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. |
17th November 2011, 08:44 | #11001 | Link |
Registered User
Join Date: Jun 2011
Posts: 288
|
Thanks madshi. Do you have any more details about this?
Does the bug also affect the other thing with having to enable the "limit rendering times to avoid glitches" or is it only related to the frame drops?
__________________
SETUP: Win 10/MPC-HC/LAV/MadVR HARDWARE: Fractal Design Node 804 | Xeon E3-1260L v5 | Supermicro X11SSZ-TLN4F | Samsung 2x8GB DDR4 ECC | Samsung 850 EVO 1TB | MSI GTX 1650 Super | EVGA G2 750 |
17th November 2011, 10:03 | #11002 | Link |
Registered User
Join Date: Dec 2007
Posts: 652
|
madshi, this version is working really well with JR's MC 17 Videoclock as madvr reports anticipated dropped/repeated frames swinging around by many hours or even days. Since the audio is slaved to madvr would it be worth reporting it as infinity or "slaved" in the OSD?
|
17th November 2011, 13:04 | #11003 | Link |
Registered User
Join Date: Nov 2010
Posts: 80
|
With this update, now my queues do not fill-up, thus I drop frames (unless I drop the scaling Algorithm to something low, like Bilinear.) I went back and forth between .78 and .79 multiple times and am confident that I can only get .78 to work.
I have always been pushing my graphics card[s] to the max with madVR (pair of NVidia 8600GT's in SLI, but with only 256MB of video-ram.) Thus I have my backbuffers/frames in advance set really low, to something like 2 or 3. This has worked in the past and allowed me to use 4-tap Spline in exclusive mode without drame-drops. However, as mentioned, .79 will not fill the queues anymore. I have been messing around with flushing settings (right now I have: don't flush/flush and Wait(sleep)/flush/flush, with "run presentation in seperate thread.") Is this the best for my situation? What settings can I try to squeeze every bit I can from my low-memory GPU's in order to continue using madVR? (Unfortunately I cannot afford a new GPU right now.) I have this issue with or without interlaced videos. However, like many others, I like madVR's new de-interlacer better than ffdShow's yadif. Of course, this puts even more strain on my GPU, but it still seems to work if I set de-interlacing in a different thread. Is this correct reasoning? I do not know what it is about .79 that puts it over the edge where I can no longer play without frame-drops...does anyone have any thoughts on that too? Thanks in advance for your help! MikeY PS - I am running software decoding of course instead of H/W decoding. (That is what has allowed me to get this far with my old GPU's!) |
17th November 2011, 14:11 | #11004 | Link |
Registered User
Join Date: Sep 2006
Posts: 2,197
|
no, atm not. dont know why or whats different. but it seems to be fine atm.
__________________
Laptop Lenovo Legion 5 17IMH05: i5-10300H, 16 GB Ram, NVIDIA GTX 1650 Ti (+ Intel UHD 630), Windows 10 x64, madVR (x64), MPC-HC (x64), LAV Filter (x64), XySubfilter (x64) (K-lite codec pack) |
17th November 2011, 16:03 | #11005 | Link | ||
Registered User
Join Date: May 2009
Posts: 212
|
Quote:
It would show the very similar issue I described in below. I have retried the normal even-field-first MPEG-2 1080i60 contents to HDMI 1080i60 output by madVR (w/o deinterlace) on GTX260+. It looks like the presented field sequence is still incorrect by GPU. The problem is very obvious when the picture contains some moving objects. If the video contents' presentation order are Frame0[E0 O1] Frame1[E2 O3] Frame2[E4 O5] ... The field signal sent to HDMI should be something like: E0 O1 E2 O3 E4 O5 ... On the 1080i PDP TV, howerver, I observed it often looks like: E0 O1 E0 O3 E4 O5 ... That is one reason why I keep using frame-rate-doubing deinterlace at 1080i50 / i60 output signals. At least HDMI 1080i signal always contains all right fields. Why I do not use 1080p50 / 1080p60 signals on my PDP TV. The reason is simple. The TV will perform some anti-flicking process if the input signal is 1080p24/p25/p30/p50/p60. Thus the static objects on the 1080i video's will have a slight vertical image resolution loss. Quote:
