madVR - high quality video renderer (GPU assisted)

[tyrindor]

Originally Posted by LigH

@tyrindor: Why do you believe that madVR is the reason? It appears to me that the most it does is reporting issues, and rather indirectly (as a video renderer, it does not process audio at all).

I don't really know what the reason is, I was hoping someone here would know. I am running a typical setup and bitstreaming lossless codecs, so I figured someone else might have experienced this. I have no idea if it's related to the frame repeat every few minutes or not.

[madshi]

Sorry for the long absence, everyone. Have been busy with commercial work, still am, but hope to get back to madVR development soon(ish).

Originally Posted by jeanlee20xx

I hope it can add 2d to SBS function，thank you!!!!

What do you mean exactly?

Originally Posted by Snap

Anyone know of a player that can capture screenshots with the HDR data preserved? From some searching it seems there's JPEG-HDR and RGBE PNG for formats but not sure what the limitations might be.

I'm not aware of any player which supports that. Currently also the madVR screenshot interfaces don't really support transporting all the necessary information to the media player. I guess I could add that if there were a media player developer interested in adding this kind of functionality. However, creating such screenshots would only be one step, you'd also need software to actually render such screenshots properly. Is there any software that does? It would probably have to do tone mapping, gamut mapping and all that kind of stuff...

Originally Posted by runner

I would like to know how to use the control function for the videoprojector, in particular I would like at the end of the playback to return to position 1 of the lens. thank you

That's not too easy because "end of playback" is difficult to detect for madVR, and changing the lens position takes a LONG time. E.g. let's assume you have a playlist with 2 files in it. At the end of the first file madVR might think that playback has ended. But you wouldn't want the lens to return to position 1 now, would you? So basically madVR would have to add extra stuff, e.g. wait for 20 seconds and only then restore the original lens position.

Currently madVR only supports setting the lens position to a specific slot, but it doesn't support restoring the original lens position (or even activate a specific lens position). If I were to implement this kind of functionality, I wouldn't want to offer a "after playback activate lens memory 1", instead I would offer "after playback restore original lens memory", which makes more sense to me.

But since you're the first user to ask for this, I'm not sure how high the demand for this really is. Usually, I require at least 3 users to ask for a specific feature before I put it on my to do list (unless I really want the feature myself).

Originally Posted by Tomah

1. I would know why madVR, when i use NGU 4x, after oversize, downscale image two times in luma y with some source resolutions?

This could be a bug. Unfortunately you cropped some important lines from the OSD away, so the information I got is rather minimal. I'd need to know how to reproduce this issue on my PC exactly. So which display resolution, do you have? Which movie resolution? Which exact scaling settings? Etc etc...

Originally Posted by Tomah

2. Always with NGU 4x, with Low or Medium quality, chromaupsampling seems forced to Bicubic60 during the doubling/quadrupling process. Can i use "chroma settings" for entire chroma upsampling ?

NGU Low or Medium settings practically get Bicubic60 chroma upscaling "for free", that's why it's forced in this situation, to get the best possible speed. I understand if you care more about quality than speed. But in that case you should really spend your GPU power on using NGU High quality instead of increasing the chroma upscaling quality, because luma quality is *much* more important.

If your GPU is too slow for NGU High quality, then you need to save any GPU power you possibly can, and getting Bicubic60 "for free" should make you happy. If your GPU is fast enough for NGU High quality, then use that and don't worry about chroma quality too much.

Originally Posted by el Filou

Hi, I have a question about HDR: MediaInfo for Dunkirk UHD Blu-ray says:
However, madVR's OSD is regularly showing much higher values than 323 cd/m2, and even the credits at the end with pale white text on black show 170-200 cd/m2.
Is the OSD 'correcting' the values somehow, or is MediaInfo reporting wrong values?

maxCLL shows the highest RGB component, not really the true luminance of the brightest pixel, so it can differ from what madVR measures. However, as far as I understand, although maxCLL can be higher than the peak luminance, it shouldn't be lower. So if maxCLL says 323nits and madVR measures more, then most likely maxCLL was set incorrectly when the movie was authored/encoded. Of course it's also possible that MediaInfo has a bug, but I'd rather guess that the movie is badly authored/encoded.

