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. |
13th April 2014, 22:31 | #25921 | Link | |
Registered User
Join Date: Jan 2014
Posts: 93
|
Quote:
http://www.microsoft.com/en-us/download/details.aspx?id=3387 http://www.microsoft.com/en-us/download/details.aspx?id=5638 no change, same error... |
|
14th April 2014, 01:36 | #25922 | Link | |
Registered User
Join Date: Feb 2002
Location: San Jose, California
Posts: 4,406
|
Quote:
Some examples thinking about vertical resolution only through the same reasoning holds true to for horizontal. 480 -> 1080 source lines want to display on 2.25 lines of the destination. 480 -> 2160 source lines want to display on 4.5 lines of the destination. 4 out of 5 (80%) of the lines are the same values as the source at 2160 while only 2 out of 3 (66%) of the lines are the same at 1080. Of course the math doesn't really work like that, no original source pixels are left after either resize (except for NN which is why tp4tissue likes it), but the error in the scaling does decrease this way. Another way to phrase it; If you are trying to approximate an image it is easier to do if you have more pixels to use to do the approximation, the error at each pixel can be smaller. It doesn't matter if the image is a real image or a 720x480 set of pixels, more pixels still allow a better approximation. |
|
14th April 2014, 02:45 | #25923 | Link | |
Registered User
Join Date: Sep 2010
Posts: 10
|
Quote:
http://www.microsoft.com/de-de/download/details.aspx?id=26347 Last edited by vaporizer; 14th April 2014 at 03:18. |
|
14th April 2014, 03:14 | #25924 | Link |
Registered User
Join Date: Aug 2013
Posts: 71
|
Updated the survey/spreadsheet To be easier to input information. You can enter settings relatively painlessly now. It looks like you can't do too much with the forms, unless you created a whole separate google app, which is possible I guess.
Last edited by Procrastinating; 14th April 2014 at 03:18. |
14th April 2014, 04:19 | #25925 | Link | ||
Registered User
Join Date: Feb 2002
Location: San Jose, California
Posts: 4,406
|
Quote:
Edit: at your default set of settings my GPU declocks so I get... less than comparable rendering times. I am not sure how to deal with this except using default settings that do not down clock on anyones system. That means 200ms+ rendering times on some systems though so probably isn't a solution. Maybe include a field for GPU clock speed? Quote:
I wonder what is actually happening with SLI on? The second card gets some significant GPU usage but much less than the main card. 1280x720p24 -> 2560x1440 @ 72Hz, Bicubic75 AR chroma, Bicubic75 AR image, NNEDI3 128 Luma doubling, No smooth motion, no debanding, Ordered Dither, 3DLUT calibration, Windowed Overlay. SLI on 41.2ms* GPU0 81% @ 1097 MHz, 19% PCI-E, 7% memory controller GPU1 18% @ 836 MHz, 16% PCI-E, 0% memory controller SLI off 29.9ms* GPU0 67% @ 1097 MHz, 5% PCI-E, 7% memory controller GPU1 00% @ 324 MHz, 0% PCI-E, 0% memory controller *rendering times measured without nvidia inspector open but it didn't seem to affect them. GTX Titans, 3770K @ 4.6 GHz, Z77 chipset, each GPU is on PCI-E 3.0 x8, 32GB DDR3-2133CL9. Note: This is with Zoom Player forced to "Single-GPU" SLI rendering in the drivers, not that the setting seems to help vs nothing set for Zoom Player (forcing AFR rendering or something on Zoom Player flashes black frames trying to give you a seizure). edit: I cannot seem to change the SLI behavior anymore, windowed overlay, FSE, Old Windowed, or New Windowed. They all act the same no matter what I set in the drivers (except disabling or enabling SLI completely of course). This isn't bad per say as the only thing I could do before was break playback. Last edited by Asmodian; 14th April 2014 at 08:00. |
||
14th April 2014, 07:28 | #25926 | Link | |
Registered User
Join Date: Jan 2014
Posts: 93
|
Quote:
|
|
14th April 2014, 08:43 | #25927 | Link |
Registered User
Join Date: Aug 2013
Posts: 71
|
Added clocks and changed a few things, Google doesn't like moving columns linked to forms around, apparently. Learned regex.