1. An interlaced frame is vertically scaled by frame to either interlaced or progressive video signal. 2. An interlaced frame is shown as still image on interlaced output device (CRT / Hitachi PDP) 3. Playback frame / field rate is different from output signal. (Tricky play) Method1: Cuda deinterlace http://img706.imageshack.us/img706/9...ncudadeint.png Method2: madVR DXVA2 deinterlace http://img257.imageshack.us/img257/9...dxva2deint.png I did the comparision of single frame by two playback methods. To be honest, the difference is very minor even through I did not capture the exact time frame. Thus the difference could come from the fact that Japanese HDTV contents have very different bit rate distribution among MPEG-2 I / P / B pictures. Yet I am sure that my Hitachi PDP can show somewhat magic nartural human skin colors and 3D-alike stereo effect on those well-captured + well-encoded high bit-rate video contents. Have played those contents on my friend's Panasonic FullHD PDP, it does not show me the same natural color / stereo feeling which I have seen on my TV. Last edited by pie1394; 17th November 2011 at 16:12. |
||
17th November 2011, 17:38 | #11006 | Link |
Registered User
Join Date: Jan 2009
Posts: 625
|
Hi Madshi,
Regarding my timestamp issue with index.bdmv and mpls files, I've found an MKV that plays in exactly the same way (hundreds of dropped frames). The MKV was made with MakeMKV straight from disc and is a music concert 1080i 29.97fps. Log file is here: http://www.mediafire.com/?oo6676cjrouncna |
18th November 2011, 13:34 | #11007 | Link |
Registered User
Join Date: Aug 2008
Posts: 176
|
madshi
Another strange beast for your zoo . http://dl.dropbox.com/u/43376972/PAL...0artifacts.mkv PAL SD movie in .mkv. If it is DXVA deinterlaced, we have a lot of interlacing artefacts. And "DXVA deinterlacing off" (plain weaving?) does not help. |
18th November 2011, 14:40 | #11008 | Link | |
Software Developer
Join Date: Jun 2005
Location: Last House on Slunk Street
Posts: 13,248
|
Quote:
I would suggest you "repair" that clip prior to playback. You may try something like this in Avisynth: Code:
FFVideoSource("PAL film deinterlacing artifacts.mkv") NNEDI3(field=-2) Merge(Last.SelectEven(), Last.SelectOdd()) Code:
FFVideoSource("PAL film deinterlacing artifacts.mkv") SeparateFields() Merge(Last.SelectEven(), Last.SelectOdd()) NNEDI3(dh=true)
__________________
Go to https://standforukraine.com/ to find legitimate Ukrainian Charities 🇺🇦✊ Last edited by LoRd_MuldeR; 18th November 2011 at 15:49. |
|
18th November 2011, 16:01 | #11009 | Link |
Registered User
Join Date: Sep 2009
Location: Europe
Posts: 5
|
madshi, just to clear things out
"use a separate device for presentation" option give any good on xp64 sp2 or can be safely disabled, because i see it says vista and win7 only ? and what about "present queue" also on xp64 sp2 ? is it possible to make it > "3" ? even setting in driver to "4" still limits it to "3"... |
18th November 2011, 16:17 | #11010 | Link | |
Registered User
Join Date: Apr 2009
Posts: 1,019
|
Quote:
I'm currently running the latest Nvidia beta drivers (seems like this is something I'll have to keep doing as a gamer) and was concerned they were the cause, but lowering the queues to 16 results in glitch-free playback. If I lower the GPU queue to 16 but leave the CPU queue at 32 playback is fine initially with no presentation glitches, but at random it will suddenly start dropping a significant number of frames. So with Nvidia, at least when running the latest beta drivers, I wouldn't recommend going over 16 at all. I suspect that whatever causes this is the same thing that Nevcariel found when going over 12 queues on his system in previous versions. I also encountered a film where madVR decided that deinterlacing was necessary (it wasn't) and that caused it to drop massive numbers of frames. Will see if I can get a sample later, been quite busy lately. I also had a problem with the OSD not showing up, and "uninstalling" & "reinstalling" fixed this. All I did was run the two batch files though. I suspect this was another issue caused by running J-River MC17 and uninstalling it. (same thing as ffdshow becoming unregistered) I guess they have a build or preference that lets them disable the OSD? |
|
18th November 2011, 16:42 | #11011 | Link |
Registered User
Join Date: Nov 2010
Posts: 57
|
madVR doesnt work with Intel IGED Drivers
I am forced to use the Intel IGED Drivers to add a custom Interlaced Resolution for my TV.
