madVR - high quality video renderer (GPU assisted)

[Asmodian]

Originally Posted by madshi

Thanks for your feedback. Some votes for monoColor coming in late in the game now.

I have to say; oppositeColor_dynamic looks great too. Only if I had to choose.

[Ver Greeneyes]

Originally Posted by madshi

Actually, that was my plan, yes. If I do add those options (and it seems I will have to) then they will apply to random dithering, too.

Nice! I'll have to play around with it on my laptop (which can't do ED) if you do add the options

Originally Posted by madshi

Yes, I'm quite sure that is going to happen...

To be fair there's already a few options in madVR where we tell people "just use the default unless you see a clear advantage" - I don't think these will be different.

[*Touche*]

Originally Posted by Ver Greeneyes

To be fair there's already a few options in madVR where we tell people "just use the default unless you see a clear advantage" - I don't think these will be different.

A "Do not touch unless you know what you're doing!" section could be added for these advanced options.

[cyberbeing]

Originally Posted by madshi

FWIW, I plan on releasing one more test build set in 1-2 days, probably containing 3 (very different) new test builds.

Well just keep in mind, I was also the one few who was strongly bothered by the patterns produced by the initial low noise non-random error diffusion builds, when watching actual videos. If any of the new dithering ideas produces a discernible patterns, you could predict in advance that I personally won't like it. So if I'm hearing you correctly that you're now moving back to deciding on the non-random "low-noise" algorithm, I was always planning to sit out and not influence the decision about a dithering method I'd never use myself.

[madshi]

Originally Posted by sajara

I've been testing oppositeColor over previous builds and my subjective opinion is that is my favorite so far.

I've tested this time in all my TV's, a 55" LED, a 32" LED, as well as my 15.4" laptop screen at 768p

Tested with blueray; B&W DVD media, several transcoded sample clips in SD and 1080p.

Now, I can only see dithering in 4bit mode. Period. In 8 bit is pretty damn impossible. Even zooming in with the magnifier is indiscernible.

Ok, thanks for your feedback!

Originally Posted by Asmodian

I have to say; oppositeColor_dynamic looks great too. Only if I had to choose.

Ok, thx...

Originally Posted by cyberbeing

Well just keep in mind, I was also the one few who was strongly bothered by the patterns produced by the initial low noise non-random error diffusion builds, when watching actual videos. If any of the new dithering ideas produces a discernible patterns, you could predict in advance that I personally won't like it. So if I'm hearing you correctly that you're now moving back to deciding on the non-random "low-noise" algorithm, I was always planning to sit out and not influence the decision about a dithering method I'd never use myself.

Only one of the 3 new test builds will be aimed at low-noise.

[iSunrise]

Originally Posted by 6233638

Done. (and marked it as a "major" bug by mistake - sorry)

Turns out it's the combination of NNEDI3 chroma upscaling (not doubling) + video mode deinterlacing that is causing it.

Yes, that is the same bug I reported and madshi had confirmed to be reproducible.

Thanks for the bug report, it sounded like madshi already fixed that in his sources. Apparently not.

[leeperry]

Originally Posted by madshi

A4 is already 0.97.

That would be oppositeColor...

Yep I know, but how about a less noisy luma-only 0.97/NL6/A4 monster? Like with a more diffuse noise that would look better in 4bit? I'm sure you could pull this performance without resorting to the blinking RGB noise madness. I kinda got used to that pretty much daily game-changing build euphoria, enough is never enough

Originally Posted by Asmodian

Actually this isn't the main reason BFI is so nice. It is actually retention on your retina, the light receptors have a change time and flashing black helps with motion blur, like Film, CRT, Plasma, and backlight based BFI LCD.
edit: I should have said the only reason BFI is so nice.