Originally Posted by leeperry

I'd love to see if any possible please would be "activate only if it doesn't drop frames" for luma quadrupling because I'm never really sure whether I can afford it and those "set and forget" options are wonderful.

It's on my to do list to look into things like that, but it's not too easy because performance measurements aren't always fully reliable. So it will take some time to implement a proper solution for that.

Originally Posted by d3rd3vil

Can you work with this for implementing DV in madvr maybe? ^^

https://www.dolby.com/us/en/technolo...ormat-v2.0.pdf
https://www.dolby.com/us/en/technolo...ormat-v1.1.pdf
https://www.dolby.com/us/en/technolo...ormat-v1.1.pdf
https://www.dolby.com/us/en/technolo...ormat-v1.1.pdf

These are helpful (I already had them, though), but the real problem is not *my* interest in DV. The real problem is that I can't do anything, as long as no splitter/decoder transports the DV information to me. So you're barking up the wrong tree here. You don't need to convince *me*. You need to convince a splitter/decoder dev. If you find one who's interested in implementing DV, I might look into it, too. But without an interested splitter/decoder dev, there's literally nothing I can do.

Originally Posted by Apache553

madVR lags too much on my HTPC.It consumes 75% of CPU but only 2% of GPU.I tried to reinstall the video card driver and the OS,but that didn't work.Is there anyone who knows how to solve this problem?Thanks!

My HTPC:
CPU: AMD Phenom II x4 965
Video card: R7 360 2gb
RAM: 8GB DDR3
MB: GIGABYTE GA-970A-DS3P

It seems that "uploading" 10bit+ textures to your GPU causes the CPU consumption. This transfer is done through the PCIe bus. This could be a bad setting in your BIOS, or a bad BIOS firmware, or a bug in the GPU BIOS, or a problem with the OS installation of the GPU driver. It's really hard to pinpoint. I had one user in the past who had a similar problem and he updated some firmware (not sure if it was the mainboard BIOS or the GPU BIOS) which fixed the problem for him. I'm not sure if it's the same for you.

You can probably work around the issue by forcing LAV Video Decoder to only send 8bit video frames to madVR (by unchecking all 10bit+ formats in the LAV configuration), but obviously that's not a good thing to do for image quality.

Originally Posted by takenori

hi, i want to ask, for a samsung monitor (32chg70) madvr reports the bitdepth of the monitor is 8bit, and the windows 10 reports it a 10bit.

Where does Windows 10 report the display bitdepth? It doesn't to do that for me.

Originally Posted by neno

Hi, madshi. In the latest 0.92.12 whenever I try to take a screenshot when paused, the potplayer crashed. And taking a screenshot when playing will get me an abnormal image(strange colors like a thermal image of the original one).

I haven't changed any settings and it can just take screenshots without any problem in older 0.92.X versions.

I use WIN10 with RX480.

Maybe the problem in changelog of 0.92.11 "screenshots in paused state sometimes crashed madVR" is still not fixed? Just a guess.

Anyone have some ideas or the same problem?

----------

And another problem.

I use lav 0.71 with latest madVR in D3D11 native mode, and the picture is always(no matter what video) split into the top and bottom part. The top part is the picture content turned into a flat size, and the bottom part is often a green image.

While D3D11 CB can just play normally.

Is there any idea about the problem?

nevcairiel said it sounds like a driver problem and madVR handles the processing of "native" frames. So I report it here.

Have you tried reinstalling or updating the GPU driver?

Does the screenshot problem go away if you use "copyback" decoding or software decoding?

Originally Posted by Jong

I have an LG E6 and an HTPC with an Nvidia GTX1050 GPU (390.65 driver). In Nvidia control panel, if I select 2160p YCbCr 4:2:2 and tell MadVR I have a 10-bit capable display all works well for HDR video.

