, dxva native (AMD) and software. all output 23p flagged as 29p.

there is no repeated frame in the stream and no waved either.

The .m2v extracted straight from the DVD does the same thing, as does playing the DVD. Putting it in a .mkv container didn't change the playback behavior of the MPEG-2 content at all.

correct but why is VFR and missing all stream flags except range and interlaced?

adding FPS=23 should be an ok workaround for now for soft telecine DVDs. it would be nice if the source filter will do this by default with a disc that has clear flags like these:

Video
ID : 224 (0xE0)
Format : MPEG Video
Format version : Version 2
Format profile : Main@Main
Format settings, BVOP : Yes
Format settings, Matrix : Custom
Format settings, GOP : Variable
Duration : 15mn 8s
Bit rate mode : Variable
Bit rate : 7 726 Kbps
Maximum bit rate : 9 800 Kbps
Width : 720 pixels
Height : 480 pixels
Display aspect ratio : 16:9
Frame rate : 23.976 (24000/1001) fps
Standard : Component
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Scan order : 2:3 Pulldown
Compression mode : Lossy
Bits/(Pixel*Frame) : 0.932
Time code of first frame : 01:00:05:00
Time code source : Group of pictures header
Stream size : 837 MiB (82%)
Color primaries : BT.601 NTSC
Transfer characteristics : BT.601
Matrix coefficients : BT.601

[Stereodude]

It doesn't matter if you change containers if the same demuxer is being used. Which demuxer are you using?
The LAV filters (internal with MPC-HC 1.7.10)

[Stereodude]

i tested now quicksync, dxva native (AMD) and software. all output 23p flagged as 29p.

there is no repeated frame in the stream and no waved either.
I'm not sure I agree with you on all those test cases / permutations. If what you're saying is true there would be no 3:2 pattern for madVR to see (after enabling deinterlacing with forced film mode turned on).
It would see a 2:2 pattern due to the progressive content. Using quicksync HW decoding it definitely finds a 2:2 pattern, which is not what it finds using software (avcodec). Further, the IVTC algorithm wouldn't get confused and pass combed frames if it was being fed progressive frames.

correct but why is VFR and missing all stream flags except range and interlaced?

adding FPS=23 should be an ok workaround for now for soft telecine DVDs. it would be nice if the source filter will do this by default with a disc that has clear flags like these:
Regardless, you currently can't play it back correctly at 24/1.001Hz unless you enable deinterlacing and forced film mode (IVTC) in madVR. And, the IVTC detection pukes twice in this scene.

[huhn]

I'm not sure I agree with you on all those test cases / permutations. If what you're saying is true there would be no 3:2 pattern for madVR to see (after enabling deinterlacing with forced film mode turned on).
It would see a 2:2 pattern due to the progressive content. Using quicksync HW decoding it definitely finds a 2:2 pattern, which is not what it finds using software (avcodec). Further, the IVTC algorithm wouldn't get confused and pass combed frames if it was being fed progressive frames.
i agree i have no clue why madVR is able to see a 3:2 pattern. there is no combed frame with disabled deinterlacing.

but why would it get confused in the first place this file is clearly soft telecine this means the repeated field is bit identical to another field.

Regardless, you currently can't play it back correctly at 24/1.001Hz unless you enable deinterlacing and forced film mode (IVTC) in madVR. And, the IVTC detection pukes twice in this scene.

it works flawlessly with disabled deinterlacing at 23p here for me.

[Georgel]

Hello again!

Thanks a lot for answering my question about my crash log file.

After your response, I think that something might be off with my drivers, because my laptop has drivers from Acer and not from Nvidia, so they might not be compatible with something.

Nvidia drivers return blue screen 5 minutes after installing, video memory error, so something might be wrong on my side.

I cannot wait to upgrade to a laptop with a gtx980 desktop inside:devil:




Now I have another question or two for people who can help.

I have dithering set to error diffusion 2 without colored noise or changing patterns.

1.But does LAV video filter apply another layer of dithering?

2.Another question : I might want to disable the entire dithering process, is it possible, or would it degrade image quality?

3.And another thing: How can I set in mpc-hc the player to copy the entire file playing in RAM? I have 16 gb of RAM and most things I do never use more than 2-4gb, so I have enough to spare, and I would want my videos in RAM while playing for better decoding, and maybe reduction of micro stutter (?).

[e-t172]

1.But does LAV video filter apply another layer of dithering?

No. LAV only dithers if it has to convert to RGB. Which is not the case when using madVR (or any other sane video renderer for that matter).

2.Another question : I might want to disable the entire dithering process, is it possible, or would it degrade image quality?

Dithering improves image quality by preventing banding artifacts that arise from conversion to integer in the final stage of the video rendering pipeline. Disabling it will only make things worse.

3.And another thing: How can I set in mpc-hc the player to copy the entire file playing in RAM? I have 16 gb of RAM and most things I do never use more than 2-4gb, so I have enough to spare, and I would want my videos in RAM while playing for better decoding, and maybe reduction of micro stutter (?).

You won't get "better decoding" or "reduction of micro stutter" by doing that. Whether the file is played from RAM or disk has absolutely zero effect on decoding. It doesn't have any effect on the timing of video presentation either, because it's massively buffered at various stages of the video chain anyway precisely to avoid such issues. Note that LAV Splitter, IIRC, will by default buffer a large portion of the video anyway (hundreds of megabytes).

[nevcairiel]

No. LAV only dithers if it has to convert to RGB. Which is not the case when using madVR (or any other sane video renderer for that matter).

Thats only half-true, it dithers whenever it reduces the bitdepth - well at least on the most common optimized conversion paths.
But it doesn't dither when you use madVR and didn't screw up the LAV settings (ie. don't disable any output formats).

[Georgel]

No. LAV only dithers if it has to convert to RGB. Which is not the case when using madVR (or any other sane video renderer for that matter).



Dithering improves image quality by preventing banding artifacts that arise from conversion to integer in the final stage of the video rendering pipeline. Disabling it will only make things worse.



You won't get "better decoding" or "reduction of micro stutter" by doing that. Whether the file is played from RAM or disk has absolutely zero effect on decoding. It doesn't have any effect on the timing of video presentation either, because it's massively buffered at various stages of the video chain anyway precisely to avoid such issues. Note that LAV Splitter, IIRC, will by default buffer a large portion of the video anyway (hundreds of megabytes).

Thanks a lot for all the data!

About the image in the first post, where you fiind out what type of input your monitor accepts, if I have 4:4:4 what does it mean, and how do I make use of it when using mpc-hc and madVR?

[Asmodian]

About the image in the first post, where you fiind out what type of input your monitor accepts, if I have 4:4:4 what does it mean, and how do I make use of it when using mpc-hc and madVR?

Please read Chroma Subsampling (https://en.wikipedia.org/wiki/Chroma_subsampling).

You do not need to do anything to take advantage of a 4:4:4 display, it simply means you have full resolution chroma. If you have a 4:2:2 or 4:2:0 display chroma scaling options are less important.

[Georgel]

Please read Chroma Subsampling (https://en.wikipedia.org/wiki/Chroma_subsampling).

