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. |
10th August 2016, 12:48 | #39081 | Link | ||
Registered User
Join Date: Mar 2009
Posts: 3,645
|
Quote:
Quote:
|
||
10th August 2016, 17:36 | #39083 | Link |
Registered User
Join Date: Mar 2007
Posts: 934
|
No, newer does not necessarily mean better. Often it does but not always. Sometimes newer masters have more DNR or whatever that has a detrimental effect on the image.
__________________
TV Setup: LG OLED55B7V; Onkyo TX-NR515; ODroid N2+; CoreElec 9.2.7 |
10th August 2016, 18:34 | #39084 | Link |
Registered User
Join Date: Sep 2014
Posts: 280
|
I'm just doing some testings on my dropped frames problem since my Win10 update...
problem only occurs when display switches into 24hz mode. In 60hz mode the render queues are stable and no dropped frames. Did a clean windows install with complete new madvr setup, but problem still exists... Anyone else got this problem since anniversary update?
__________________
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 |
10th August 2016, 23:02 | #39086 | Link | ||
Registered User
Join Date: Mar 2009
Posts: 3,645
|
Quote:
Quote:
That doesn't stop me from answering the question. |
||
10th August 2016, 23:04 | #39087 | Link |
French Registered User
Join Date: Aug 2016
Location: France
Posts: 333
|
Hello,
I'm new on this forum and I'm French I would like to know if one day we can hope for a madVR "special encoding" ? I think a kind of filter like "ffdshow video encoder". It would be very cool, because the quality of madVR is top ! And the anti-ringing filter is very useful with Jinc filter. I tried with avisynth for encoding, but the ringing... it's too bad ! Thansk for your help ! (sorry for my simple English) P.S. : Your random questions are very hard ! |
10th August 2016, 23:26 | #39088 | Link | ||||||||||||||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
Quote:
Which means comparison screenshots between SuperRes strength 2 with and without AB are not very useful because one will be much softer than the other. So which one is better? It's impossible to judge. The only proper comparison can be to try to create 2 screenshots which have a roughly similar sharpness, where 1 screenshot achieves that with a higher SuperRes strength and AB turned on, while the other uses a lower SuperRes strength with AB turned off. If in this situation the SuperRes+AB screenshots looks objectively better, that would count as evidence for the usefulness of AB. Quote:
Just my 2 cents, of course. Quote:
Quote:
Instead of setting black level output to 16, you can set it to e.g. 16.25. Sounds weird. But with 10bit output this would be 65 instead of 64, so it does make sense. I thought about moving the control to 10bit. But most people would be confused with having a 64 value for black. So I decided to stick with 16/235, but allow fractions for 10bit support. This is rather complicated, it was discussed in great depth in the last couple of pages. Basically after tone mapping some pixels might be both too saturated and too bright, thus making them "out of gamut". The question is how to move them "in gamut"? We can either reduce luminance or saturation or both. Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Portable or not doesn't matter to madVR. The media player is the key. E.g. JRiver MC loads madVR from a subfolder without registering it and it works just fine. Also the madTPG test pattern generator loads madVR without needing it to be registered. The key problem to solve is that the media player must somehow locate madVR, and usually that's done through the DirectShow mechanism, which works through the registry. That's why madVR usually needs to be "installed" (registered in the registry). However, if the media player is able to locate the madVR in some other way (e.g. look in some specific path, or maybe in a subfolder with a specific name) then madVR doesn't need to be registered. So basically, this is something you need to ask the media player devs, not me. Quote:
Some Intel users seem to have problems with that. I haven't been able to reproduce it yet, unfortunately. You may have to stick to D3D9 and ordered dithering for now. |
||||||||||||||
10th August 2016, 23:29 | #39089 | Link | |||||||||||||||||||||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Compression isn't actually more accurate. It modifies a large amount of pixels away from their intended values. So one could argue that clipping is more accurate because it keeps more pixels at their exactly intended values. However, clipping loses details in highlight areas, which is most often worse than the accuracy loss caused by compression. Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Encoding support is not planned. However, maybe some day madVR could be used as a filter in AviSynth or VapourSynth. Probably not very soon, though. |
|||||||||||||||||||||
11th August 2016, 00:13 | #39090 | Link |
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
madVR v0.90.24 released
http://madshi.net/madVR.zip Code:
* HDR: improved tone mapping algorithm * HDR: added gamut mapping compression (instead of clipping) * HDR: slightly improved luminance vs saturation reduction formula * HDR: added options to use an external HDR tone mapping 3dlut * HDR: added option to do HDR processing, but still output PQ * fixed: measuring HDR luminance + NNEDI3 chroma upscaling sometimes crashed * compiled OpenCL kernels are now stored separately for 32bit and 64bit * when using the "ShowRenderSteps" tweak, a smaller OSD font is used now |
11th August 2016, 00:31 | #39091 | Link | |
French Registered User
Join Date: Aug 2016
Location: France
Posts: 333
|
Quote:
Thanks for your job madshi It's awesome ! P.S. : I really don't understand these %*$! random questions when I want to post a reply |
|
11th August 2016, 00:31 | #39092 | Link |
Registered User
Join Date: May 2012
Posts: 447
|
Okay, I'll try something higher then. I don't really understand why this is though - do the maths break down for these low display luminances? Should I be doing some sort of conversion to account for the fact that I'm sitting ~2 feet away from my monitor instead of looking at a TV from the sofa in my living room? I find it really hard to judge what the "right" brightness is for this material, so I'd love some sort of objective metric.