I noticed by accident that if I change Nvidia video mode to RGB 8-bit then when I play a 60Hz HDR test clip (eg. LG's "Chess") I was getting "sparkles", showing HDMI errors.

I noticed that MadVR was outputting 10-bit depth even when using RGB and 4K/60 RGB 10-bit is not allowed by HDMI 2.0.

Are we talking about HDR passthrough, as in switching the TV into HDR mode? For Nvidia GPUs I use a custom API to do that, which bypasses some of the OS and GPU driver safety checks. So that might explain why the GPU tries to output something which is impossible to output. I guess I could add a safety check to prevent madVR from trying to output 4Kp60 10bit. Will add that to my do to list...

[madshi]

Originally Posted by steakhutzeee

I'm reading a lot and hardly trying to figure out how I know when I simply need to enable deinterlacing and when I need IVTC.

I never seen IVTC turned on automatically, don't know why.

But in my dvd profile, how I know if I only need deinterlacing or IVTC?

Ok, so here comes a full blown explanation, hopefully it will put the topic to rest:

With default settings ("disable automatic source type detection" unchecked), madVR always uses DXVA deinterlacing. The exact DXVA deinterlacing algorithm is provided by the GPU hardware/driver. It's supposed to automatically detect the source type and internally switch between whatever deinterlacing technique is needed for best results, even per pixel. Since the algorithm is provided by the GPU, not by madVR, the quality will depend on the GPU manufacturer, GPU model, maybe even OS and driver version.

There's one thing DXVA deinterlacing cannot technically do, which is decimation. DXVA deinterlacing *can* (at least in theory) detect telecined movies and deinterlace them correctly, e.g. from 60i to 30p. However, from those 30p only 24p are unique, the other 6p are duplicates. Removing those is called "decimation".

Because I wasn't happy with the lack of decimation support, and with the general state of the DXVA film mode deinterlacing, I added a "forced film mode" to madVR. If you want to use it, you have to enable it manually. Of course you should do that only for true movie content. Otherwise you'll get nasty artifacts.

Practically, most PAL movie DVDs don't need any deinterlacing at all. But there are exceptions. Using forced film mode should work for both those PAL movie DVDs which don't need deinterlacing, and also for the exceptions.

NTSC movie DVDs and ATSC movie broadcasts almost always need deinterlacing. For those forced film mode should work great and shoudl be able to restore the original 24p.

Forced film mode doesn't work for natively interlaced content (e.g. sports, concerts etc), and also for field blended stuff. Sometimes you can't know if forced film mode will work or not. In that case you'll simply have to try. If it doesn't work, you'll probably see that immediately.

Originally Posted by Schwartz

Got a madVR / Driver Stopped Responding crash here, latest version and latest 18.2.2 AMD drivers on a R9 270X. Windows 8.1.
The crash occured when I was seeking a passthrough Twitch stream in MPC-HC, while a second instance of MPC-HC had a movie paused.

https://pastebin.com/ZvprjHaD

This is a crash in the AMD GPU driver, in the moment when madVR tries to release the D3D11 "compute" device. Currently the compute device is only used for error diffusion and for the "measure frame luminance" HDR feature. Did you happen to use either of those?

Generally, if you only got this problem once, it's probably safe to ignore it. If you get it often (or once in a while), try disabling the mentioned features. If that still doesn't help, try also disabling the "use D3D11 for presentation" option in madVR.

Also, try updating the GPU driver.

Originally Posted by sauma144

What about image refinements or smoothmotion?
Are there some possible improvements to do?

What about a reclock alternative since custom resolutions are still problematic with old and current hardware?

I've on my to do list to look into AdaptiveSharpen some more.

Reclock is mainly an audio renderer, which is not something I'm interested in developing.

Originally Posted by cork_OS

Well, since I began of dreaming I would also ask madshi what he think about upscale detection and about upscale reversion possibility.

