View Full Version : madVR - high quality video renderer (GPU assisted)
clsid
8th April 2011, 16:11
BTW, is it ok that MPC still loads VSFilter along with madVR? As you've implemented internal subtitles support there is no more need in VobSub.
There is code in MPC that automatically blocks use of vsfilter when the internal sub renderer is enabled and a compatible video renderer is used. Madshi needs to add his renderer to that list.
madshi
8th April 2011, 16:33
Jetico Firewall's HIPS detects it on playback start. I'm not familiar with WinAPI, but I guess it can be triggered by something like WSAStartup().
Hmmmm... And LAN is definitely disabled in the madVR tray icon menu? Can anybody else reproduce this?
There is code in MPC that automatically blocks use of vsfilter when the internal sub renderer is enabled and a compatible video renderer is used. Madshi needs to add his renderer to that list.
Do you happen to know where that code is in MPC? Thanks!
Andy o
8th April 2011, 17:24
In case you prefer to bitstream audio (as is) you can't use fps adaptation, that makes using ReClock pointless.Reclock at least gives you an option now if you prefer to have video frame errors or audio errors. You can also apply video adaptation with bitstreaming, but then audio will suffer. Sometimes it's not noticeable.
But also, there are some instances where there are problems with bitstreaming (some TMT versions and possibly PowerDVD too, with some ATI driver versions IIRC) where bitstreaming doesn't work unless you're using ReClock. Using ReClock also fixes some titles with MPC-HC that otherwise don't bitstream (silent) on my setup.
Abradoks
8th April 2011, 17:26
And LAN is definitely disabled in the madVR tray icon menu?
Yeah. And I've tried with default settings, and tried to enable/disable LAN support, etc. There is no real network activity from
madVR with disabled LAN support, but it still triggers HIPS.
Do you happen to know where that code is in MPC? Thanks!
Should be \src\apps\mplayerc\FGManager.cpp:2207
Thunderbolt8
8th April 2011, 17:26
IMO there is a sence to use reclock in combination with madvr if audio output is set to PCM (not bitstream). This way you have perfect fps adaptation for video and WASAPI/KS output for sound. In case you prefer to bitstream audio (as is) you can't use fps adaptation, that makes using ReClock pointless. Note, you can set ReClock to convert PCM to AC3 on the fly after fps adaptation (the only way to use S/PDIF bitstream in this case).
thats the way I use reclock as well, to bitstream PCM to my receiver. therefore, I wonder why you guys still seem to need it for video when madvr is doing all the rest. thats why disabling reclocks video renderer in order to avoid all reclock problems right from the start.
pankov
8th April 2011, 19:49
Can someone explain what exactly should I do to be sure that I've disabled Reclock's video renderer.
I think I've done it but just want to be sure.
nevcairiel
8th April 2011, 19:57
ReClock still is useful with madVR if you use PCM output. It adjusts the reference clock to be in sync with the video clock, so your video refresh rate matches the screen refresh perfectly - and then resamples the audio to make it fit the new refresh rate. People might argue that ReClock is not the perfect place to fix the video timings, but its currently the only way to really get perfectly fluid playback for me (my GFX card can not do 100% perfect 23.976)
SamuriHL
8th April 2011, 20:05
Does it show up in the filters menu of MPC-HC when you're playing a video?
chogster
8th April 2011, 20:47
For what it's worth on the HAM issue. I've had another play & can confirm that frames are only dropped during playback of 1080 material, dvd playback is fine (on my system anyway). I've tried a few 1080 25fps films with refresh rate set to 50Hz & frames are still dropping. Baring this evidence mind, hopefully a fix wouldn't prove too difficult.
Thunderbolt8
8th April 2011, 21:18
Can someone explain what exactly should I do to be sure that I've disabled Reclock's video renderer.
I think I've done it but just want to be sure.
I think I mixed something up :eek:
what you can actually do is to bypass the audio resampler which is always active and modifies the audio. you need to do the following:
open regedit and go to "COMPUTER\HKEY_CURRENT_USER\Software\Reclock\Config"
right click to add a new value, select "DWORD (32 bit)" and name it:
BypassResampler
double click the value, and set it to "1".
there you can also check whether bypassdefault renderers is set to 1. if not, go advanced tab in reclock configuration and set the following:
forced feclock to be loaded instead of default renderers - checked
only load reclock and show application dialog for the following applications - checked
add MPC-HC and set to "Load Always"
but afaik, in the end it looks to be the same. when you now set those different fps rates in reclock properties tab while playback, it has no effect any more (also tick slave reference clock to audio if you have to and tick lock original speed).