You do not need to do anything to take advantage of a 4:4:4 display, it simply means you have full resolution chroma. If you have a 4:2:2 or 4:2:0 display chroma scaling options are less important.

Thanks a lot!:thanks:







I took a look at the samples posted on the first page, madTestPatternSource. I wanted to ask if for 23 , 24 , and 25 fps files, it is normal that the vertical black bars would jump from one position to the next, without being overly smooth.

This happens with all bars, except the top, bottom and sometimes the second from top to bottom. But the rest of the bars are jumpy for me, not like jumping dropping frames, but more like micro stutter. I have smooth motion active.

Another strange thing is that I get a reported frame rate of

60fps (not constant, it starts from 180fps then jumps to 170 and so on until it touches 62 then it is lowering itself with 0.1 per 2-3 seconds) for all 23, 24, and 25 test files.

[Aktan]

Except the scenes are not fades. They just cause misdetections. I will cut one of them from the DVD and upload a sample.

Edit: Here's a sample (https://www.sendspace.com/file/tp2lq7). Enable deinterlacing and watch.

Okay I've taken a look at the sample. I'm not sure what I'm missing here. With default settings, madVR turns off deinterlacing because the upstream says it's not needed. Fine, you said to turn it on. So I did turn it on with automatic source type detection. It detected fine and I saw no problem. Okay, you said IVTC. I turned off automatic source type and force film mode. Then and only then do I see the problem you describe, but I forced it to film mode when originally it completely thought there was no need to deinterlace. Why force it?

Edit: I reread later posts where you explain why. My bad. Hmm...

Here is what ffmpeg says about the sample:

Input #0, matroska,webm, from '..\..\..\Castle S05E04 IVTC Failure sample.mkv':
Metadata:
encoder : libebml v1.3.1 + libmatroska v1.4.2
creation_time : 2015-12-17 04:04:26
Duration: 00:00:22.61, start: 0.004000, bitrate: 5476 kb/s
Stream #0:0(eng): Video: mpeg2video (Main), yuv420p(tv), 720x480 [SAR 32:27
DAR 16:9], Closed Captions, max. 9800 kb/s, SAR 186:157 DAR 279:157, 59.94 fps,
59.94 tbr, 1k tbn, 59.94 tbc (default)
Metadata:
BPS : 5093403
BPS-eng : 5093403
DURATION : 00:00:22.606000000
DURATION-eng : 00:00:22.606000000
NUMBER_OF_FRAMES: 542
NUMBER_OF_FRAMES-eng: 542
NUMBER_OF_BYTES : 14392684
NUMBER_OF_BYTES-eng: 14392684
_STATISTICS_WRITING_APP: mkvmerge v7.7.0 ('Six Voices') 32bit built on Feb
28 2015 23:23:00
_STATISTICS_WRITING_APP-eng: mkvmerge v7.7.0 ('Six Voices') 32bit built on
Feb 28 2015 23:23:00
_STATISTICS_WRITING_DATE_UTC: 2015-12-17 04:04:26
_STATISTICS_WRITING_DATE_UTC-eng: 2015-12-17 04:04:26
_STATISTICS_TAGS: BPS DURATION NUMBER_OF_FRAMES NUMBER_OF_BYTES
_STATISTICS_TAGS-eng: BPS DURATION NUMBER_OF_FRAMES NUMBER_OF_BYTES
Stream #0:1(eng): Audio: ac3, 48000 Hz, 5.1(side), fltp, 384 kb/s (default)
Metadata:
BPS : 384000
BPS-eng : 384000
DURATION : 00:00:22.176000000
DURATION-eng : 00:00:22.176000000
NUMBER_OF_FRAMES: 693
NUMBER_OF_FRAMES-eng: 693
NUMBER_OF_BYTES : 1064448
NUMBER_OF_BYTES-eng: 1064448
_STATISTICS_WRITING_APP: mkvmerge v7.7.0 ('Six Voices') 32bit built on Feb
28 2015 23:23:00
_STATISTICS_WRITING_APP-eng: mkvmerge v7.7.0 ('Six Voices') 32bit built on
Feb 28 2015 23:23:00
_STATISTICS_WRITING_DATE_UTC: 2015-12-17 04:04:26
_STATISTICS_WRITING_DATE_UTC-eng: 2015-12-17 04:04:26
_STATISTICS_TAGS: BPS DURATION NUMBER_OF_FRAMES NUMBER_OF_BYTES
_STATISTICS_TAGS-eng: BPS DURATION NUMBER_OF_FRAMES NUMBER_OF_BYTES
At least one output file must be specified

[Aktan]

I also noticed if you uncheck "only look at pixels in the frame center..." The messing up happens less often.

Edit: After doing more testing, since it seems the video decoder honor the flags anyway and outputs at 23p, I agree with huhn and just disable deinterlacing and set output 23p.

[nanpan]

I use Windows 10 Pro x64(10586.29)+AMD Fury X.
When I update VGA driver to new driver【radeon-crimson-15.12-win10-64bit】

The NNEDI3 in madVR(chroma upscaling or image doubling)all could not work.
Maybe the madVR NNEDI3 OpenCL kernel something wrong in new VGA driver.

I always use SVP4 Pro+madVR together.
SmoothVideo Project SVP(3.17a and 4 pro)has the same error when using GPU-acceleration(OpneCL).

Any idea to fix it ?

If there is no way to fix it, I just only go back to old VGA driver.
Because old VGA driver(not crimson)all works fine.

[djsolidsnake86]

0.89.18 playback is stopped (only audio work) with a cloned 2nd monitor
reverting to .17 is fine

[Asmodian]

0.89.18 playback is stopped (only audio work) with a cloned 2nd monitor
reverting to .17 is fine

v0.89.18 works with cloned monitors on my system. Windowed Overlay only shows an image on one screen but D3D9 or D3D11 windowed or exclusive works fine, I have an image on both screens.

Win10 x64, Zoom Player 11.1, Nvidia 980 Ti with driver version 359.06.

[CarlosCaco]

Hi madshi and everybody, first i want to thanks madshi for this great work and all the people in the forum, i never posted anything here because everytime i have a doubt someone already explaned here
i have a question that i don't know if has already been discuted here, i searched but did not find

i watch my movies on an relative old laptop with an disgusting optmus system,
after milions of tests i can assume that my display output 8 bit, my question is:

if i play a 10 bit content on my 8 bit display, i'm loosing some color information?
i thought that madvr converts or dither down to 8 bit, but in my osd shows: h264, 10bit, 4:2:0 -> P010, 10 bit, 4:2:0

my laptop display output 1920x1080p with fixed refresh rate 60 hz (59.63046)