I'm not sure if upscales can be easily detected and reverted. I mean if I *knew* a certain video was upscaled with a specific algorithm from resolution X to resolution Y, I could probably find an algorithm to fix that. But I would have to know all the details. And I don't think I can get them from just looking at the pixels.

Originally Posted by neno

Hi, madshi. About the problem, I have tried various settings in the screenshots tab.

I find that if I draw subtitiles(uncheck "don't draw" by default), then it will crash whatever other settings are(even if there aren't any subtitles). Can this be helpful to you? Anyone have the same situation?

Interesting. Which subtitle renderer are you using? Does switching to a different renderer help?

Originally Posted by neno

And a question: Do I have to re-optimize the custom timing everytime when I update the graphic card driver? (RX480 in my situation)

As long as the custom timings survive the GPU driver update, probably not.

Originally Posted by neno

PS: A tiny UI visual suggestion. The "color & gamma" tab in "devices" settings: the beginning position of the first letter "c" is some space backward compared to the other tabs' title.

True. Thanks!

Originally Posted by Oguignant

Hi. someone with experience to create a 3dlut for HDR content with Calman? Is there any guide or manual to consult?

Do you want that 3dlut to convert the HDR content to SDR for you? Or do you just want to calibrate your display, but still passthrough HDR content to the display? I'm not sure if Calman supports either of these two options, to be honest. HDR calibration is a difficult topic. My preferred approach would be to let madVR convert the HDR content to SDR and then run the result through a normal SDR 3dlut (though maybe for BT.2020 or DCI-P3 gamut).

Originally Posted by stefanelli73

I wanted to know if any of you could help me about the subtitles of blu ray scaled to 3840x2160, in practice when the fixed subtitles come out the image slows down by going for a few seconds, increasing the rendering and after the subtitles disappear it is replaced, this it happens even if I use JINC as IMAGE UPSCALING.
My video card is AMD RX480, media player JRIVER, os WINDOWS 10.

I don't really understand your explanation of what's happening. Can you explain in more detail?

Originally Posted by arabesc

I have a stable madVR crash on a movie file when the following settings are enabled

That's a weird one. It crashes while trying to upload the video frame to the GPU. Does this crash occur with all video files, or only some? Does it occur regardless of madVR settings, or do some madVR settings play a role in the crash (if so which)?

Originally Posted by Sunset1982

Do you know if madvr's madTPG works with the new autocal feature on the new 2018 lg oled tvs?

I'm not sure I understand the question. Are you asking if the LG TVs are able to remote control madTPG? If so, as far as I'm aware, the answer is no.

Originally Posted by MrGlasspoole

8GB RAM and task manager says 38% usage.
MPC-HC CPU usage goes up to 70% with madVR and QS in LAV.
With copy-back it's ~45%

With EVR the Sony HEVC 59/60fps Videos work up until ~25 seconds and sometimes one minute. Then the sound starts to stutter and slowly the Frame rate in the MPC-HC statistics window is going down.

The attachment is:
4K H.264-AVC MPEG-4 [60.000fps][8-35Mbps][8bits][DD 5.1 (AC-3)] Big Buck Bunny
in madVR.

With EVR this movies video is smooth but sound is totally out of sync. Sync Offset, dropped and Jitter show all 0.

According to your screenshot your decoder is too slow.

Originally Posted by jespermart

Does the scripting support unc path

Probably. Why don't you simply try?

Originally Posted by oldpainlesskodi

I have been following your work on the AVS forum re HDR to SDR conversion with corrective tone mapping - what are the chances of MadVR being able to trigger HDR, but perform the same corrective type of processing (as some suspect the OS HDR Toggle does)?

What do you mean with "corrective tone mapping" and "corrective type of processing"? I'm not really sure what you mean/want exactly.

Originally Posted by mclingo

what about having a virtual HDR mode that conversts SDR to something that looks like HDR, now that would be cool.

Not planned.

Originally Posted by huhn

yes and if that happens with "let madVR decide" i see this as a real problem and it should be changed to at least 3.0. quadrupling for 20 missing pixel is just a total overkill.