Andy o
8th April 2011, 21:22
Why are you going to the registry when the options are right there in the Reclock settings panel?
It adjusts the reference clock to be in sync with the video clock, so your video refresh rate matches the screen refresh perfectly
Thanks, i couldn't explain it better in english.
Does it show up in the filters menu of MPC-HC when you're playing a video?
No, in case of bitstream (packed streams) it's just doesn't load.
I think it's wrong place to discuss ReClock though.
Thunderbolt8
8th April 2011, 21:39
Why are you going to the registry when the options are right there in the Reclock settings panel?afaik the bypassresampler thing is not to be found via settings.
Mark_A_W
9th April 2011, 05:55
I doubt you'll hear a difference if you set reclock to upsample, rather than just resample around 48khz.
It may even sound better ;)
A lot of worrying about nothing IMO.
Andy o
9th April 2011, 09:48
I agree about audible differences, but was just saying that you can disable resampling (and media adaptation) by setting PCM output "same as input", and then while playing just clicking on the clock icon, then "slave reference clock to audio" and set media adaptation at "original speed" and have it locked so it sticks. I might be missing something?
Thunderbolt8
9th April 2011, 10:36
seems right to me. also got same as input (and format 24 bit int padded to 32, accept bitstream formats and disable media speed correction with bitstream audio) and then it all gets forwarded to the receiver without windows or reclock doing any modification.
disable media speed correction with bitstream audio) and then it all gets forwarded to the receiver without windows or reclock doing any modification.
The only benefit here is bit perfect PCM output. IMO in movies or even live misuc shows perfect video quality (perfect vsync) is more important. No matter I've spent 2-3 times more money on sound system than on TV... OK, it depends of content though.
Plutotype
9th April 2011, 12:29
ReClock still is useful with madVR if you use PCM output. It adjusts the reference clock to be in sync with the video clock, so your video refresh rate matches the screen refresh perfectly - and then resamples the audio to make it fit the new refresh rate. People might argue that ReClock is not the perfect place to fix the video timings, but its currently the only way to really get perfectly fluid playback for me (my GFX card can not do 100% perfect 23.976)
Agree. Newcairel and folks, one more question to the A/V synchoronisation theme. If Reclock ( media adaptation mode set to auto) says I have 23.976 video stream and my video hardware refresh rate is 23.975 (ATI5670 ), why Reclock is not syncing them @23.976 but instead I see:
"CINEMA adaptation: media speed changed to 24.000 fps. Audio resampled at 48048 Hz."
Does this mean, is it in such Reclock behaviour better to set 24hz refresh rate in ATI Catalyst?
Thanks
Does this mean, is it in such Reclock behaviour better to set 24hz refresh rate in ATI Catalyst?
You can just select more relevant refresh rate manually from list.
namaiki
9th April 2011, 12:43
"CINEMA adaptation: media speed changed to 24.000 fps. Audio resampled at 48048 Hz."
Does this mean, is it in such Reclock behaviour better to set 24hz refresh rate in ATI Catalyst?
Thanks
The message is misleading. It is synced to a multiple of your monitors refresh rate closest to 24fps. (in this case, 23.975)
However, make sure that the refresh rate of your screen is being detected correctly in ReClock filter properties.
http://img855.imageshack.us/img855/8885/dsfdajfdkfj.th.jpg (http://img855.imageshack.us/img855/8885/dsfdajfdkfj.jpg)
noee
9th April 2011, 12:44
You can just select more relevant refresh rate manually from list.
No, no, no. This is a known "display issue" with Reclock. The message Reclock shows there is not entirely accurate (though it is doing it's magic). You guys need to go over to the Slysoft Reclock forum and read up a little bit.
Andy o
9th April 2011, 12:48
That's right. Auto in media adaptation is best if you want Reclock to do what it's supposed to do. You can check the actual audio sampling rate where it says (I think) "audio clock" to the left.