madVR works with the Intel GMA Drivers but refuses to work with the IGED Drivers. One thing i noticed that my custom Resolution is reported as 25 Hz Interlaced with the IGED Drivers but under the GMA Drivers interlaced resolutions are reported as 50 Hz Interlaced. D3D games working fine. Is this a bug in madVR ? Is there hope you can fix it ? (I cant live without madVR ) thanks Last edited by italospain; 18th November 2011 at 16:45. |
18th November 2011, 19:17 | #11012 | Link | ||
Registered User
Join Date: Aug 2008
Posts: 176
|
Quote:
Mediainfo reports for that stream: Codec ID : V_MPEG2 Width : 720 pixels Height : 576 pixels Frame rate : 25.000 fps Standard : PAL Scan type : Progressive Color primaries : SMPTE 170M Transfer characteristics : SMPTE 170M ??? Matrix coefficients : SMPTE 170M ??? So progressive frames are in that stream and Quote:
|
||
18th November 2011, 20:58 | #11013 | Link |
Registered User
Join Date: Jun 2006
Posts: 452
|
Dropped frames / Vsync
Madshi,
On certain files I have a lot of dropped frames. Because I have fast hardware (core i7 -970 with GTX-570, W7 SP1), I assumed it is not a performance issue. The drops also happen if a switch all scaling to bilinear + lower quality settings. I started logging and this messages were often repeated in the log : Code:
00092845 00161358 Render drop frame 9246, no free vsync slot, plannedVSync: 4121 00161358 Render drop frame 9247, no free vsync slot, plannedVSync: 4122 00167112 VSync drop frame 9570, can't delay, plannedVSync: 4448, present vsync: 4467, next frame's vsync: 4449 00167112 Upload drop frame 9578, can't be rendered in time, plannedVSync: 4456, timeLeft: -186891, vsync interval: 16666 00167112 Upload drop frame 9579, can't be rendered in time, plannedVSync: 4457, timeLeft: -170265, vsync interval: 16666 00167112 Upload drop frame 9581, can't be rendered in time, plannedVSync: 4459, timeLeft: -137000, vsync interval: 16666 00167112 Upload drop frame 9583, can't be rendered in time, plannedVSync: 4461, timeLeft: -103738, vsync interval: 16666 00167112 Upload drop frame 9585, can't be rendered in time, plannedVSync: 4463, timeLeft: -70481, vsync interval: 16666 00167112 Upload drop frame 9587, can't be rendered in time, plannedVSync: 4465, timeLeft: -37213, vsync interval: 16666 00167112 Upload drop frame 9589, can't be rendered in time, plannedVSync: 4467, timeLeft: -3946, vsync interval: 16666 00167113 Upload drop frame 9571, no free vsync slot, plannedVSync: 4449 00167113 Upload drop frame 9572, no free vsync slot, plannedVSync: 4450 00167113 Upload drop frame 9573, no free vsync slot, plannedVSync: 4451 00167113 Upload drop frame 9574, no free vsync slot, plannedVSync: 4452 00167113 Upload drop frame 9576, no free vsync slot, plannedVSync: 4454 00167113 Upload drop frame 9575, no free vsync slot, plannedVSync: 4453 00176248 Render drop frame 9598, can't be rendered in time, plannedVSync: 3766, timeLeft: -20821486, vsync interval: 16666 00176249 Render drop frame 9599, can't be rendered in time, plannedVSync: 3767, timeLeft: -20804860, vsync interval: 16666 00176249 Render drop frame 9600, can't be rendered in time, plannedVSync: 3768, timeLeft: -20788236, vsync interval: 16666 00176249 Render drop frame 9601, can't be rendered in time, plannedVSync: 3769, timeLeft: -20771616, vsync interval: 16666 00176249 Render drop frame 9602, can't be rendered in time, plannedVSync: 3770, timeLeft: -20754994, vsync interval: 16666 00176249 Render drop frame 9603, can't be rendered in time, plannedVSync: 3771, timeLeft: -20738364, vsync interval: 