Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
5th February 2017, 02:34 | #42241 | Link | ||||||||||||||||||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
Quote:
Quote:
Quote:
Quote:
How did you create the chroma subsampled image (4:2:0) from the original image, though? Did you take the chroma shift into account? Quote:
Quote:
It seems to me the user just wants to switch between 2 different profiles, one for 2D and one for 3D. In that case why not simply using "if (3D) "1080p 3D" else "1080p""? Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Have you tried software decoding? Quote:
If your display can't handle HDR on its own, then do a normal SDR calibration on the calibration tab. That calibration will also cover HDR content, as long as you allow madVR to convert HDR to SDR by using pixel shader math. |
||||||||||||||||||
5th February 2017, 02:35 | #42242 | Link | |||||||||||||||||||||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
Quote:
Quote:
Quote:
In all my scientific tests (using groundtruths and measurements like PSNR and SSIM), NGU runs circles around NNEDI3. Fact is: NNEDI3 has a specific purpose (originally: deinterlacing; somewhat unintended: upscaling of highly aliased sources) and works great for that specific purpose. But it's rather bad for upscaling high-quality video/photo sources. NGU also has a specific purpose (mainly upscaling of high quality video/photo sources) and works great for that specific purpose. NGU fails totally if you feed it with crappy sources. NGU doesn't like aliased sources, or sources with lots of visible compression artifacts. Quote:
Already YCbCr -> RGB conversion means you need to apply dithering (noise), otherwise you get much more dramatic artifacts then noise. Quote:
Quote:
Dithering is a whole different topic. There's a law in digital processing: Whenver you reduce bitdepth, you *HAVE* to apply dithering. It's good, it's necessary, and if you don't do it, very bad things will happen to image quality. Quote:
Quote:
Quote:
Quote:
Quote:
I suppose that PowerDVD etc use private APIs from AMD, NVidia and Intel to make other 3D formats work. I don't really have much interest in implementing custom support for 3 different GPU driver APIs. Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
|
|||||||||||||||||||||
5th February 2017, 02:36 | #42243 | Link | ||||||||||||||||||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
Sure, a nice graph would be lovely, but it also costs development time, which I'd rather spend elsewhere right now, to be honest. Quote:
And I'm glad to hear that madVR proved to be the most accurate renderer - that's what I was always aiming for! From your screenshots it looks as if the chroma placement keeps getting worse, though. Is it possible that your capture card is responsible for that? I'm pretty sure that madVR does the chroma placement correctly. Well, at least for MPEG2, VC-1 and h264 content. Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
http://www.avsforum.com/forum/139-di...argyllcms.html Quote:
Quote:
Quote:
Quote:
Could be DPC latency issues. Google for "dpc latency checker". |
||||||||||||||||||
5th February 2017, 02:48 | #42245 | Link | |
Registered User
Join Date: May 2013
Posts: 77
|
Quote:
I'm not sure why there was no issue before, maybe it never worked with 8 but updating madVR reset that setting somehow (when it introduced new settings UI), maybe it worked before but something triggered that bug. Anyway, 6 is fine too) |
|
5th February 2017, 02:49 | #42246 | Link |
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
So, here's a little test build. Simply replace the v0.91.5 files with these:
http://madshi.net/madVRpixart.zip I've created a new NGU variant which I'm going to call maybe "NGU pixart" or "NGU smooth", I'm not sure yet. This NGU variant will be optionally available in addition to the already existing NGU algorithm. The "pixart" variant has a look that is quite similar to NNEDI3. Which means it's very smooth and works great to reduce aliasing artifacts in the source. However, just as NNEDI3, "NGU pixart" is quite soft. My development goal was to create a NNEDI3 successor which has similar properties, but achieves a better "quality per watt" ratio. Currently available are NGU pixart in "low", "medium" and "very high" quality settings. The "high" quality setting internally uses the "medium" algorithm for now. There'll be a separate "high" algo in the next official build. In order to keep my development cost low, I haven't changed the settings dialog yet. So this test build replaces the main NGU algorithm with "NGU pixart". Which means this test build does not do the main NGU algorithm at all. The one and only purpose of this test build is so you guys can directly compare "NGU pixart" to NNEDI3, for both luma doubling and chroma upscaling. So now I need your FEEDBACK: 1) Do you think NNEDI3 looks better? Or the new NGU pixart variant? 