But yeah, we should take this to the ReClock forum (http://forum.slysoft.com/forumdisplay.php?f=85).
Plutotype
9th April 2011, 12:50
Hi all,
As every LCD TV has small or more significant input lag, following question evolves: Does the video, being played on such LCD TV with PCM/Reclock adaptation setup guarantees, that there is no need for audio delay correction because of the input lag?
I have done a test with madVR running parallely in windowed mode on my DELL2311H and Sony EX500 at once ( doubled desktop ). I have not idea how to measure the lag exactly, but SONY TV was of course a very little behind the DELL LCD Monitor in terms of picture frame rendering times. I have read on avsforum that SONY EX500 has approx. a generic 30ms input lag. Should I use audio delay setting in MPC-HC to match this?
Thanks
namaiki
9th April 2011, 12:53
You might be able to equal the delay for audio with video if you pass it though your TV as well.
sneaker_ger
9th April 2011, 12:55
30 ms? That's unnoticeable for humans.
I had to do some 400ms correction for a Samsung TV.
pankov
9th April 2011, 13:10
Plutotype,
here is a tool to measure your input/display lag
http://www.flatpanelshd.com/focus.php?subaction=showfull&id=1229335064
Set your video card in "Clone mode" and take a picture (with a decent camera, not the average smartphone camera) and you'll see what's the difference between your two displays. To measure an absolute value you'll probably have to do the same test and compare against a CRT display, because even your DELL has a minimal amount of lag.
Plutotype
9th April 2011, 13:24
Plutotype,
here is a tool to measure your input/display lag
http://www.flatpanelshd.com/focus.php?subaction=showfull&id=1229335064
Set your video card in "Clone mode" and take a picture (with a decent camera, not the average smartphone camera) and you'll see what's the difference between your two displays. To measure an absolute value you'll probably have to do the same test and compare against a CRT display, because even your DELL has a minimal amount of lag.
Thanks Pankov, will try do the tests.
Anyway, sorry to all for questioning all this stuff here and thanks for your input.
leeperry
9th April 2011, 13:28
Does the video, being played on such LCD TV with PCM/Reclock adaptation setup guarantees, that there is no need for audio delay correction because of the input lag?
I use headphones, and I can detect 5ms A/V delay. The input lag of my DLP pj isn't worth mentioning.
OTOH, when using speakers it's more complicated because there will be a 3ms delay per meter between the speakers and your ears. And apparently, some flat screens do have a pretty big input lag...so find a movie scene with a lot of tight and clear noises, and sync them to your ears?
Some lips close-up might be good too.
Thunderbolt8
9th April 2011, 13:41
I'd say no one can detect a 5ms A/V delay. its all just chance or imagination.
leeperry
9th April 2011, 13:46
no one can detect a 5ms A/V delay. its all just chance or imagination.
and who cares about what you think again :confused:
using headphones, it's very real...because there's no delay between the transducer and the eardrum.
anyway 1) it's very much OT 2) I've got nothing to prove to anyone and 3) I was merely answering to someone else.
if you're sitting 10 meters away from your speakers, you already have a 30ms A/V desync de facto.
pankov
9th April 2011, 13:51
guys,
does anybody of you have a short sample that I can use to test A/V sync?
I'm talking about the ones that show something like a clock face with big numbers and beeps at the 00 (12 o'clock) mark.
I've seen such in the beginning of some movies but never got my hands on one
madshi,
I apologize for the off topic talk but I guess we are (or at least I am) very eager for your next version and that's why we all keep checking the thread and chat while waiting
;)
Thunderbolt8
9th April 2011, 13:57
and who cares about what you think again :confused:natural limitations do.
leeperry
9th April 2011, 14:07
natural limitations do.
you know, I was also called on bs when I said that 23.976 sounded too slow compared to 24fps.
it's easy to A/B in Reclock, and a 0.1% pitch variation is very much audible...especially on music. Everything sounds sloooooooow, like a FUBAR revox if you see what I mean..or a slightly out of tune musical instrument.
http://www.google.com/url?q=http://hyperphysics.phy-astr.gsu.edu/hbase/sound/earsens.html&sa=U&ei=IlmgTcHuLoK6hAePkdXzBA&ved=0CBgQFjAC&usg=AFQjCNEAH3LzlgPEcSD4sbwzudV1LZMyLw
The normal human ear can detect the difference between 440 Hz and 441 Hz
I had better links, but I can't be hassled to search for them...as long as I don't have to listen to a 400ms desync'ed movie rig, I'll be just fine :p
does anybody of you have a short sample that I can use to test A/V sync?