16666 00176249 Render drop frame 9605, can't be rendered in time, plannedVSync: 3773, timeLeft: -20705102, vsync interval: 16666 00176249 Render drop frame 9607, can't be rendered in time, plannedVSync: 3775, timeLeft: -20671834, vsync interval: 16666 00176249 Render drop frame 9610, can't be rendered in time, plannedVSync: 3778, timeLeft: -20621902, vsync interval: 16666 00176249 Render drop frame 9612, can't be rendered in time, plannedVSync: 3780, timeLeft: -20588606, vsync interval: 16666 00176249 Render drop frame 9594, no free vsync slot, plannedVSync: 3762 00176249 Render drop frame 9595, no free vsync slot, plannedVSync: 3763 00176249 Render drop frame 9596, no free vsync slot, plannedVSync: 3764 00176249 Render drop frame 9597, no free vsync slot, plannedVSync: 3765 00176249 Render drop frame 9609, no free vsync slot, plannedVSync: 3777 00177763 Render drop frame 9617, no free vsync slot, plannedVSync: 69 00177763 Render drop frame 9618, no free vsync slot, plannedVSync: 70 --------------------------------------------------------------------------------------------------------------------------- 00041624 Upload drop frame 2119, can't be rendered in time, plannedVSync: 1936, timeLeft: 17220, vsync interval: 19992 00042140 VSync drop frame 2132, can't delay, plannedVSync: 1960, present vsync: 1961, next frame's vsync: 1961 00042460 VSync drop frame 2146, can't delay, plannedVSync: 1976, present vsync: 1977, next frame's vsync: 1977 00042780 VSync drop frame 2160, can't delay, plannedVSync: 1992, present vsync: 1993, next frame's vsync: 1993 00043100 VSync drop frame 2174, can't delay, plannedVSync: 2008, present vsync: 2009, next frame's vsync: 2009 00043740 VSync drop frame 2202, can't delay, plannedVSync: 2040, present vsync: 2041, next frame's vsync: 2041 00044379 VSync drop frame 2230, can't delay, plannedVSync: 2072, present vsync: 2073, next frame's vsync: 2073 00045019 VSync drop frame 2258, can't delay, plannedVSync: 2104, present vsync: 2105, next frame's vsync: 2105 00045660 VSync drop frame 2286, can't delay, plannedVSync: 2136, present vsync: 2137, next frame's vsync: 2137 00046300 VSync drop frame 2314, can't delay, plannedVSync: 2168, present vsync: 2169, next frame's vsync: 2169 00046620 VSync drop frame 2328, can't delay, plannedVSync: 2184, present vsync: 2185, next frame's vsync: 2185 00046939 VSync drop frame 2342, can't delay, plannedVSync: 2200, present vsync: 2201, next frame's vsync: 2201 00047579 VSync drop frame 2370, can't delay, plannedVSync: 2232, present vsync: 2233, next frame's vsync: 2233 00047667 Render drop frame 2375, can't be rendered in time, plannedVSync: 2237, timeLeft: -7939, vsync interval: 19992 00047680 Upload drop frame 2377, can't be rendered in time, plannedVSync: 2239, timeLeft: 18509, vsync interval: 19992 00047808 Render drop frame 2381, can't be rendered in time, plannedVSync: 2245, timeLeft: 10524, vsync interval: 19992 It seems it most happens in Window mode (Aero on), much less in exclusive mode. Also if interlace is active, it seems to happen more often, but it could also be the content itself. I can't reproduce it every time, it seems to happen sometimes. My buffers settings are : decoder=20, GPU=10, backbuffer =10, and they are filled to about max-2/(max-1 or max), just the backbuffer goes to 0 if the frame gets deleted from the queue. The GPU load < 35 %, CPU load < 5%. I use Lav-filters to decode, player is MPC-HC and MadVR v0.79. System : Core i7 6+6 cores @3.6 Mhz / 24 GB RAM @ 1600 Mhz Win7 x64 including latest updates (auto-update) / Aero ON NV drivers : latest WHQL Lav filters 0.39 + MadVR v0.79 with latest MPC-HC |