I'm using madVR v0.89.18 on latest mpc-hc and lav nitghly buid

software decoding on lav with untouched output levels (thats another question should i left untouched or pc levels on lav? there´s any effect on the image if is madvr that does the conversion to rgb? )

my madvr settings:

d93d exclusive
the display is calibrate to bt 709 power curve 2.20 with i1d2 (not perfect acurated but good)
deband=off
chroma= Lanczos4 AR
Downscale: catmul-rom AR LL
Upscale: Lanczos3 AR
enhance details= 1.00 on image refinements
smooth motion = on
ordered dithering with colored noise=enable, change patterns =disabled
trade quality options are the standard but with 16bit buffer option disabled (don´t see perfomance improvement if i active this option)


the movie is 1080p fps=23.976 16:9 so theres no image scaling or AR adjust

i notice that the movie has no banding at all, but seens slighty blurry, i do know if is because the 10 bit on 8 bit display, or because the encode,or the movie itself, or smooth motion (don´t see look like the same slighty blurry aspect on other movies with smooth motion on) is a web serie: the man on the high castle, i watched other rips and seens that the movie was studio mastered with some sort of heavy fxaa filter like pc games

its the first 10bit encode that i watch so i don´t have other references to compare

i watched the life of pi e and exodus HDR trailers that are 10 bit too, but my gpu can´t handle downscale content to 1080p even with overclock, bilinear or dxva scalling, so looked like a slide show (but look likes great, vivid colours and contrast and great image quality, artificial because special effects but very pleasent experience)

i will found other 10bit encodes to compare, but i want to know if i loosing my time watching this on a 8bit panel
thats my question, i'm sorry about my bad english, i'm from Brasil
thanks

[Thunderbolt8]

Im a bit confused by the terms 8-bit, 10-bit etc. in relation to the number of colours. my monitor has 16.7 million colours, but I dont know if that is 8-bit or 10-bit.

[Stereodude]

I also noticed if you uncheck "only look at pixels in the frame center..." The messing up happens less often.
That might work for DVDs, but it doesn't work for broadcast 1080i60 HDTV where there are lots of logos crawls and other crap outside of the center that will mess it up if you don't limit it to the center.

Edit: After doing more testing, since it seems the video decoder honor the flags anyway and outputs at 23p, I agree with huhn and just disable deinterlacing and set output 23p.
It seems the HW decoders ignore the soft pulldown flags, not honor them. Further, I'm not sure what you two are seeing, but disabling deinterlacing with software decoding (avcodec) with the video card outputting 24/1.001Hz looks terrible and the render times are higher to boot. Maybe it's engaging smooth motion. I haven't spent a lot of time with it.

[Aktan]

That might work for DVDs, but it doesn't work for broadcast 1080i60 HDTV where there are lots of logos crawls and other crap outside of the center that will mess it up if you don't limit it to the center.

Ah, well then damn if you do, damn if you don't? I thought for broadcast, it be pure interlace anyway, not film video. I guess you need to change when needed? I don't know..

It seems the HW decoders ignore the soft pulldown flags, not honor them. Further, I'm not sure what you two are seeing, but disabling deinterlacing with software decoding (avcodec) with the video card outputting 24/1.001Hz looks terrible and the render times are higher to boot. Maybe it's engaging smooth motion. I haven't spent a lot of time with it.

Sorry, I forgot to mention I was using LAV Video to software decode. It being MPEG2 and DVD resolution, I didn't think there be a need for HW acceleration. When you said looks terrible, what you mean. Lack of post processing or jittery movement?

[CarlosCaco]

Im a bit confused by the terms 8-bit, 10-bit etc. in relation to the number of colours. my monitor has 16.7 million colours, but I dont know if that is 8-bit or 10-bit.

I had this doubt too, since is not register in edid, or in display tecnical specs or in any site,
i found a lot of posts asking about, in one madshi says to use the test paterns especialy the colours and try different bitdepths, and see what looks best, also i made the procediment to know if my display supports 10 bit (there is a thread about in the forum) i tested milion times 6bit, 7bit and 8bit now i clear see the diference in the patterns in my case bellow 8bit produce banding and above 8bit i can see more banding and seens that the images has no dithering
after all of this i calibrated my display with dispcalgui and on iccprofileinfo was showing 8 bitdepth

[ibius]

Im a bit confused by the terms 8-bit, 10-bit etc. in relation to the number of colours. my monitor has 16.7 million colours, but I dont know if that is 8-bit or 10-bit.

When you do the math, 16.78 million is 8-bit, displaying 256 red, 256 green and 256 blue shades. Look up specs of your monitor, depending on how they decided to phrase it you might have to dig deeper, better yet find a detailed review. Look for 'Colour Palette' of your panel, if it says '1.06 billion' but the 'Colour Depth' says '8-bit', then it's not a true 10-bit panel (like 99% out there), but is capable of producing 1.07 billion colours. That this is done with an 8-bit colour depth and an additional Frame Rate Control (FRC) stage (Temporal Dithering).
But I doubt you will see a difference outputting 10-bit on a '8-bit + FRC' panel vs properly dithered madVR output on 8-bit panel.

On a side note, did anyone with a '8-bit + FRC' panel make some tests displaying 10-bit with madVR with dithering disabled to compare panel's dithering to that of madVR?
And, is madVR's Ordered dithering enough to smooth out the gradients of a dithered 10-bit on a 8-bit panel or is Error Diffusion required?

[CarlosCaco]

When you do the math, 16.78 million is 8-bit, displaying 256 red, 256 green and 256 blue shades. Look up specs of your monitor, depending on how they decided to phrase it you might have to dig deeper, better yet find a detailed review. Look for 'Colour Palette' of your panel, if it says '1.06 billion' but the 'Colour Depth' says '8-bit', then it's not a true 10-bit panel (like 99% out there), but is capable of producing 1.07 billion colours. That this is done with an 8-bit colour depth and an additional Frame Rate Control (FRC) stage (Temporal Dithering).
But I doubt you will see a difference outputting 10-bit on a '8-bit + FRC' panel vs properly dithered madVR output on 8-bit panel.

On a side note, did anyone with a '8-bit + FRC' panel make some tests displaying 10-bit with madVR with dithering disabled to compare panel's dithering to that of madVR?
And, is madVR's Ordered dithering enough to smooth out the gradients of a dithered 10-bit on a 8-bit panel or is Error Diffusion required?

so i not loosing information outputing 10bit on 8bit display?

i don´t find any specs about my panel searched everywere,i don´t know if has FRC
i know that if i go d113d exclusive 10 bit with any dither algo the image is similar to 8bit with dithering disabled on madvr

[Knight77]

Try SuperRes(3) + crispen edges(1.0) for 2x scaling factors. I prefer this to SuperRes + thin edges.

Do you suggest this also for 1.5x scaling or in that case you would stick with SuperRes + thin edges?

[michkrol]

i thought that madvr converts or dither down to 8 bit, but in my osd shows: h264, 10bit, 4:2:0 -> P010, 10 bit, 4:2:0
This line informs you what format the video is and what format madvr gets from the decoder, not display depth.
In your case the file is h264 10bit and madvr gets 10bit color format from the decoder. This is good because it means madvr does the dithering to 8bit before sending it to your display. This gets you higher quality than decoder's dithering.
so i not loosing information outputing 10bit on 8bit display?
Technically you might be losing some information, but most people wouldn't notice the difference.
Another question is whether the studio used 10bit while mastering the source, which usually isn't the case.
i know that if i go d113d exclusive 10 bit with any dither algo the image is similar to 8bit with dithering disabled on madvr
It's actually the other way round: 8bit + dither is really close to 10bit without dither.
That said, you should have dither always enabled.