Fully agreed, BFI is on a necessary evil basis....and I'm willing to believe that 4:4:4 might somewhat matter for the fluctuating RGB noise mVR builds but +95% of TV's in the world are used to watch movies in 4:2:2.

mVR is conservative enough to still supports XP, it cannot be seriously expected for every mVR user to watch movies in 4:4:4 when pretty much all TV's turn down all their best features(24/50Hz support, REC.709 gamut, BFI, etc).

This said madshi said that 4:4:4 didn't matter much if at all, I guess he'd know

Originally Posted by *Touche*

It's not that I don't notice when something is different, or appreciate subtle improvements in audio or video, but I tend to be more critical, sceptical, scientific, down to earth with it. I have to keep in mind that some things are real and some are placebo. That is why I renamed the builds for testing and later compared the hashes to see which ones I could honestly identify.

Every A/V reviewer is well aware that placebo & nocebo are very real, that the human brain is a very poor measuring tool and all that. The DBT game has been run and madshi already drew his own conclusions.

Originally Posted by bacondither

Well, DLPs and Plasmas are 1-bit native, and by using a high speed temporal(dynamic) dither(at several khz),gives an apparent bitdeph somewhere around 6-7 bits for DLP.
I am sure they are using some spatial dithering in conjunction with the latter.

Strangely I can stand fast enough DLP(many ppl claim that Mitsu are/were some of the best at mastering it) but yeah it's very obviously a dithering feast and I would presume that this is the foremost reason to the grainy appearance of those two technologies. DLP is one nasty trick(and no studies were made regarding its influence on a long time basis, that doesn't seem too healthy IMHO) but Plasma is really a flickering feast.

[Olivier C.]

Hi,

I think the calibration process madVR + dispcalGui is not very consistent because madVR 0.87.4 - madTPG does not work.
I had to replace with another release (november 2013) : MadVR Test Build with madTPG fixes wich does not contain last (very nice) work on Error Diffusion.

When i extract and overwrite the contents of MadVR Test Build with madTPG fixes to madVR folder, new features of 0.87.4 are not available (especially Error diffusion, but also Image doubling, etc).
For profiling / calibration, i have to use an old madVR release which does not contain Error Diffusion great and for displaying movies, i want to use the last release 0.87.4 for Error Diffusion great stuff.

May be i am wrong

Thanks a lot

PS : Thanks for all your great stuff madshi
And sorry for my poor english (i'm french)

[leeperry]

Originally Posted by Olivier C.

Hi

BTW, what is it that you meant on HCFR regarding the "ITC Processing" option in the AMD GPU drivers and its effect on mVR exactly?

[YxP]

I may be too late for the party since new test builds are coming, but after a close battle my favourite is opposite colour static. Earlier in this thread someone said the dynamic was "energetic", and I have come to the same conclusion. Static somehow feels more peaceful and relaxing to the eye.

[Olivier C.]

Originally Posted by leeperry

BTW, what is it that you meant on HCFR regarding the "ITC Processing" option in the AMD GPU drivers and its effect on mVR exactly?

Hi leepery ;-)

I realized that disabling "ITC processing" in AMD GPU drivers has a very noticeable effect on first % of luminance (maybe between 0% and 5%). It seems that disabling "ITC processing" crushes blacks but does not lower black level, according to measurements.
In my opinion, the effect of ITC Processing is more general, and does not only affect madVR (picture display, games, etc).

[omarank]

Madshi, would it be too much work to implement Directcompute support for DX10 cards (only for ED algorithms)? I find it too difficult resisting myself from testing these ED builds. I understand though that performance might not be great, but I would be happy using ED even if it works only for SD content.

[magus]

Hey guys I am benchmarking some videos with madvr using a new vbios of my GPU and was wondering what a good way to test the limits of the GPU related to video rendering(and more specifically madvr) would be? Would downloading a very high-bit video and checking Cntrl+J to see framerate and dropped frames be good?


Also I notice some tearing on videos when Aero is disabled when using madvr but other renderers don't seem to have as much a problem. Does anyone else see this or does everyone just leave Aero on?

