madVR - high quality video renderer (GPU assisted)

[6233638]

Originally Posted by bacondither

Random dithering was changed in the debanding testbuilds. If the pixel(s) have a integer value of 0,0,0 ie black, no dither will be applied and local dimming LCDs can shut down the backlight completly.

So why would error diffusion dither be better for local dimming LCDs?

This does not apply if you are using a 3DLUT.
Random Dither only gets disabled on black if you are using madVR with the "disable calibration controls for this display" or "my display is already calibrated" options.
Error Diffusion seems to work correctly with 3DLUTs.

[DarkSpace]

I finally got suspicious of my excellent OpenCL performance (my card is an AMD HD6970M), and decided to verify that Image Doubling is actually being used, and it turns out that no NNEDI features appear to be used at all. I also updated my drivers and re-tested, but the issue persists.
To test, I set my Image Upscaling algorithm to Nearest Neighbor (so I'd be able to spot it easily), and played some video at doubled resolution (with Image Doubling enabled, of course). The output had obvious NN scaling applied. I also tried to increase the video scaling factor to something a little more than 2, but the issue stayed the same.
Is this a bug (it seems to silently fail and fall back to using Image Upscaling or Image Downscaling settings, depending on whether image doubling should have resulted in a resolution smaller or larger than the output resolution), or am I doing something incredibly wrong?
Do you need a Debug Log?

[cca]

Originally Posted by DarkSpace

I finally got suspicious of my excellent OpenCL performance (my card is an AMD HD6970M), and decided to verify that Image Doubling is actually being used, and it turns out that no NNEDI features appear to be used at all. I also updated my drivers and re-tested, but the issue persists.
To test, I set my Image Upscaling algorithm to Nearest Neighbor (so I'd be able to spot it easily), and played some video at doubled resolution (with Image Doubling enabled, of course). The output had obvious NN scaling applied. I also tried to increase the scaling factor to something a little more than 2, but the issue stayed the same.
Is this a bug (it seems to silently fail and fall back to using Image Upscaling or Image Downscaling settings, depending on whether image doubling should have resulted in a resolution smaller or larger than the output resolution), or am I doing something incredibly wrong?
Do you need a Debug Log?

Assuming you're using the latest drivers for your card, you're not supposed to increase the NNEDI factor, but decrease it. That number the lower it is, the sooner NNEDI will be used, not the opposite.

[Stereodude]

Originally Posted by pie1394

However I have posted it previously that HD7970 is even unable to handle DVD 720x480i60 16:9 contents to 1080p60 output with following options:

1. DXVA2 MPEG2 VLD (very minor mem bandwidth used on GPU)
2. Vector-adaptive deinterlacing + frame rate doubling
3. NNEDI3 Image Doubling (default --> Luma 64, Chroma16)
4. All Jinc3AR upscaling
5. debanding_with_Angle_detection

FWIW, I can do the following on my 7790 for DVD to 1806x1016 (odd resolution due to overscan compensation on my CRT HDTV).

1. Software MPEG2 decoding
2. madVR IVTC
3. NNEDI3 Image Doubling (Luma 64, Chroma 32)
4. All Jinc3AR upscaling
5. smooth motion on

I'm not using OpenCL error diffusion though.

[DragonQ]

If you're restricting yourself to 30 fps material you're obviously going to be able to use higher settings than would work for 50/60 fps material.

[iSunrise]

Originally Posted by Q-the-STORM

it kinda sucks that madVR doesn't give a message "opencl not available, using different settings" something like that...

Maybe madshi can include something like an OpenCL checkbox like GPU-Z somewhere or even better, grey out all the related OpenCL settings if it doesn´t detect working OpenCL support, so you can´t even select them. This would lower the confusion a lot.

[The 8472]

Originally Posted by DragonQ

If you're restricting yourself to 30 fps material you're obviously going to be able to use higher settings than would work for 50/60 fps material.

With profiles there's no need to restrict yourself anymore. Optimal settings for any scenario.