To sum this up: when in doubt leave the settings at defaults to get highest quality. 10bit videos on 8bit displays should look so close to 10bit display that you can't notice a difference.
Your settings are alright, so don't over-think it, just enjoy the movies ;)

[madshi]

Is it possible for madVR to adjust display backlight over DDC/DI based on profile rules for HDR/SDR media? That way, like calibration curves, backlight can be adjusted automatically as well.
I've no idea if that's possible. And if it worked I've no idea if that would screw up things like calibration and stuff. Might be worth trying at some point, although I don't know if the OS/GPU driver even gives me access to DDC/DI. But this is not going to happen any time soon.

I have exhibited random crashes in the past couple of weeks. I have updated madVR, LAV Filters, MPC-BE and graphic drivers to the latest versions but I can't narrow down the culprit. I'm hoping this log provide some kind of clarity. It is a large file so please forgive me.
Unfortunately for crashes debug logs don't really help at all in most cases. I'd need a crash report. When those crashes occur, do you get a window which looks somewhat similar to this?

http://madshi.net/exc-ss3.gif

If so, click on "show bug report", then press Ctrl+C. After that you have the crash report in your clipboard. Upload it somewhere for me to look at.

If the crash box looks different, it's probably a crash in the media player (or decoder or splitter or ...) and not in madVR.

Are there any guidelines for setting peak luminance? Does this even matter with non-HDR content? If so, this option could use better labelling.
It only matters for HDR content at this point. This is just the very first HDR version, so there are no real guidelines yet. Will need some testing and feedback from users.

madshi in his infinite wisdom make possible to run any executables on (de)activating profile ;)
Well, the option is there, but I think I've not actually implemented that feature yet? Not sure right now...

FSE, but I don't think we should spent to much time on this problem if FRC isn't the right way to go for this case and ReClock solves it
Ok.

By the way, I have my monitor calibrated to 120 nits (or 120 lux? I always get them confused), well below the minimum of 400 that madVR allows. Is 400 simply the point below which normal (non-HDR) luminance values would start getting cut off? In other words, what is the HDR luminance value of 'normal white'?
You know, I've asked this exact question to my UHD insider, and he didn't know, either. He said he was going to ask his contacts, but I'm not sure we'll get a useful answer. The HDR specs don't specify the luminance value of "diffuse white". Currently I'm assuming diffuse white to be 100 nits. But that's just a guess because that's what DVDs and Blu-Rays are usually mastered for, I think.

Oh an dteh ENhance Detail option, (sorry, just found out about it) i do like. how does it work exactly? at first glanze it look just like an sharpener to me, but its different, i cant tell what it is, but call me interessted.
It increases the contrast between darker and brighter pixels in non-edge areas, which makes fine image detail easier to see, but it may also increase noise, grain and compression artifacts.

The other sharpeners all concentrate on edges, only, and intentionally leave non-edge areas untouched. So the "enhance detail" algorithm basically works on different image regions compared to all other sharpening algorithms.
BUT, is there a way to go lower them 400nits?

Its still a bit too dark for my taste, maybe 200?
@madshi: I played a couple of HDR sample videos with the new build, at normal 100 nits peak brightness level. I too feel that I would prefer a brighter output than it is at 400 nits setting of “display peak luminance”. Can you please offer 100 nits option in the dropdown for “display peak luminance”?
So my display is currently calibrated to 100 nits, since that is the intended reference level for SDR.
I can get that up to 300-350 or so if I disable backlight scanning, trading motion clarity for brightness.
But with HDR content not being very common right now, I'd be more inclined to just compress HDR sources to 100 nits than mess with creating a new preset on the TV that I have to switch to. Do you plan on extending the range of options for this?
Ok, the next build will have options for 265, 180 and 120 nits.

After downloading the new madTestPatternSource samples, i wanted to try them, but found out that my MPCHC64bit didnt want to. cant handle the file.
I did install teh new madTestPatternSource stuff and the bat and all..

It did work with MPCHC32bit.
So i guess they only work with 32bit player?, or am i missing something?
Yes, the test patterns only work with 32bit players atm.

Something that I was wondering though: I believe that one of the videos I have is HDR, but lacking any of the metadata, as it has that very desaturated, washed-out look that HDR videos did prior to this update.
Is there any way to tag a file as HDR, or is it basically beyond repair if that metadata is missing?
Seems like it doesn't even have the proper flags for the SMPTE 2084 transfer function. I guess it might be possible to make it work ok by providing the missing information via file name tags, but at this point I don't think it's worth adding those tags, especially because you'd probably have to guess the exact values, anyway.

So having support in madVR would be a much better solution.
I don't expect it to be a high priority item, but I would like to see BFI support at some point.
This is unlikely to happen before we get displays which can do really high FPS. Maybe in 2-3 years when we get fast OLED displays, we can revisit BFI. At the moment most displays top out at 60Hz, which makes BFI impossible without major flickering.

Madishi on a 9500LC crt projector what is the nit capabilities of the projector? Have all the latest direct chain mods and Moomee card and I can see changes all the way up to 10000nit.
I've no idea. Just try which value looks best to your eyes.

I can't think of many consumer displays, CRT or not, which would exceed 400 nits brightness.
True. However, things are more complicated than simply measuring the brightness. E.g. if you're watching a movie in a "bat cave" with black walls/floor/ceiling and a good projector, your eyes will adjust to the much lower overall brightness level compared to an LCD flat panel. So the same e.g. 400 nits setting might work well for such front projection, too, due to the magic of our eyes.

Basically this means that we have to take ambient light levels and that kind of stuff into account, too.

I've come across a few scenes in some NTSC DVDs where the madVR IVTC algorithm freaks out and drops the 3:2 pattern and goes unknown before recovering when the shot changes./QUOTE]
[QUOTE=Stereodude;1750059]Except the scenes are not fades. They just cause misdetections. I will cut one of them from the DVD and upload a sample.

Edit: Here's a sample (https://www.sendspace.com/file/tp2lq7). Enable deinterlacing and watch.
Thanks for the sample. The more samples you can provide, the better. So if you stumble over this problem again, please create more samples.

Can someone tell if there is any difference at all between choosing "disable calibration controls for this display" and "this display is already calibrated?"
"disable calibration control" completely disables any gamma and gamut adjustments. If you set madVR to "this display is already calibrated", then depending on the video properties, madVR might do some adjustments or not.

Blu-Rays are usually encoded to BT.709 primaries. NTSC DVDs are usually encoded to SMPTE-C primaries. PAL DVDs are usually encoded to EBU/PAL primaries. In theory, in order to watch all these 3 different sources with correct colors, you'd have to calibrate your display 3 times with 3 different presets, one for each of those gamuts. I don't think anybody is doing that, not even professional calibrators.

Most displays these days try to get near to BT.709. Most calibrators aim to achieve BT.709, too. If your display is calibrated to give you correct colors for BT.709, then Blu-Rays will look correct in your setup, regardless of whether you set madVR to "disable calibration control" or "this display is already calibrated". However, if in this situation you play e.g. an NTSC DVD, there will be a difference. With "disable calibration control" disabled, the content will be sent to the display as is, and the colors will be slightly incorrect. When using "this display is already calibrated to BT.709", madVR will convert the NTSC colors to look correct on a BT.709 calibrated display.