2) How does performance compare for you? 3) Can I replace NNEDI3 with NGU pixart in the next official build (pretty please)? 4) Can I replace super-xbr (for luma doubling, only) with NGU pixart in the next official build? If you prefer NNEDI3, please let me know why. If you think NNEDI3 looks better, please provide comparison screenshots (for luma doubling comparisons always include the original unscaled source, please). Thanks! |
5th February 2017, 03:17 | #42247 | Link |
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Here's a little comparison comparing various algorithms when upscaling pixel art (highly aliased sources):
low-res original -- | -- Jinc -- | -- waifu2x -- | -- NGU very high -- | -- super-xbr 100 -- | -- NNEDI3-16 -- | -- NNEDI3-256 -- | -- NGU pixart low -- | -- NGU pixart med -- | -- NGU pixart very high As you can see, with sources like this, linear resamplers like Jinc fail. Also the standard "NGU Very High" and even the mighty waifu2x fail here. super-xbr does fine, after all that's exactly what it was made for. But best in class are NNEDI3 and NGU pixart. Which do you prefer? (Zooming a little into the images will reveal more.) |
5th February 2017, 03:26 | #42248 | Link | |||
Registered User
Join Date: Oct 2012
Posts: 7,903
|
Quote:
there are no in deep test for x4 because i didn't had card to use it. Quote:
i personally prefer "set and forget" scaler like super XBR. Quote:
i would even say a luma only super XBR version would be very nice for iGPUs at least i don't see NGU low to be faster. |
|||
5th February 2017, 03:31 | #42249 | Link |
Registered User
Join Date: Dec 2010
Posts: 254
|
Ok, will test.
1) Does the pixart also apply to NGU Chroma or Upsampling only? 2) Is it possible for Chroma to scale in 1-step like Super-XBR when doubling? Like If I've selected NGU High for Chroma Scaling and selected NGU High for Luma Doubling, to just have the option under it for Chroma Doubling grey out and have it scale in a single step without the need for doing Bicubic 60? NGU-VeryHigh still performs quite slow on my 1080 GTX and drops frames down to 0-2 render queue, so it's still unuseable. Maybe in another 2 years when there's video cards fast enough. Same for Chroma. High works ok, but cuts it pretty close for 24p content. 30p and higher content is not possible with NGU because it can't render fast enough with any current graphics card. I never figured out how to make profiles, so I just make one setting for all content. Last edited by AngelGraves13; 5th February 2017 at 03:50. |
5th February 2017, 03:38 | #42250 | Link | |
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
NGU low will probably be slower on Polaris, but on other GPUs it's nearly the same speed as super-xbr. |
|
5th February 2017, 04:01 | #42251 | Link | |
Registered User
Join Date: Oct 2012
Posts: 7,903
|
Quote:
i think the general lines on NGU pixart low are to thin it get's better with medium but with very high the background lines are really strange/uneven and nnedi3 wins there hands down. even NGU veryhigh wins there most of the part the line thickness is really good! but there or other obvious issues with NGU very high. mario him self looks "best with pixart very high. his eyes are to small but compared to super XBR they are pretty good. i have a huge problem selecting a winner here... so my last standing argument is that NGU is "hit or miss". super XBR works with everything. |
|
5th February 2017, 04:12 | #42252 | Link | ||
Registered User
Join Date: Oct 2012
Posts: 7,903
|
Quote:
and on top of it quadrupling chroma is questionable... Quote:
|
||
5th February 2017, 04:19 | #42253 | Link |
Registered User
Join Date: Mar 2009
Posts: 3,646
|
NICE. The low preset is terrible though IMO, I would remove it as the improvement from low to medium is quite considerable. I need to play with it more to state which I prefer, it's going to take some time..