What sort of default would you guys recommend, namely madshi? One which has sub-40ms render times for the high end say, and above-40 for low-end. Maximizing nnedi of course. Tears of steel seems like a pretty reasonable standard video. If all is decided tomorrow, I can reorganise the survey. Last edited by Procrastinating; 14th April 2014 at 11:11. |
14th April 2014, 15:07 | #25928 | Link |
Registered User
Join Date: Apr 2009
Posts: 1,019
|
480p content is 720x480, not 848x480. (and it should actually be scaled to 853x480)
A large amount of the SD content I watch is PAL, which would be 720x576. 40ms as a target would only be suitable for film content. I watch a lot of videos which require rendering times to be below 16ms. (60fps) Does 10-bit content put any additional strain on the GPU for scaling, or is it just more demanding because decoding has to be done in software? (thus irrelevant) I wouldn't know, as I thought that format was only used with pirated content. |
14th April 2014, 15:11 | #25929 | Link |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,340
|
It does not. Although since its twice the size, it might put additional load on the memory controller during the upload stage, which may however not be all that important.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
14th April 2014, 15:44 | #25931 | Link | |
Registered User
Join Date: Nov 2012
Posts: 167
|
Quote:
It might make sense to deinterlace/ivtc and scale the video to correct aspect ratio if you want to watch it on an older TV with built-in video player or a (cheap) hardware media player with very limited format support and you are re-encoding it anyway, since it was broadcasted at some crazy bitrate and you want to (re)watch it later. And once again it's perfectly legal in some countries. As for the 10-bit H264, the need to decode in software is the biggest difference, so perhaps a "DXVA-N/DXVA-CB/software decoder" option and always using 8-bit 4:2:0 content (most common) would be good for this tests. Once again it's not used for pirated content only, you might opt to use 10-bit while encoding your streams to save space or improve quality/size ratio. I hear it helps even with 8-bit sources. It is useful with computer-rendered movies/animations to avoid banding with lower bitrate. |
|
14th April 2014, 16:35 | #25932 | Link | |||
Registered User
Join Date: Apr 2009
Posts: 1,019
|
Quote:
Quote:
Quote:
As far as I can see, there is zero reason to enable chroma doubling though. |
|||
14th April 2014, 16:58 | #25933 | Link |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,340
|
I use NNEDI3 chroma scaling for 1080p content on a 1080p screen, just because the GPU is idle otherwise, but if I need to scale the image itself, I would prefer higher neurons of Luma doubling any day.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
14th April 2014, 17:10 | #25934 | Link | |
Registered User
Join Date: Oct 2012
Posts: 7,903
|
Quote:
Watching a 3 gb TV stream with 2:2/3:2/4:2:2:2 telecine deint by MadVR doesn't look nice at all reencoding. it to about 200-1000 mbsaves a lot of space. |
|
14th April 2014, 17:44 | #25935 | Link |
Registered User
Join Date: Dec 2008
Posts: 496
|
@madshi:
Blaire linked to your recent workaround (your changelog for 0.87.9) for the NV driver issue and asked me via PM, if there still is a driver fix needed. I kinda feared this would happen, since your woraround takes the pressure off of NV to fix an issue no one else (besides you and people that use NV hardware with madVR) seems to care about (that's how I interpret it). It looks to me that they were in the process of working on the fix, but they re-checked if they have the newest madVR version to test against. Now, what should I tell him? Some technical details would probably be helpful. Also how you (if memory serves right, a madVR user actually came up with the idea) worked around the bug, so NV knows where and what to search for. Last edited by iSunrise; 14th April 2014 at 17:53. |
14th April 2014, 20:25 | #25936 | Link | ||
Registered User
Join Date: Sep 2013
Posts: 919
|
Quote:
Quote:
I CAN see a small difference in chroma aliasing with NNEDI3 vs Lanczos, but the question is whether its worth the extra 30% GPU hit? I'm also using Ordered Dithering instead of Error Diffusion (another 30% GPU hit). I guess I prefer a cool running system (and a low electricity bill ) instead of squeezing every last drop of "almost invisible" performance from the system.