So basically the recommendation is the following: If your display comes close to BT.709 (either by factory, or by manual calibration), then you should use "this display is already calibrated". If your display is not calibrated at all, then the colors you'll see are likely to be incorrect in any case. In that situation you could just as well use "disable calibration control" because there's no way to achieve correct colors, in any case. The difference between a properly calibrated display and a non-calibrated display is probably much bigger than the difference between BT.709 and SMPTE-C.

I use Windows 10 Pro x64(10586.29)+AMD Fury X.
When I update VGA driver to new driver【radeon-crimson-15.12-win10-64bit】

The NNEDI3 in madVR(chroma upscaling or image doubling)all could not work.
Maybe the madVR NNEDI3 OpenCL kernel something wrong in new VGA driver.

I always use SVP4 Pro+madVR together.
SmoothVideo Project SVP(3.17a and 4 pro)has the same error when using GPU-acceleration(OpneCL).

Any idea to fix it ?

If there is no way to fix it, I just only go back to old VGA driver.
Because old VGA driver(not crimson)all works fine.
I'm not sure why it fails. Probably a GPU driver problem. Nobody else reported this specific problem with Crimson, though. So it might be worth a try simply reinstalling the very same driver. Maybe the OpenCL driver didn't get installed properly or something. If all else fails, you may have to go back to an older driver.

0.89.18 playback is stopped (only audio work) with a cloned 2nd monitor
reverting to .17 is fine
Need more details. Does the madVR debug OSD (Ctrl+J) still work in this situation? Do the media player buttons and menus still react to your mouse in this situation? Does it make any difference if you disable/enable D3D11 presentation and/or FSE mode?

if i play a 10 bit content on my 8 bit display, i'm loosing some color information?
No. madVR uses dithering to automatically convert higher bitdepths to lower bitdepths. Dithering works very well, the only disadvantage is a raised noise floor.

Im a bit confused by the terms 8-bit, 10-bit etc. in relation to the number of colours. my monitor has 16.7 million colours, but I dont know if that is 8-bit or 10-bit.
8bit = 2^8 = 256 steps per channel.
10bit = 2^10 = 1024 steps per channel.

If each RGB channel (R, G and B) has 8bit, that is 256 * 256 * 256 = 16.777216 million colors.

[madshi]

I have posted in the LAV thread (http://forum.doom9.org/showthread.php?p=1746830#post1746830) an issue.

With the new madVR0.89.17 play the DVD Aliens(test dvd without the big-vobs (http://forum.videohelp.com/attachments/34047-1444852648/Aliens.7z)) with 4:3 format (menu and video), but it should be 16:9. With EVR it plays fine with 16:9 format.
It seems to play fine as 16:9 on my PC. But I always get to the "Deutschland / Österreich" copyright screen, not any further, due to missing VOBs. Which media player are you using? And which aspect ratio settings in the media player?

When using the current madVR on a 4K UHD display with high DPI say 300%, the madVR CTRL J renderer stats are too small to read.
OSD font size will scale with the rendering window size in the next build.

I think it's more like the madVR isn't requesting the frames from the capture source. Here is the screenshot of Ctrl+J in Potplayer using the capture source: http://i.imgur.com/8KwExdw.jpg and here is the screenshot of Ctrl+J in Potplayer using a captured AVI from the same capture source: http://i.imgur.com/7anTyFl.jpg.
According to the screenshots the decoder isn't delivering the frames to madVR fast enough. madVR is not "requesting" frames. That's not how it works. The decoder should be sending frames as fast as it can decode them. The decoder has the active role here. It's madVR which has the passive role. And if madVR doesn't receive decoded frames fast enough, obviously all queues stay empty.

Have you tried different decoders?

Hi
I think I have found a bug. I am using the last development version of MPC-HC with the last version of XYSubFilter and the last version of MadVR
The bug happens in MadVR plus XYSubFilter when displaying upscaled VobSub subtitles.
It is an anime episode, 848x480 internal resolution, MPEG4 (H264) video, softcoded VobSub subtitles.
When displaying the vid at 1x resolution the subtitles are correctly displayed at the bottom of the screen.
When the vid is at 50% resolution (in window mode) it doesn't show subtitles at all.
If the vid is upscaled, resolution 2x (window mode) or it is upscaled fullscreen at 1920x1080 the subtitles appear on the left and in the middle of the screen and the subtitles themselves are not upscaled (they appear with the same size than in the resolution 848x480).
MadVR with other subtitles filters (VSFilter and Internal MPCHC filter) display correctly the subtitles upscaled and the bottom of the screen.
The MPCHC Personalized Enhanced Video Renderer with XYSubFilter shows the subtitles correctly too.
You can see screenshots of the vid here with the subtitles. In all these screenshots I am using MPCHC, MadVR and XYSubFilter.
If you need more info, please, tell me.
http://www72.zippyshare.com/v/8bgIjJfZ/file.html
Hmmmm... I suppose you're using DXVA upscaling, is that correct? Can you confirm that this problem goes away if you use e.g. Catmull-Rom upscaling? If the problem still occurs with Catmull-Rom upscaling, can you create a small sample for me, please, so that I can reproduce the problem on my PC? If the problem is fixed by using Catmull-Rom, then I don't need a sample.

- movie file here (http://wikisend.com/download/540628/qs-adaptive.mkv) (35Mb)
- settings files here (http://wikisend.com/download/262952/settings.zip); when everything is fine, jinc is used for resizing; when it's not fine, DXVA is used for resizing
- screenshots here, for how it looks when DXVA resizing is used, in windowed and fullscreen modes, because of deinterlacing performed by madVR
Thanks, I can reproduce the problem. Will try to fix it for the next build.

- VIDEO_TS folder here (http://wikisend.com/download/573694/CaroDiario.7z) (67Mb)
This seems to work just fine on my PC. The intro is shown as 4:3, the menu switches to 16:9. Which media player are you using? And which aspect ratio settings?

[6233638]

This is unlikely to happen before we get displays which can do really high FPS. Maybe in 2-3 years when we get fast OLED displays, we can revisit BFI. At the moment most displays top out at 60Hz, which makes BFI impossible without major flickering.There are a lot of monitors out there which support 72Hz inputs. Though it would flicker, most Plasma TVs ran 24p inputs at 48Hz, which would be 72Hz with 2 video frames for every 1 black frame. The question is not "will it flicker?" but rather "do I prefer motion clarity over flicker?" Growing up with 50Hz CRTs, I don't mind flicker one bit.

And higher refresh rates are becoming increasingly common. There's an ever-increasing number of TVs which support 120Hz inputs now, and I don't know anyone that's bought a new monitor in the last few years which wasn't 120-144Hz - and the latest models now offer 165-200Hz. (and that 200Hz monitor is a 21:9 "ultrawide", which seems to be gaining popularity)

It's less common with higher-end home theater equipment like projectors - though there are some 120Hz DLPs, but there should be 4K120 displays on sale next year. I just hope that LG or Samsung OLEDs are a part of that.