I still do prefer the slightly softer more natural look of NNEDI3, but a softer downscaler or soften edges 1 might make it comparable. Any possibility for control of sharpness within the algorithm or do we have to rely on additional resizers and postprocessing? Last edited by ryrynz; 5th February 2017 at 04:26. |
5th February 2017, 04:26 | #42254 | Link | |
Registered User
Join Date: Dec 2010
Posts: 254
|
Quote:
Guess I'll stick with either Jinc or Super-XBR. Looking forward to getting that 27 inch 4K HDR monitor this year from Asus Last edited by AngelGraves13; 5th February 2017 at 04:44. |
|
5th February 2017, 10:23 | #42256 | Link |
Registered User
Join Date: Sep 2014
Posts: 280
|
Is there a way to set Hotkeys for different profiles?
If there will be a new NGUsmooth variant for lower quality sources, it might make sence to be able to change the profile ota when watching high or low quality material.
__________________
Intel i5 6600, 16 GB DDR4, AMD Vega RX56 8 GB, Windows 10 x64, Kodi DS Player 17.6, MadVR (x64), LAV Filters (x64), XySubfilter .746 (x64) LG 4K OLED (65C8D), Denon X-4200 AVR, Dali Zensor 5.1 Set |
5th February 2017, 10:44 | #42257 | Link |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,342
|
Check the settings of the profile groups, you can assign a keyboard shortcut there.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders Last edited by nevcairiel; 5th February 2017 at 10:57. |
5th February 2017, 11:17 | #42258 | Link |
Registered User
Join Date: Sep 2014
Posts: 280
|
ah ok, thx. Is there a madvr key table anywhere, to see what keys are free and which are used?
__________________
Intel i5 6600, 16 GB DDR4, AMD Vega RX56 8 GB, Windows 10 x64, Kodi DS Player 17.6, MadVR (x64), LAV Filters (x64), XySubfilter .746 (x64) LG 4K OLED (65C8D), Denon X-4200 AVR, Dali Zensor 5.1 Set |
5th February 2017, 12:07 | #42259 | Link | |||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
Quote:
For example, here are NNEDI3-256 and NGU pixart med at 300% zoom. Where does NNEDI3 look more natural here? The "M" in "TIME" looks better with NNEDI3-256. In every other part of the image, I personally prefer NGU pixart med. NNEDI3 also has a lot of very noticeable artifacts. And btw, NNEDI3-256 is about 3000% slower compared to NGU pixart med (!!!). Or let's look at a photo instead of a game screenshot. The low-res photo was downscaled using a box filter so it has more aliasing than usual. Here's again how NNEDI3-256 and NGU pixart med compare. I suppose I could offer even softer NGU pixart variants, but is it really needed? I posted a pixart image comparison with a very old very aliased game screenshot because that's really the hardest image type to get upscaled nicely. I think with real world images NGU and NNEDI3 look very similar. Would be great if you could do some real world comparisons. Quote:
I do not want to leave Polaris users behind, though, so I might keep super-xbr for the time being, just because it's probably still noticeably faster than NGU pixart for Polaris users. For comparison sake, here are the super-xbr 300% Mario zoomed image, and the clown zoomed by super-xbr. Look at the front wheel in the clown image. It's still quite aliased when using super-xbr. Less so with NNEDI3-256, and even less so with NGU pixart med. It's true that "medium" is a noticeable improvement over "low", but it's also a bit slower. There may be users whose GPU can perform "low" but not "medium". So why remove "low"? Is it really *that* terrible? If I compare it to NNEDI-16, IMHO NGU pixart "low" competes just fine. Or try comparing "low" to the regular (non-pixart) NGU variant! |
|||
5th February 2017, 12:07 | #42260 | Link |
Registered User
Join Date: Mar 2016
Posts: 160
|
New NGU med+ is definitely outperform sXBR in terms of quality and could replace it completely. NGU smooth also looks even a bit softer than NNEDI3.
However, right now it's hard to talk about speed because NGU smooth a) doesn't have high mode; b) seems doesn't have 4x mode. Personally I rarely use NNEDI3 and I have nothing to say in it's favor. But why it's so important to drop out NNEDI3 instead of drop out something like Spline? Additionally, it's still a big difference in sharpness between sXBR/NNEDI3/NGU smooth and plain NGU. Maybe there is a room for some intermediate NGU variant?
__________________
I'm infected with poor sources. |
Tags |
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling |
Thread Tools | Search this Thread |
Display Modes | |
|
|