[pie1394]

Just retested various samples on hand to determine what the limitation is on HD7970 with madVR 0.87.4. The safe profile rules are listed as below:

if (srcWidth > 1920) and (srcHeight > 1080) "UHD"
else if (srcWidth <= 1024) and (srcHeight <= 576) "SD"
else if ( deintFps >= 50 ) "HDi"
else "HD"

Basically all profiles share the same configuration like
- debanding + angle_detection
- image upscaling: Jinc3AR
- image downscaling: Catmull-Rom + linear-light
- Turn off all "trade quality for performance" settings
- no dithering
- no GPU DXVA2 decode

The different part is just the chroma upscaling to control the madVR's workload always below 16.6ms. In reality for Jinc3AR mode it only needs 6 ~ 10 ms with various source resoultion to 1920x1080, and 12ms for 3840x2160 source's Luma with Catmull-Rom + linear-light. Chroma NNEDI3 2x doubler + D3D-OpenCL inter-op make it increased to 13 ~ 3x ms depending on content resolution and neuron count.

[UHD]
Jinc3AR

[HD]
NNEDI3 32 neurons

[HDi]
NNEDI3 16 neurons

[SD]
NNEDI3 128 neurons


I just can say that madVR's chroma NNEDI3 4:2:0 to 4:4:4 somehow also helps the TV's super-resolution engine performance over Jinc3AR scaling on motion objects of clean BD contents with my subjective observation. Yet it is quite difficult to tell any big difference in single frame comparsion.

ps: The TV is Sony 65HX920, adjusted for normal lighting room viewing condition -- thus higher peak white and more light leakage viewed in dark room condition.

[mindz]

I use a Windows font size of 125%, with this setting i cannot choose NNEDI3 as Image upscaling method:



IMG doesnt work i think. Check the following link for picture:

https://www.dropbox.com/s/dd3mo7mk8vcgalm/nnedi3noshow.png

[shimaflarex]

Originally Posted by mindz

I use a Windows font size of 125%, with this setting i cannot choose NNEDI3 as Image upscaling method:



IMG doesnt work i think. Check the following link for picture:

https://www.dropbox.com/s/dd3mo7mk8vcgalm/nnedi3noshow.png

No one can select NNEDI as an image upscaling method.
There's no radio button under Jinc.
You need to select NNEDI3 as an image doubling algorithm (NNEDI3 can only double the image size).

[DragonQ]

Originally Posted by mindz

I use a Windows font size of 125%, with this setting i cannot choose NNEDI3 as Image upscaling method:



IMG doesnt work i think. Check the following link for picture:

https://www.dropbox.com/s/dd3mo7mk8vcgalm/nnedi3noshow.png

What do you mean? That window has no other options. To use NNEDI3 you need to go to the "image doubling" tab (which to be honest is a bit confusing).

[DarkSpace]

Originally Posted by cca

Assuming you're using the latest drivers for your card, you're not supposed to increase the NNEDI factor, but decrease it. That number the lower it is, the sooner NNEDI will be used, not the opposite.

What do you mean by "NNEDI factor"?
If you mean the image scaling factor, I strongly believe it should be dependent only on my madVR Image Doubling settings (only if factor is 2x or greater in this case), and if you mean how many neurons I use, I tried 256 neurons and 16 neurons, and both gave the same result.

[iSunrise]

Originally Posted by DarkSpace

What do you mean by "NNEDI factor"?
If you mean the image scaling factor, I strongly believe it should be dependent only on my madVR Image Doubling settings (only if factor is 2x or greater in this case), and if you mean how many neurons I use, I tried 256 neurons and 16 neurons, and both gave the same result.

He is refering to the fact that you stated "I also tried to increase the scaling factor to something a little more than 2, but the issue stayed the same." which leads someone to believe that you meant the scaling factor description from the NNEDI3 double/quadruple settings. After re-reading that about 10 times and since you cannot go any higher than 2.0x without enabling "always...", you seem not to be refering to the scaling factor settings at all, but the actual scaling factor that you applied manually when you played back a clip.

If whether NNEDI3 4:2:2 -> 4:4:4 chroma upscaling nor image/chroma doubling/quadrupling does make a difference on low-resolution clips there is clearly something wrong, though.