[madshi]

madVR is more meant for big TVs and projectors. I'm not going to invest time into BFI if it mostly only works on computer monitors. IMHO we need at least 96Hz before madVR BFI can start making sense, ideally 120Hz. The number of users with such TVs is probably very low atm. Furthermore there's no way for me to test this properly without having such a high-refresh-rate TV or monitor myself, which is currently not the case.

[e-t172]

I've no idea if that's possible. And if it worked I've no idea if that would screw up things like calibration and stuff. Might be worth trying at some point, although I don't know if the OS/GPU driver even gives me access to DDC/DI. But this is not going to happen any time soon.

Do typical consumer HDTVs and projectors actually support DDC/CI, anyway? Even computer monitors rarely support it, except maybe professional ones.

[chros]

But I doubt you will see a difference outputting 10-bit on a '8-bit + FRC' panel vs properly dithered madVR output on 8-bit panel.

On a side note, did anyone with a '8-bit + FRC' panel make some tests displaying 10-bit with madVR with dithering disabled to compare panel's dithering to that of madVR?
And, is madVR's Ordered dithering enough to smooth out the gradients of a dithered 10-bit on a 8-bit panel or is Error Diffusion required?
I agree with you. As it turned out I have one, 8bit+FRC, if it's true this is the one http://www.panelook.com/LC550WUD-SCA1_LG%20Display_55_LCM_overview_11833.html and it passes the 10bit gradient test ( (I had to enable FullRGBHDMI hack for the Intel driver, thanks madshi for this!)).
I'll take a good look and try to see the difference, but I didn't see at the first time. :)

an relative old laptop with an disgusting optmus system,
after milions of tests i can assume that my display output 8 bit,
Honestly, I'd be surprised if it was 8 bit at all :) I vote for 6bi+FRC at most!!! :)
What's your laptop? And what string of the display can you see in madvr under Devices? (I need the full string)
Mine is: LG LP156WF1-TLF3, and panellook reports it: http://www.panelook.cn/LP156WF1-TLF3_LG%20Display_15.6_LCM_overview_12274.html

[Stereodude]

I thought for broadcast, it be pure interlace anyway, not film video.
Lots of broadcast 1080i60 HTDV in the US is film based material that has gone through a telecine process. Things like the local news, or talk shows, no. But the primetime scripted shows are usually IVTC-able to 1080p24.

When you said looks terrible, what you mean. Lack of post processing or jittery movement?
Well, smooth motion was driving up the render times to higher than 1/23.976... Turning it off helped solved that issue, but I still can't quite put my finger on what it's doing. I think motion looks slightly different, but it might just be in my head. I guess I'd need to remove the soft pulldown and compare the playback of the file. My HTPC has an AMD R9 380 and a old Intel E5300 CPU. So no CUVID or Quicksync HW decoding options. Both the Radeon HW (copyback) and SW decoding leave a 3:2 pattern for madVR to find, which I don't understand if it's pure progressive output. However, I will agree that stepping through the video in slow motion doesn't show any interlaced or duplicated frames.

When I play it on another PC and use Quicksync as the HW decoder madVR only finds a 2:2 pattern which AFAIK means it is pure progressive. HW decoding with CUVID also leaves a 3:2 pattern for madVR to find, like SW and the Radeon HW unit. :confused:

[CarlosCaco]

"disable calibration control" completely disables any gamma and gamut adjustments. If you set madVR to "this display is already calibrated", then depending on the video properties, madVR might do some adjustments or not.

Blu-Rays are usually encoded to BT.709 primaries. NTSC DVDs are usually encoded to SMPTE-C primaries. PAL DVDs are usually encoded to EBU/PAL primaries. In theory, in order to watch all these 3 different sources with correct colors, you'd have to calibrate your display 3 times with 3 different presets, one for each of those gamuts. I don't think anybody is doing that, not even professional calibrators.

Most displays these days try to get near to BT.709. Most calibrators aim to achieve BT.709, too. If your display is calibrated to give you correct colors for BT.709, then Blu-Rays will look correct in your setup, regardless of whether you set madVR to "disable calibration control" or "this display is already calibrated". However, if in this situation you play e.g. an NTSC DVD, there will be a difference. With "disable calibration control" disabled, the content will be sent to the display as is, and the colors will be slightly incorrect. When using "this display is already calibrated to BT.709", madVR will convert the NTSC colors to look correct on a BT.709 calibrated display.

So basically the recommendation is the following: If your display comes close to BT.709 (either by factory, or by manual calibration), then you should use "this display is already calibrated". If your display is not calibrated at all, then the colors you'll see are likely to be incorrect in any case. In that situation you could just as well use "disable calibration control" because there's no way to achieve correct colors, in any case. The difference between a properly calibrated display and a non-calibrated display is probably much bigger than the difference between BT.709 and SMPTE-C.

Testing the HDR new feature i can see cleary the diference, if i check disable calibration controls, madVR not do the conversion to the correct colour space of HDR content, the result are bad compared to this display already calibrated to bt 709

Do my calibration settings in dispcalgui i thought the correct was use the already calibrated to bt 709 with pure power curve transfer function

but testing the hdr content i only achieve goods resuls if i set the bt 709/601 curve

could someone explain the difference between the the two transfer functions?

There was a shader in sweetfx that simulated the appearence of HDR, was terrible because even in the minimal seetings the black were so darker e the withe so brighter that vanishes a lot of elements in the darkker/brighter areas

tweaking madVR i achieve the same results setting the transfer function calibration to bt 709/601 and setting gamma processing to pure power curves(terrible results, i was only messing the gama to see what happens), so i wondering what the difference of the transfer functions and in what cases should we use then?
iits possible to a tv or monitor were calibrated to bt 709 with pure power curve? what should correct?
in dispicalgui there´s option to use bt 709 tone curve or gama 2:20, what should the correct transfer function to work properly on madVR color conversions?

[CarlosCaco]

Honestly, I'd be surprised if it was 8 bit at all :) I vote for 6bi+FRC at most!!! :)
What's your laptop? And what string of the display can you see in madvr under Devices? (I need the full string)
Mine is: LG LP156WF1-TLF3, and panellook reports it: http://www.panelook.cn/LP156WF1-TLF3_LG%20Display_15.6_LCM_overview_12274.html

Hey man, my laptop is a Dell Xps 15 l502x, and the display is a AU Optronics KYYVK€B156HW1 full hd
I serchead so many time for some information/specs about that, but didn not find anything relevant (i had hard times to do a correct calibration profile for use with madVR, i fid a lot of people in foruns complaining about the colours of this display, some people sayng that is wide gamut, so i searched a lot, but with no results, the colors are a bit saturated but i do know nothing about, found that are different versions of this panel that came with lenovo, acer and hp laptops, but in my case there is no documetation about, so i don´t if is v1, v2, v3,v4,v5

http://www.panelook.cn/rfqs_cn.php?panel_part_number=AU+Optronics+KYYVK%E2%82%ACB156HW1


tried to search AU Optronics KYYVK-B156HW1 too, because dispcalgui reports that way
just not informations at all, if you help me to findi will be very glad

edited

searched again and found but that are 8 versions on the site, thanks a lot! but i still don´t know what version is my...

http://www.panelook.cn/modelsearch.php?panel_part_number=B156HW01

[huhn]

Testing the HDR new feature i can see cleary the diference, if i check disable calibration controls, madVR not do the conversion to the correct colour space of HDR content, the result are bad compared to this display already calibrated to bt 709

Do my calibration settings in dispcalgui i thought the correct was use the already calibrated to bt 709 with pure power curve transfer function

but testing the hdr content i only achieve goods resuls if i set the bt 709/601 curve

could someone explain the difference between the the two transfer functions?