18th November 2011, 23:52 | #11014 | Link | |
Software Developer
Join Date: Jun 2005
Location: Last House on Slunk Street
Posts: 13,248
|
Quote:
But they can't tell you the nature of the content! Progressive content can be encoded as interlaced, which costs some compression efficiency, but nevertheless often happens in reality. Encoding interlaced content as progressive will "destroy" it. Having said that, what we see in your case is progressive content. Just "bad quality" progressive content - which, apparently, once was created from an interlaced source. It's impossible to know, how they managed to screw up the content. And, unfortunately, it's impossible to simply "revert" the process. As long as we don't have access to the original "interlaced" source, we can only try to "repair" the mess as good as possible. A "normal" bobbing deinterlacer is not the right tool in this case though*. And, as this looks like some kind of "line doubling" has been applied, my quick&dirty AVS script interpolates a full frame from each field, then blends the two results together. (*) A dumb "skip field" deinterlacer might work in a way for this case, but would simply discard 1/2 of the lines.
__________________
Go to https://standforukraine.com/ to find legitimate Ukrainian Charities 🇺🇦✊ Last edited by LoRd_MuldeR; 19th November 2011 at 00:18. |
|
19th November 2011, 00:05 | #11015 | Link | |
Registered User
Join Date: Sep 2011
Posts: 1
|
Quote:
|
|
19th November 2011, 10:10 | #11017 | Link |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,347
|
Hi madshi,
i've been trying to figure out why my HTPC stopped going into FSE mode, but i couldn't find anything the log regarding a window covering the display or anything like that. Maybe you can see anything in it? http://files.1f0.de/madVRnofse.zip Edit: Reinstalling MC17 seems to have fixed it for now, but if you still see something in the log, maybe it helps avoid the problem in the future.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders Last edited by nevcairiel; 19th November 2011 at 11:44. |
19th November 2011, 15:00 | #11018 | Link |
Registered User
Join Date: Mar 2011
Posts: 90
|
i just assumed my pc wasnt fast enough do use madvr with 1080 stuf but yesterday i loaded lav video and it works. why is this when i dont have a nvidia card?
also if anyone can help i cant get reclock to work with lav cplitter /audio it gives an error. can someone give me the right settings for lavfilters and madvr? cheers |
19th November 2011, 15:33 | #11019 | Link |
Registered User
Join Date: Jan 2010
Posts: 76
|
Hi madshi,
first of all: Great work you are doing with madVR! At least since the latest build 0.79 I am getting very often a black screen when starting a video in MPC-HC. I already tried to play around with the madVR options, but nothing changed the situation. The sound of the video is audible, but I do not see the picture. If I press CTRL+ALT+DEL the video sometimes appears (sometimes have to press "Switch to" before), but sometimes MPC-HC closes immediately. I prepared a log, where I opened MPC-HC in fullscreen, the picture was black and MPC-HC closed after pressing CTRL+ALT+DEL. http://www.mediafire.com/?a538o2rial1gxpl Maybe this log helps and I would be glad if you could find the reason for this behaviour. |
19th November 2011, 16:47 | #11020 | Link | |
Registered User
Join Date: Nov 2006
Posts: 799
|
Quote:
|
|
Tags |
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling |
|
|