[DarkSpace]

Originally Posted by iSunrise

He is refering to the fact that you stated "I also tried to increase the scaling factor to something a little more than 2, but the issue stayed the same." which leads someone to believe that you meant the scaling factor description from the NNEDI3 double/quadruple settings. After re-reading that about 10 times and since you cannot go any higher than 2.0x without enabling "always...", you seem not to be refering to the scaling factor settings at all, but the actual scaling factor that you applied manually when you played back a clip.

Oops. I can see what you mean, and you're right, I meant the video scaling factor. I edited my original post to prevent further misunderstandings.

Originally Posted by iSunrise

If whether NNEDI3 4:2:2 -> 4:4:4 chroma upscaling nor image/chroma doubling/quadrupling does make a difference on low-resolution clips there is clearly something wrong, though.

I only tested Image scaling, because I'm not sure how obvious the difference in 4:2:0-to-4:4:4 would be, seeing how a Lanczos-3 AR is used after NNEDI automatically to compensate for the shift, while with Image Upscaling, I could specify which algorithm to use after Doubling.

[Asmodian]

Originally Posted by XMonarchY

Specs - GTX 770 2Gb with 327.23 drivers and Windows 8.1.

I use NNEDI3 Chroma Upscaling (64 neurons) with NNEDI3 Chroma & Luma Image Doubling (16 neurons) for any SD or 720p content on my 1080p LCD TV. I keep the same settings for 1080p content, so I guess that means Image Doubling is not used for 1080p, but NNEDI3 Chroma Upscaling (64 neurons) is still used. I also use OpenCL dithering, MPC-HC, LAV Video - CUVID decoder, ArgyllCMS 3DLUT.

I see that many people prefer JINC3AR for Chroma Upscaling when using NNEDI3 Image Doubling - is that a higher quality combination than using NNEDI3 for both - Image Doubling AND Chroma Upscaling? To my eyes, 1080p videos look better with NNEDI3 than with JINC3AR, but I could be wrong...

I mentioned I was using Jinc3AR instead of NNEDI3 chroma upscaling but that was only because that was where I was maxing out my card with the NNEDI3 image doubling options I was using. Using suggestions from both madshi and 6233638 I did more testing; turning off NNEDI3 chroma doubling and turning on NNEDI3 chroma scaling. I now also believe NNEDI3 chroma scaling and only luma doubling is a better tradeoff. I agree with them (and you): I really like NNEDI3 chroma upscaling (4:2:0 -> 4:4:4).

Originally Posted by XMonarchY

My TV is set to 4:2:2 (HDMI to HDMI, 0-255) because 4:4:4 does not allow me to calibrate CMS properly. Does that mean I am not getting the benefits of Chroma Upscaling, OpenCL dithering, and Image Doubling??? Maybe its a placebo, but NNEDI3 Chroma Upscaling with NNEDI3 Image Doubling and OpenCL dithering + extremely accurate 3DLUT looks absolutely stunning!

I do not understand how your TV would not allow you to calibrate CMS properly when in 4:4:4, are you using it for content other than madVR? If you are only using madVR you can do all the color management with ArgyllCMS's 3DLUT in madVR and leave the TV in a "just display what you get" mode. That said I bet you do get some benefits now but not the full glory.

Originally Posted by XMonarchY

NNEDI3 is shown to have some "other artifacts" but I have not seen any... What are they? When would I see them?

The NNEDI3 artifacts are odd details at are "different" after NNEDI3 compared to the original, they are really hard (impossible?) for me to notice unless comparing to the original. madshi posted some examples a few pages back.

Originally Posted by Asmodian

I do not understand how your TV would not allow you to calibrate CMS properly when in 4:4:4, are you using it for content other than madVR? If you are only using madVR you can do all the color management with ArgyllCMS's 3DLUT in madVR and leave the TV in a "just display what you get" mode. That said I bet you do get some benefits now but not the full glory.

