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. |
24th June 2017, 04:07 | #44181 | Link |
Registered User
Join Date: Dec 2016
Posts: 65
|
hi madshi , The authors of SRCNN have published another paper about accelerating SRCNN(FSRCNN).now it can run in real-time performance
http://mmlab.ie.cuhk.edu.hk/projects/FSRCNN.html |
24th June 2017, 13:58 | #44183 | Link |
*****
Join Date: Feb 2005
Posts: 5,647
|
For those with performance issues:
First reset the settings. That will undo all silly performance killing changes that you may have made, like supersampling. Press Ctrl+J to display the OSD. Look at the details that are shown and first check if the algorithms that are used match with what you expect. You can show extra details by creating an empty folder called ShowRenderSteps in the installation folder of MadVR. Post a screenshot if you need help. |
24th June 2017, 20:56 | #44184 | Link | |
Registered User
Join Date: Nov 2016
Posts: 15
|
Quote:
There is also ESCPN or VESCPN specifically designed for video upscaling. But there are either patent issues or knowledge issues on madashis site, so an implementation is very unlikely. If you look for an very good upscaler (albeit a specialized one) than there is enhancenet https://arxiv.org/abs/1612.07919 , which might be feasible in the future. But now it is not realtime |
|
24th June 2017, 22:02 | #44185 | Link | |
Registered User
Join Date: Jan 2009
Posts: 625
|
Quote:
The weird thing is, for me, when the queues just drop, pausing the video and letting the queues refill and then unpausing it fixes it (until the next time they drop). If i dont pause and then unpause the queues stay empty and continues dropping gazillions of frames infinitely (well ok to the end of the video). I've even made changes so that render times stay between 10-15ms so lots of headroom. So i'm stumped. |
|
25th June 2017, 09:25 | #44188 | Link |
Quality Checker
Join Date: Aug 2013
Posts: 285
|
when i turn on Exclusive Mode and exit i noticed madvr reset my calibration setting in the system
even when i set [disable calibration controls for this display] on and [disable GPU gamma ramps] off sorry for my english
__________________
I love Doom9 |
25th June 2017, 09:45 | #44189 | Link |
Registered User
Join Date: Oct 2012
Posts: 7,926
|
like this: http://bugs.madshi.net/view.php?id=290 ?
|
26th June 2017, 00:02 | #44193 | Link | |
Quality Checker
Join Date: Aug 2013
Posts: 285
|
Quote:
__________________
I love Doom9 |
|
26th June 2017, 13:44 | #44196 | Link |
Registered User
Join Date: Jul 2002
Posts: 35
|
Hey, I just found out this week end that to be able to use the HDR metadata passthru with MPC-HC, I have to use the 32bit version. I have always used the 64bit... this might have been tell before in this thread and I've misse it probably...
anyway since the switch, I have been able to test out both the Nvidia API as well as the Windows 10 API and both work pretty well. On one occasion the HDR flag stayed up after exiting the movie and I couln't find any way to reset it exept rebooting my PC. But overall , works nicely. |
26th June 2017, 18:16 | #44197 | Link |
Registered User
Join Date: Sep 2008
Posts: 21
|
When playing a source video in DXVA2 copy-back mode with BT.2020 matrix & primaries and passthrough HDR, Mad-VR says "HDR 1000 nits, BT.2020 -> DCI-P3".
Should I select DCI or BT.2020 in my TV (X850C) menu settings? With BT.2020 enabled the colors look deeper without oversaturation. Is there a setting in MadVR to disable the BT.2020 -> DCI-P3 conversion (DXVA2 native mode, perhaps?) Thanks for any help! Last edited by Damie; 26th June 2017 at 18:58. |
26th June 2017, 19:37 | #44198 | Link | ||
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,348
|
Quote:
So that line is just informationl. Quote:
At this point in time, perhaps using DCI-P3 would be best, although BT.2020 would also work (but if you're not using 10-bit output, it may cause a bit of loss).
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
||
26th June 2017, 19:45 | #44199 | Link |
Registered User
Join Date: Dec 2016
Posts: 14
|
Are my colors wrong?
Hey there,
I encountered this pack containing some test videos and some of them caught my interesting: https://drive.google.com/file/d/0B9F...p1eGVnTG8/edit When I play "MP4-2c\Basic Settings\3-White Clipping.mp4" what I see on the screen matches what the video says: 235 and higher are completely white, while 230-234 are slightly gray (refer to the image), but shouldn't I see whole range up to 253? I tried checking colors in Paint and 230 column is actually 249 white, while 234 column is 254 white. Why is this considered to be correct output when colors are incorrect? Similar issue with "1-Black Clipping.mp4" file - dark colors are darker than the number says. |
26th June 2017, 19:48 | #44200 | Link | |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,348
|
Quote:
Note that when you take a screenshot of the video, it is already after various levels of processing, and the values don't match the actual values in the video.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
|
Tags |
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling |
|
|