__________________
System: i7 3770K, GTX660, Win7 64bit, Panasonic ST60, Dell U2410. |
||
14th April 2014, 21:14 | #25937 | Link | |
Registered User
Join Date: Feb 2002
Location: San Jose, California
Posts: 4,406
|
Quote:
But this makes me think not. |
|
14th April 2014, 21:28 | #25938 | Link |
Kid for Today
Join Date: Aug 2004
Posts: 3,477
|
Humm, this page really makes a strong case for GPGPU on maxwell: Nvidia’s new Maxwell-powered GTX 750 Ti is hyper-efficient, quiet, a serious threat to AMD
Apparently miners are pretty happy with its low TDP, as they said in that url "GCN’s advantages in the GPU compute market are not so great that superior power consumption and per-core scaling can’t close the gap". I just saw a sale on this 760 but it's very hot and pretty sluggish with GPGPU apparently. FWIR in a local review, a slightly o/c 750Ti is only barely slower than my o/c 7850 in some games(saving 70W in the process) and some ppl claim that AMD's success in GPGPU was more a matter of luck with their choice of architecture than anything else, very eager to try that highly o/c 750Ti, hopefully it'll get here by the end of the week Last edited by leeperry; 14th April 2014 at 22:39. |
14th April 2014, 21:59 | #25940 | Link |
Registered User
Join Date: Apr 2014
Posts: 46
|
I've been using the 750 Ti since early March, and I feel like I have it pretty dialed in for my needs. I'd like to share my current MadVr configuration so folks out there can get an idea of it's capability, and determine if it's the right card for them.
In addition to MadVr, the GPU is also responsible for LavFilters via MPC-HC. Lav set to CUVID HW Deinterlacing, 50p/60p with High-Quality Processing enabled. I'm also using LumaSharpen as a post-resize shader. This is all under a native resolution of 1920x1080. Image upscaling: Lanczos 3 AR Image downscaling: Catmull-Rom AR+LL ChromaUpscaling: 1080p NNEDI3 32 neurons / Everything else Bicubic 75 AR Image Doubling: I will share my profile so you can see the range with which each profile operates: 1080p: 16 neurons (Not with NNEDI3 ChromaUpscaling, and it's not used often.) 720p < 24fps: 64 neurons 720p > 24fps: 32 neurons 720p > 30fps: 32 neurons SD < 24 fps: 128 neurons SD > 24 fps: 128 neurons SD > 30 fps: 32 neurons Code:
if (srcWidth >= 1900) and (srcHeight >= 721) "1080p" else if (srcWidth >= 956) and (srcWidth < 1900) and (srcHeight >= 500) and (deintFps <= 23.98) "720p < 24fps" else if (srcWidth >= 956) and (srcWidth < 1900) and (srcHeight >= 500) and (deintFps > 23.98) and (deintFps < 29.7) "720p > 24fps" else if (srcWidth >= 956) and (srcWidth < 1900) and (srcHeight >= 500) and (deintFps > 29.7) "720p > 30fps" else if (srcWidth <= 956) and (srcHeight <= 576) and (deintFps <= 23.98) "SD < 24fps" else if (srcWidth <= 956) and (srcHeight <= 576) and (deintFps > 23.98) and (deintFps < 29.7) "SD > 24fps" else if (srcWidth <= 956) and (srcHeight <= 576) and (deintFps > 29.7) "SD > 30fps" Dithering: ED, option 2 with both boxes checked Nothing checked under trade quality for performance I have my card overclocked to 1250Mhz, and boosts stable up to 1388Mhz. GPU Memory is at 1463. I've been following madvr for quite a long time now, and I've noticed a lot of conjecture towards where the GTX 750 Ti stands in terms of performance. Hopefully this helps reign in some of that conjecture. i7 3770k @ 3.7Ghz GTX 750 Ti Windows 8.1 64 bit Last edited by hu1kamania; 14th April 2014 at 22:03. |
Tags |
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling |
Thread Tools | Search this Thread |
Display Modes | |
|
|