Originally Posted by Asmodian

I agree, quadrupling for less than a 3.0x upscale is total overkill, and even a negative at times. However, I thought "let madVR decide" for quadrupling was 2.4x so quadrupling should not have activated.

Originally Posted by thighhighs

2.4 default for double again mode (or NGU AA). For direct 4x "if any upscaling is needed" by default.

Please feel free to suggest better "let madVR decide" settings, I'm open for suggestions. Obviously the reason for choosing different setings for "direct 4x" is that it performs dramatically better than "double twice".

Originally Posted by BatKnight

I've been doing further testing and managed to capture better screenshots (download and present them at 1:1 resolution) to exemplify my findings:

NV HDR triggered by madVR (output 10 bit) BANDING
NV HDR triggered by madVR (output 8 bit) NO BANDING
OS HDR enabled at Windows 10 (output 10 bit) NO BANDING (no screenshot available, too bright)

This looks like a madVR bug to me. There's a technical limition in the OS right now which supports 10bit in fullscreen windowed mode - but only if the OS HDR switch is turned on. So as long as the OS HDR switch is turned off, madVR should not even try to switch to 10bit in windowed mode. If you want the OS HDR switch to be off, you have to use FSE mode to get 10bit. Or alternatively use 8bit and trust madVR's high quality dithering. It might actually look better than 10bit, depending on your display. Worth a test, at least.

[mclingo]

good to have you back MADSHI, hope everything's going well on the commercial side, thanks again for you continued support.

[madshi]

Big thanks to everyone who is helping with madVR support work. I really apprecate it. You're making it much easier for me to handle the overall madVR workload. So you're indirectly contributing to madVR development.

I planned to add a list of users who helped with support work a lot, but then I feared I might forget someone who might then be disappointed, so I decided not to list any user names. But please rest assured that I read all posts in this thread, and your support work is not going unnoticed. Thank you again!



Next madVR build will come maybe in a week or two or so, I'm not sure right now. It will feature improved HDR tone mapping quality and an additional "low" RCA quality level (which will fuse with NGU medium quality).

FYI, although I didn't have much time for madVR recently, I did start working on a totally new algorithm (not scaling related) last year, and I've been working on it ever since. I even hired external help for some things I didn't know how to do myself (I'm not a math genius, unfortunately). I'm not completely sure yet if I can manage to make it work. It might end up being much too slow, or too low quality, but I'm hopeful. If it works out as well as I hope, it could be a relatively big thing. But it's going to take several more weeks at least. Sorry, no more information at this point.

[huhn]

Originally Posted by maxkolonko123

where i can check it ? but i doubt that i do if u mean external 3d lut file

i got tick displaye is calibrated, disable gpu ramps, bt.709, pure power curve 2.20

you shouldn't have one.

[Asmodian]

Originally Posted by madshi

I'm not sure I understand the question. Are you asking if the LG TVs are able to remote control madTPG? If so, as far as I'm aware, the answer is no.

I do not believe madTPG would need to support AutoCal, that feature is talking to the TV over the network, not HDMI. madTPG would be the pattern generator but not the display.
Portrait Displays to Provide Auto Calibration for 2018 LG OLED and SUPER UHD TVs

Originally Posted by madshi

Please feel free to suggest better "let madVR decide" settings, I'm open for suggestions. Obviously the reason for choosing different setings for "direct 4x" is that it performs dramatically better than "double twice".

I still think using 2.4x would be more appropriate, direct 4x does take a little more GPU power than 2x and it would be consistent. Upscaling to 4x and then downscaling to 2.05x is not ideal.

Thanks for all your work, as always great stuff.

[Warner306]

"This looks like a madVR bug to me. There's a technical limition in the OS right now which supports 10bit in fullscreen windowed mode - but only if the OS HDR switch is turned on. So as long as the OS HDR switch is turned off, madVR should not even try to switch to 10bit in windowed mode. If you want the OS HDR switch to be off, you have to use FSE mode to get 10bit. Or alternatively use 8bit and trust madVR's high quality dithering. It might actually look better than 10bit, depending on your display. Worth a test, at least."