[cyberbeing]

Originally Posted by iSunrise

Yes, that is the same bug I reported and madshi had confirmed to be reproducible.

Thanks for the bug report, it sounded like madshi already fixed that in his sources. Apparently not.

Yeah, I remembering mentioning that issue shortly after you did as well:

Originally Posted by cyberbeing

NNEDI3 4:2:0->4:4:4 chroma upscaling is broken on my GTX 770 with Video Mode deinterlacing. The screen turns a greenish yellow. Film Mode is unaffected

Just goes to show, madshi forgets things when we don't add entries to his bug tracker.

[iSunrise]

Originally Posted by cyberbeing

Just goes to show, madshi forgets things when we don't add entries to his bug tracker.

Just to be safe, I recently added a bug report for a similiar DCI-P3 issue, which is probably related to the one you already reported, that triggers every time for me when I enable DCI-P3, whether or not I use a 3DLUT (it´s also not resolution dependent).

[cyberbeing]

It's probably about time I create a bug tracker entry for that uncompressed v210 in avi issue as well.

Edit: Done

[leeperry]

Originally Posted by magus

I am benchmarking some videos with madvr using a new vbios of my GPU and was wondering what the best way to test the limits of the GPU related to video rendering(and more specifically madvr) would be?

OpenCL benchmarks would make sense if you care about NNEDI, otherwise the best throughputs in any simple D3D benchmark the better basically.

Originally Posted by magus

I notice some tearing on videos when Aero is disabled when using madvr but other renderers don't seem to have as much a problem. Does anyone else see this or does everyone just leave Aero on?

I also get tearing when I disable Aero on W7SP1, FSE takes care of it

Originally Posted by Olivier C.

I realized that disabling "ITC processing" in AMD GPU drivers has a very noticeable effect on first % of luminance (maybe between 0% and 5%). It seems that disabling "ITC processing" crushes blacks but does not lower black level, according to measurements.

Humm interesting, thanks for the info.

Were you able to reproduce this behavior with several displays of different brands? What's the easiest way to find out? Running ArgyllCMS on the first 5%?

[James Freeman]

Originally Posted by madshi
Well, I'd like that. Not sure why Microsoft didn't introduce something like this when they implemented desktop composition. Would have been a good opportunity.

Anyone know how can we directly contact the manufacturers to clearly convey this message to them?

It'd be VERY nice too see "Dither Output" in the Nvidia/Catalyst Control Panel/Center.
Gone will be the problem of banding after calibration, and 10-bit Pro Monitors & Video Cards ($$$$) would be less important (almost obsolete) even for critical editing like Photoshop.

I am really grateful that MadVR does just that,
Thank you madshi.

[The 8472]

Originally Posted by madshi

FWIW, I plan on releasing one more test build set in 1-2 days, probably containing 3 (very different) new test builds. One of them is going to be "simple" ordered dithering (based on a modified/improved version of "void-and-cluster"). The 2nd one is going to be a combination of ordered dithering and error diffusion. The 3rd one is going to be an updated low-noise error diffusion algorithm. Straight ordered dithering should perform faster than error diffusion. The combination of ordered dithering and error diffusion could maybe bring another quality improvement to dither dot placement. The low-noise-alternative might be another option to select for users. These test builds will probably only be monoColor, to make testing simpler. This final test build set will decide which algorithms are going to be offered as options in madVR.

Nice. For day-to-day video playback I don't care much for the finer variants of the error diffusion builds, apart from an interest in tinkering with algorithms.

But a possibly faster alternative that allows me to use smooth motion together with dithering and without turning my gfx card into a heating unit is certainly welcome for practical use.

[nevcairiel]

Originally Posted by The 8472

But a possibly faster alternative that allows me to use smooth motion together with dithering and without turning my gfx card into a heating unit is certainly welcome for practical use.

You mean like random dithering? =)