There was a shader in sweetfx that simulated the appearence of HDR, was terrible because even in the minimal seetings the black were so darker e the withe so brighter that vanishes a lot of elements in the darkker/brighter areas

tweaking madVR i achieve the same results setting the transfer function calibration to bt 709/601 and setting gamma processing to pure power curves(terrible results, i was only messing the gama to see what happens), so i wondering what the difference of the transfer functions and in what cases should we use then?
iits possible to a tv or monitor were calibrated to bt 709 with pure power curve? what should correct?
in dispicalgui there´s option to use bt 709 tone curve or gama 2:20, what should the correct transfer function to work properly on madVR color conversions?

bt 709 has a pure gamma of 2.22 so bt 709/ bt 601 should be the same but the result is more like sRGB which is BT 709 with a different gamma function. this doesn't have an effect without "color or gamma -> enable gamma processing" at least for bt 709 content.

if you use this option for HDR than the results are different.
it's like madVR wants to take these settings and change it back to normal gamma 2.2.

that's why changing to BT 709/bt 601 curve is crushing black or at least rising the gamma and a higher gamma is lowering the gamma.
but "color & gamma -> enable gamma processing" does nothing with HDR content.

"correct" would be a pure gamma curve of 2.2.
if you are using DISPCAL GUI just think about an 3D LUT.

[Aktan]

According to the screenshots the decoder isn't delivering the frames to madVR fast enough. madVR is not "requesting" frames. That's not how it works. The decoder should be sending frames as fast as it can decode them. The decoder has the active role here. It's madVR which has the passive role. And if madVR doesn't receive decoded frames fast enough, obviously all queues stay empty.

Have you tried different decoders?

This is raw YUY2 from an analog capture source (device). I don't think it can deliver frames faster than realtime as it captures in realtime. I thought a setting like "delay playback start until render queue is full" would fix the problem, but there is no delay. Did you look at the 2nd screenshot that shows decoding an AVI works fine? That has decoder delay since it is losslessly compressed, but it still works fine. Another note is if I pipe the raw YUY2 from FFMPEG encapsulated in an ASF container to MPC-HC, this problem does not happen. But then there is an obvious input delay which wouldn't work for playing games.

Edit: Now thinking about it, I can see why "delay playback start until render queue is full" would not fix the problem since it is a decoder queue problem. Would it be possible to add the option of something like "delay playback start under decoder queue is full"?

[seiyafan]

I have a computer which I use for amature photo editing at home and it's calibrated with i1 Display pro and basICColor (it generates an icm file which is loaded by Windows at startup) Just today I found out about dispcalGUI and Argyll CMS. Since I also watch movies on this computer, what's the advantage of dispcalGUI over my existing calibration besides the 3D LUT? Is the current movie playback with MadVR using the .icm of the existing calibration?

[Aktan]

When I play it on another PC and use Quicksync as the HW decoder madVR only finds a 2:2 pattern which AFAIK means it is pure progressive. HW decoding with CUVID also leaves a 3:2 pattern for madVR to find, like SW and the Radeon HW unit. :confused:

I'm surprise it finds a 2:2 pattern at all as I thought that is usually used on PAL DVDs. But it is interesting that 1/3 of the time when I force FILM deinterlacing from madVR, I do get a 2:2 pattern as said by the OSD.

[retrue]

Quote:
Originally Posted by retrue View Post
Hi
I think I have found a bug. I am using the last development version of MPC-HC with the last version of XYSubFilter and the last version of MadVR
The bug happens in MadVR plus XYSubFilter when displaying upscaled VobSub subtitles.
It is an anime episode, 848x480 internal resolution, MPEG4 (H264) video, softcoded VobSub subtitles.
When displaying the vid at 1x resolution the subtitles are correctly displayed at the bottom of the screen.
When the vid is at 50% resolution (in window mode) it doesn't show subtitles at all.
If the vid is upscaled, resolution 2x (window mode) or it is upscaled fullscreen at 1920x1080 the subtitles appear on the left and in the middle of the screen and the subtitles themselves are not upscaled (they appear with the same size than in the resolution 848x480).
MadVR with other subtitles filters (VSFilter and Internal MPCHC filter) display correctly the subtitles upscaled and the bottom of the screen.
The MPCHC Personalized Enhanced Video Renderer with XYSubFilter shows the subtitles correctly too.
You can see screenshots of the vid here with the subtitles. In all these screenshots I am using MPCHC, MadVR and XYSubFilter.
If you need more info, please, tell me.

Hmmmm... I suppose you're using DXVA upscaling, is that correct? Can you confirm that this problem goes away if you use e.g. Catmull-Rom upscaling? If the problem still occurs with Catmull-Rom upscaling, can you create a small sample for me, please, so that I can reproduce the problem on my PC? If the problem is fixed by using Catmull-Rom, then I don't need a sample.

At Madshi

Yes, I am using DXVA upscaling, but when I use any other upscaling with this video (using last versions of MPC-HC, XYSubFilter and MadVR) I don't get any image at all. The sound plays fine but the screen is black. With some of the upscaling methods I can see the info when typing Ctr+J, but in most cases I see nothing at all. All the upscaling methods work fine with other videos that I have tried.

How I can create the sample?
I took a sample with Avidemux but it ignored the subtitles. Then I extracte d the subtitles from the original video and I added them to the sample. But something is wrong. Selecting DXVA upscaling it was possible to reproduce the problem: subtitles not upscaling and appearing in a wrong position, but when I tried Catmul-Rom on the sample the computer crashed.

[huhn]

try mkvtoolnix.

the splitting settings are under output.

[CarlosCaco]

bt 709 has a pure gamma of 2.22 so bt 709/ bt 601 should be the same but the result is more like sRGB which is BT 709 with a different gamma function. this doesn't have an effect without "color or gamma -> enable gamma processing" at least for bt 709 content.

if you use this option for HDR than the results are different.
it's like madVR wants to take these settings and change it back to normal gamma 2.2.

that's why changing to BT 709/bt 601 curve is crushing black or at least rising the gamma and a higher gamma is lowering the gamma.
but "color & gamma -> enable gamma processing" does nothing with HDR content.