FSE is now required for 10-bit HDR passthrough, from madVR to GPU to display? I didn't know FSE was a requirement. This could be a problem for users getting occasional black screens on playback stop, or having FSE fail completely at playback start with Windows 10, which reverts to windowed mode without the user knowing, unless they read the message from madVR that FSE has failed. It's not that big of deal, but I'd like to know that FSE is required for full, 10-bit HDR passthrough, as most users are using windowed mode and madVR is reporting 10-bits during HDR playback. As a result, they appear to be getting banding.

Getting people to believe that they're not missing anything with 8-bits with dithering seems to be the bigger challenge. If FSE is problematic, they don't want to downgrade to 8-bits because their display is 10-bits and they think they need to take advantage of those extra bits.

Edit: The user I was talking to was using 12-bits at the GPU with AMD card for GPU passthrough, so this was actually a full 10-bit pipeline form madVR to the GPU to the display, not 8-bits at the GPU as previously reported. I assuming it is still necessary to have a complete 10-bit pipeline with AMD cards for HDR passthrough from madVR to the display?

[Warner306]

And one more thing...I assume the Windows OS is sending the metadata to the display untouched and isn't altering it to change the tone mapping, just like the private APIs do? So the Windows HDR calibration tool (which is simple as can be) only applies to the Windows desktop and not to video playback?

[Warner306]

"I have been following your work on the AVS forum re HDR to SDR conversion with corrective tone mapping - what are the chances of MadVR being able to trigger HDR, but perform the same corrective type of processing (as some suspect the OS HDR Toggle does)?"

This user wants to know if you can somehow send the PQ transfer function to the display to trigger HDR mode and use madVR's adjustable and sometimes superior tone mapping to replace the display's tone mapping? Sounds technically impossible, but would be nice.

Edit: I forgot about "process HDR content by using pixel shader math" as pointed out by Asmodian. How far would that get him? Are all of the YCbCr values tone mapped and stripped of illegal color values and then sent to the display to be tone mapped again? Does the metadata remain the same? Is this recommended?

[huhn]

and how do you tell a display to not tone map.

the closes to that is a HDR 3D LUT.

[Asmodian]

Originally Posted by Warner306

This user wants to know if you can somehow send the PQ transfer function to the display to trigger HDR mode and use madVR's adjustable and sometimes superior tone mapping to replace the display's tone mapping? Sounds technically impossible, but would be nice.

Isn't that what both "process HDR content by using pixel shader math" and "process HDR content by using an external 3DLUT" are for? At least as far as it can go, you cannot tell the display to turn off its tone mapping. However, you do get some of the benefits from madVR's superior tone mapping, or DisplayCal's when using a 3DLUT, and you can set the metadata so you can influence the display's tone mapping.

[Polopretress]

Originally Posted by LigH

Running x86-64 (AMD64) Windows applications requires both a CPU with x86-64 instruction set and a 64-bit code variant of a Windows operating system (optional and common as private user OS since Vista, more or less mandatory since 8).

Thanks to the extended compatibility in the x86-64 architecture developed by AMD (in contrast to the Itanium IA-64 architecture by intel which is incompatible to x86 / IA-32), 32-bit x86 processes still run in a SysWoW64 (Windows on Windows-64) environment, but all DLL's used by a 32-bit process must also be programmed for 32-bit x86 architecture, and all DLL's used by a 64-bit process must be programmed for 64-bit x86-64 architecture. Windows will check the DLL header and refuse to load libraries not matching the bitness of the calling process.

To be able to load the 64-bit variant of madVR (and LAV Filters), you have to run a 64-bit player application on a 64-bit Windows.

64-bit code can already be faster simply because CPU registers in 64-bit mode have a) twice the width and b) twice the number, compared to the 32-bit mode. Every instruction which can avoid RAM access by using data still stored in another CPU register is a little improvement.