__________________
Test patterns: Grayscale yuv444p16le perceptually spaced gradient v2.1 (8-bit version), Multicolor yuv444p16le perceptually spaced gradient v2.1 (8-bit version) |
11th August 2016, 00:40 | #39093 | Link | ||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
Quote:
I'm not sure if the math breaks down. 120-180 nits looks quite ok with some demos, but not with all. Maybe Life of Pie is badly encoded? I've no idea. The same approach is valid here as always: Simply try different settings and pick what looks best to your eyes. |
||
11th August 2016, 01:28 | #39094 | Link |
Registered User
Join Date: Mar 2009
Posts: 3,645
|
It would seem that you are wrong.
The latest LG BD-XL drives work and apparently Leawo's blu-ray player can decode the discs. |
11th August 2016, 02:02 | #39095 | Link | |
Registered User
Join Date: Jul 2002
Location: Maryland
Posts: 52
|
Quote:
ERIC |
|
11th August 2016, 04:48 | #39096 | Link |
Visual Novel Dev.
Join Date: Nov 2015
Location: Bucharest
Posts: 200
|
@madshi Thank you for your great work!
New demo seems to work well for me, but I have not tested with HDR content yet. (lack of full movies in HDR at the moment) Quick advice for Optimus laptop users: Don't enable exclusive mode at all. With Optimus, it can cause a little worse experience, or at least it caused for me with most driver versions.
__________________
Visual Novels! https://www.facebook.com/seventh.heart.studios/ https://audiophile-heaven.blogspot.com/ https://www.youtube.com/watch?v=27VWU3ydRY8 |
11th August 2016, 05:22 | #39097 | Link | |
Registered User
Join Date: Oct 2015
Location: Brasil, SP, São Paulo
Posts: 154
|
Quote:
I m on optimus on two different laptops... One with gt 525m that hdmi outputs by nvidia and one with 930m that hdmi outputs by Intel hd 5500, on the two laptops fullscreen exclusive mode only improve perfomance and smoothnes of the vídeo, inclusive lowering a litros the render times... Same improvment on hdmi or laptop display... The two laptops with the latest drivers...
__________________
Desktop, i5 2500, 8GB, N570 GTX TF III PE/OC Asus X555LF, i7-5500U, 6GB Ram, Nvidia 930m/HD 5500 Windows 8.1 Pro x64 |
|
11th August 2016, 07:29 | #39098 | Link |
Registered User
Join Date: Feb 2014
Posts: 139
|
@madshi:
Nowhere in either intel or ATI graphics panel can I specify such a setting. I haven't bothered to dig through the registry, as I know the frame queue number is set and controlled via a reg key. The issue has persisted across various ATI driver updates. No idea why this is, but I'm glad it's fixed. Edit: I want to second uoppi's suggestion to add the old upscale refinement option as a trade performance tickbox. Upscaling 720p to 1080p is probably one of the most common upscaling factors, and it usually involves a double->downscale for best quality. With SR2, I can get render times of 33ms from 720p->1080p with SXBR->bicubic150, but I can't even use SR1 with the same settings without exceeding 45ms in the new implementation. As far as I know, there's no way to get the old behavior using profiles. Last edited by har3inger; 11th August 2016 at 07:42. |
11th August 2016, 07:44 | #39099 | Link |
UTAVATU
Join Date: Feb 2010
Location: Gdynia, Poland
Posts: 57
|
Dear Friends
After Anniversary Update for Windows 10 in MPC-HC + LAV Filters + madVR with Chroma Upscalling option set to NNEDI3 64 neurons, when watching movies and clicking CTRL+J, there is chroma > Jinc AR instead NNEDI3. When I choose other options everything is correct, only with NNEDI3 it shows Jinc AR. Do you also have such problem? Sincerely |
Tags |
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling |
Thread Tools | Search this Thread |
Display Modes | |
|
|