"correct" would be a pure gamma curve of 2.2.
if you are using DISPCAL GUI just think about an 3D LUT.


the gamma processing wast a test i was doing with nonHDR content only to see what happens, i don´t use gamma processing at all



i tried many times make a 3dlut but not achieve any descent result, all are too saturated, darker with banding, tried many configurations but nothing

here i take some screen shots to compare the gamma calibration in HDR,

[KoD]

- VIDEO_TS folder here (http://wikisend.com/download/573694/CaroDiario.7z) (67Mb)
This seems to work just fine on my PC. The intro is shown as 4:3, the menu switches to 16:9. Which media player are you using? And which aspect ratio settings?

I'm using MPC-HC 64bit nightly 1.7.10.28 from Dec 1st, and I've exported its settings to a file (http://wikisend.com/download/128016/mpc-hc-settings.7z) to make it easy to test.

in the Video Frame menu:
- Touch window from inside = selected
- Keep Aspect Ratio = enabled
- Override Aspect Ratio = Default

What I describe below is when using software decoding, and with madVR performing DXVA deinterlacing:
- the copyright warning shows, and in the OSD it says: deinterlacing off [says upstream]
- then the studio logo display starts, but for some reason the player moves from 00:00 to 00:01 to 00:02 to 00:03 seconds slower than real-time (it takes more than 3 seconds); while this happens, the OSD still says: deinterlacing off [says upstream]
- then the playback of the studio logo resumes to real-time, and the OSD says: deinterlacing on [says upstream]
- while the studio logo is displayed there is a moment close to its end when it switches to 16:9, stays like this for half a second, then switches back to 4:3 before the logo clip ends
- then the menu starts, and the display ratio is still 4:3; OSD still says deinterlacing on [says upstream]

The effect when using hardware decoding and deinterlacing performed by QuickSync is the same, just that the madVR OSD always says "deinterlacing off [says upstream]".

The settings used for madVR are the same as for the other movie clip. You can disable smooth motion, and it still happens. The TV refresh rate is 60Hz (but it doesn't matter). The same thing happens on older MPC-HC versions too, but did not try to see how far back.

[Stereodude]

I'm surprise it finds a 2:2 pattern at all as I thought that is usually used on PAL DVDs. But it is interesting that 1/3 of the time when I force FILM deinterlacing from madVR, I do get a 2:2 pattern as said by the OSD.
So I tested the clip with the soft pulldown removed from the elementary stream. With that file and any of the decoders madVR only sees a 2:2 pattern, never a 3:2 pattern, so I think it's fair to say there is some difference in the output of the various decoders (except for Quicksync) depending on if a soft pulldown flag is present or not.

[huhn]

So I tested the clip with the soft pulldown removed from the elementary stream. With that file and any of the decoders madVR only sees a 2:2 pattern, never a 3:2 pattern, so I think it's fair to say there is some difference in the output of the various decoders (except for Quicksync) depending on if a soft pulldown flag is present or not.

yes but only the IVTC mode from madVR sees the 3:2 pattern.

without deinterlacing a 23p stream is rendered.
with film mode deinterlacing a 23p stream is rendered with some combing frames at the end. (the core issue)
with video deinterlacing each frame is repeated resulting in "47"p.

i'm still wondering why the soft telecine flag isn't ignored all the time and the stream flag as 23p. i mean what should go wrong on a PC?

[retrue]

At Madshi

Yes, I am using DXVA upscaling, but when I use any other upscaling with this video (using last versions of MPC-HC, XYSubFilter and MadVR) I don't get any image at all. The sound plays fine but the screen is black. With some of the upscaling methods I can see the info when typing Ctr+J, but in most cases I see nothing at all. All the upscaling methods work fine with other videos that I have tried.

How I can create the sample?
I took a sample with Avidemux but it ignored the subtitles. Then I extracte d the subtitles from the original video and I added them to the sample. But something is wrong. Selecting DXVA upscaling it was possible to reproduce the problem: subtitles not upscaling and appearing in a wrong position, but when I tried Catmul-Rom on the sample the computer crashed.

Thanks to huhn I was able to create a sample.
There you go: http://www14.zippyshare.com/v/xX8OZRey/file.html

[Stereodude]

yes but only the IVTC mode from madVR sees the 3:2 pattern.

without deinterlacing a 23p stream is rendered.
with film mode deinterlacing a 23p stream is rendered with some combing frames at the end. (the core issue)
with video deinterlacing each frame is repeated resulting in "47"p.
If the output from the decoder is truly progressive, why does madVR detect a 3:2 pattern? When the soft pulldown is removed and the decoder spits out progressive frames madVR can't detect a 3:2 pattern in the output. Only Quicksync HW decoding has the same downstream behavior for both permutations.

i'm still wondering why the soft telecine flag isn't ignored all the time and the stream flag as 23p. i mean what should go wrong on a PC?
I think this one is pretty easy to answer. Ignoring the flag only works if the stream is 100% soft pulldown. The decoder can't know this. If it's ignoring the pulldown flag, what is the decoder supposed to do when it gets a hard pulldown segment?

[huhn]

If the output from the decoder is truly progressive, why does madVR detect a 3:2 pattern?

When the soft pulldown is removed and the decoder spits out progressive frames madVR can't detect a 3:2 pattern in the output. Only Quicksync HW decoding has the same downstream behavior for both permutations.

that's the part i don't understand what so ever. i have to do really really wild guess to make sense of this.

with software decoding madVR IVTC detects a 3:2 pattern. but with disabled deinterlacing it is displaying 23p and only progressive frames.

but why? how could a stream be both at the same?

I think this one is pretty easy to answer. Ignoring the flag only works if the stream is 100% soft pulldown. The decoder can't know this. If it's ignoring the pulldown flag, what is the decoder supposed to do when it gets a hard pulldown segment?

a soft telecine stream is clearly flagged as one the only issue is the wrong framerate.

it has a progressive flag, a 23p flag and a 3:2 scan flag.

[Stereodude]

a soft telecine stream is clearly flagged as one the only issue is the wrong framerate.

it has a progressive flag, a 23p flag and a 3:2 scan flag.
Yes, but the DVD / .VOB doesn't have any such limitation. Two successive chapters could alternate from true interlaced content with a hard telecine encoded into it to p24 w/ soft pulldown and back. If the decoder ignores the soft pulldown and decides to output p24 content, what is it supposed to do when it gets to the next chapter that's true interlaced content?

[Aktan]

So I tested the clip with the soft pulldown removed from the elementary stream. With that file and any of the decoders madVR only sees a 2:2 pattern, never a 3:2 pattern, so I think it's fair to say there is some difference in the output of the various decoders (except for Quicksync) depending on if a soft pulldown flag is present or not.

So I must have brain farted when I said I didn't get why madVR would only detect a 2:2 pattern. If it was progressive, that is all madVR should detect. Retesting the clip over and over, madVR does keep detecting a 3:2 pattern which I have no idea why, but interesting enough, if I add ffdshow in between the MPEG2 decoder and madVR, madVR only detects a 2:2 pattern. This doesn't fix the problem as madVR still messes up at the same spots. Only thing I can think of is this: the 3:2 pattern detection is a display mistake when madVR really is doing a 2:2 pattern, but since a 2:2 pattern is really 2:2:2:2:2:2:2:2:2:2:2:3 pattern, maybe madVR assumes that and messes up every 12 frames?