OK, i was maybe not enough clear.
System is in Win10 64 bits.
The installation included lav32bits, reclock32bits, mpc-be32bits and madVR.
Linked to the issue with CPU 4k hevc decoding (multiple drop with "queue decoder" at 1 while CPU is not above 40% of usage), player potplayer64 bits has been installed without any other modification and fixed the problem.
Lav is still in 32 bits. Reclok also (of course)

It seems to work fine. i was surprised.

Question is : in this case, what is the mode used by madVR (32 or 64 bits) ?

[huhn]

if you use a 64 bit player everything else it 64 bit too.

[thighhighs]

Originally Posted by Asmodian

I still think using 2.4x would be more appropriate, direct 4x does take a little more GPU power than 2x and it would be consistent. Upscaling to 4x and then downscaling to 2.05x is not ideal.

I agree. Because "always" quadrupling in some cases does not improve the quality, but it might hurt performance.

[Warner306]

Originally Posted by Asmodian

Isn't that what both "process HDR content by using pixel shader math" and "process HDR content by using an external 3DLUT" are for? At least as far as it can go, you cannot tell the display to turn off its tone mapping. However, you do get some of the benefits from madVR's superior tone mapping, or DisplayCal's when using a 3DLUT, and you can set the metadata so you can influence the display's tone mapping.

I see. Are you still stuck with the backlight and contrast settings of the display's SDR mode? I don't completely understand how this works. Can you toggle the display's HDR mode by sending it the correct metadata by 3D LUT? Many calibration topics are not completely clear to me.

[Asmodian]

No, you are not stuck in SDR mode. You set the metadata with madTPG (madVR when the 3DLUT is used), measure the display as it responds given that metadata, and construct the 3DLUT based on that. For best results I have found you really need a 3DLUT for every movie, this can be done with DisplayCal's 3DLUT creating utility but it is a bit crazy. Because the TV is always in the same mode you only need to profile once but given that the mastering display's peak white and the video's peak white both influence the 3DLUT most videos would use a different 3DLUT. That said I get decent results as long as the peak whites for the 3DLUT are higher than the peak white in the video (if not they clip generating nasty banding artifacts).

What we need is pixel shader math that takes the source's metadata into account to convert to a specific set of metadata and then send that through an HDR 3DLUT. If that makes sense?

[BetA13]

Ha, nice one...You sneaky little developer....
Where there some performance improvements?
Because for me its like an early birthday present cause i can use NGU Medium now and still be under 20ms.
Before i could only use NGU Low.

Im impressed..I also think i did understand the whole MadVR now..what each setting does,, how it works, etc..
Using and understanding MadVR is like a study Project at the University, it took me allmost 5 years to figure it all out..
And i learned so freaking much about Video processing and upscaling filters, etc etc..
Everytime theres a new version, i get sweaty hands and start to test and fiddle wich each setting to see whats changed and such, and when iim finished i allways wonder where the Time did go..Its easy for me to get lost for hours while fiddling around with your awesome Programm..
Damn, nothing, really nothing comes close to the quality of MadVR..
Ever since i started using madvr i wondered what all the TV´s use for up and downscaling, because it looks soo bad..

anyway, what i wanted to say to you @madshi, THANK you for all youve done, and thank you for sparking my interesst in this area, this is kinda my hobby now, setting up Home cinemas for my friends haha..

have a good night

[sauma144]

I'm gonna have some more crazy dreams about this new algorithm.

[Grimsdyke]

Hi Madshi, nice to have you back and thank you for all the hard work !!
I have an issue with a few PAL-SD files in the following configuration:
-> MPC-BE + LAV (Hardware device to use: Automatic(Native)) + MadVR => File plays normally for a few seconds then screen turns completely green !! File keeps playing and I hear sound.
-> MPC-BE + LAV (Hardware device to use: (GPU selected)) + MadVR => File plays normally
Unfortunately I can' keep this configuration because the drop in performance is too big for 4K. Not that important but maybe you could look into this when you have the time. Thx !!