Thank you for explaining. My Samsung LN40B550 40" CCFL SPVA LCD simply locks CMS controls when in PC Mode, which is the mode that uses 4:4:4. In that mode color accuracy is bad enough that I will need about 4300 patches to achieve the same results as with 4:2:2 mode. Almost all Samsung TVs are like that in PC 4:4:4 Mode.

At the same time, I do not use Movie 4:2:2 mode for several reasons, one them being text sharpness. In Standard 4:2:2 mode that I use, I disable sharpness completely, but text remains much much better than in Movie 4:2:2 mode, where text is just plain bad. Let's just say that there is very little difference between Standard 4:2:2 mode and PC 4:4:4 mode, but there is a huge difference between Movie 4:2:2 Mode and PC Mode.

In Standard 4:2:2 mode I calibrated CMS manually (all of CMS readings were under dE 3) and then applied 3DLUT profile on top, using 2527 patches. The result was amazing - about 3 saturation sweeps as high dE 1.1, one ColorChecker sweep as high as dE 1.5, and all the rest is under 1.0 - measured by i1 Display 3.

I still see a stunning image using madVR rendering though. I could try to create a 3DLUT for PC 4:4:4 mode using 4300 patches over night - can't think of a better way to test the new ArgyllCMS 1.6.3 stable release (not beta) and dispcalGUI 1.7.1.6.


EDIT: Actually, my TV can't use the 48Hz judder-free mode for 24p content in 4:4:4. I would have to trade judder-free smoothness for 4:4:4 subsampling... not sure which one is worth more to me...

[Mano]

Originally Posted by cyberbeing

Do you use XySubFilter with a very high CPU queue setting of 64 in madVR?

i only have LAV Video, Audio decode and LAV spliter.

[huhn]

Originally Posted by XMonarchY

EDIT: Actually, my TV can't use the 48Hz judder-free mode for 24p content in 4:4:4. I would have to trade judder-free smoothness for 4:4:4 subsampling... not sure which one is worth more to me...

just force 60 hz and use smooth motion best you can do with samsung.

[Asmodian]

@XMonarchY

It sounds like you understand your TV very well and I trust you have it correctly configured for your conditions and preferences. My old 2010 LG Plasma's CMS did more harm than good.

Originally Posted by XMonarchY

I use NNEDI3 Chroma Upscaling (64 neurons) with NNEDI3 Chroma & Luma Image Doubling (16 neurons) for any SD or 720p content on my 1080p LCD TV.

That said I doubt NNEDI3 16 chroma doubling is worth it, I suggest trying out NNEDI3 Chroma Upscaling 32 and Luma only Image Doubling at 32 or 64 if you can.

Originally Posted by huhn

just force 60 hz and use smooth motion best you can do with samsung.

If the TV is doing BFI or something else better than smooth motion it might be worth 4:2:2. I quite like smooth motion on a 60Hz display but it does add a lot of ms to rendering times, especially when using OpenCL error diffusion dithering.

Originally Posted by Asmodian

@XMonarchY
That said I doubt NNEDI3 16 chroma doubling is worth it, I suggest trying out NNEDI3 Chroma Upscaling 32 and Luma only Image Doubling at 32 or 64 if you can.

From what I understand - Image Doubling only works for content that is below the native resolution of the display, correct? My TV is set to its native 1080p, so Image Doubling would NOT be applied to any 1080p content, right? Since Image Doubling is ignored for 1080p content, then only NNEDI3 Chroma Upscaling is being used in such cases. My PC handles 64 neurons for Chroma Upscaling without any frame losses.

I see a considerable image improvement in 720p and below resolution content when using NNEDI3 Chroma Upscaling (64 neurons) in combination with NNEDI3 Image Doubling (Chroma & Luma) with 16 neurons. I also tried using 32 neurons for Chroma Upscaling and upping Image Doubling Chroma and Luma to 32 neurons, but that was a stutter-fest. But I see what you're saying - since Chroma is always upscaled with NNEDI3, there may be no point in doubling Chroma, which would also decrease performance. Instead, I could use Chroma Upscaling with only Luma Doubling but with higher neurons if I disable Chroma Doubling.

Thanks again! I really wish there was a sure way to tell what looks better with some kind of picture analysis benchmark.