IME the best is a lips close-up, it's very easy to spot when you're perfectly synced.
maybe you could create a separate thread to make madshi's work easier in this very thread ;)
Thunderbolt8
9th April 2011, 14:22
pitch variation is something different. but audio video sync comparison by 5ms is basically impossible, especially in case of lip sync. every human has different voice onsets when producing different letters. if anything, its easier with mechanical sounds.
Sometimes i have to add tracks to movies. I prefer scenes with gun shots to find minimal delay. Can't get better than +/-50ms. I need leeperry's phones I guess.
Thunderbolt8
9th April 2011, 15:22
shots are usually pretty good, yes. in matrix, theres a scene when they enter that building and shooting is shown in slow motion. then you got the muzzle flash and the audio of few shots after another in the same rhythm, I think thats a good scene to really see how accurate you can be with delay.
sneaker_ger
9th April 2011, 15:25
I highly doubt that movies on DVD/Blu-Ray/Film are synched that exactly in the first place and personally don't believe anyone until I witness them proving it in a proper blind test. But you are correct, this is off topic unless madVR is the culprit of any delay.
madshi
9th April 2011, 16:23
Please stay on-topic, guys. Thanks.
mr.duck
9th April 2011, 16:59
madVR OSD is not very clear. Using MPC with EVR I set the OSD font to Calibri size 20. With the opaque background it looks very good and is clear and easy to read. Would like the same for madVR :)
Yeah, why not. I want OSD font to be twice bigger.
leeperry
9th April 2011, 17:27
I'm color-blind, I can't read the damn OSD w/o squinting like hell :D
madshi
9th April 2011, 19:34
Yeah. And I've tried with default settings, and tried to enable/disable LAN support, etc. There is no real network activity from madVR with disabled LAN support, but it still triggers HIPS.
So the complaint you get now is different from what you got before? (since you're saying that there's no real network activity)
Is there a free/demo version of that firewall you're using available?
Should be \src\apps\mplayerc\FGManager.cpp:2207
Thanks, I'll check that out!
madVR OSD is not very clear. Using MPC with EVR I set the OSD font to Calibri size 20. With the opaque background it looks very good and is clear and easy to read. Would like the same for madVR :)
You mean the tiny red fonts used by madVR? Yeah, they are too small. But then in windowed mode if the video is small, with bigger fonts the whole OSD contents wouldn't fit into the video window. Maybe I should scale the font size with the video resolution. But then I've so many more important things to do right now...
madshi
9th April 2011, 19:37
Just for your information: The final exclusive mode rendering path appears to work (more or less). I still need to do some fine tuning, though. Should be ready for release tomorrow...
SamuriHL
9th April 2011, 19:50
Just for your information: The final exclusive mode rendering path appears to work (more or less). I still need to do some fine tuning, though. Should be ready for release tomorrow...
That's awesome! Can't wait to see that change. Thanks, madshi! Looking forward to it.
pankov
9th April 2011, 19:52
Just for your information: The final exclusive mode rendering path appears to work (more or less). I still need to do some fine tuning, though. Should be ready for release tomorrow...
You are the best!
:thanks:
jeremy33
9th April 2011, 20:27
You are the best madshi !
For those who want to test the sync of their installation I just found this site http://editorsean.com/blog/49-audiovideosynctest with audio video sync test sample
mr.duck
9th April 2011, 21:33
You mean the tiny red fonts used by madVR?
I'm afraid I have no idea what you are talking about. I mean this OSD in the top left...
http://i.imgur.com/poGiM.jpg
It is always the same size regardless of the window/video size (as it should be).
I want it to be exactly the same as the OSD that you would get with MPC-HC + EVR.
fairchild
9th April 2011, 22:12
I'm afraid I have no idea what you are talking about. I mean this OSD in the top left...
He's talking about the font of the stat's when you do CTRL+J with MadVR rendering.
Abradoks
9th April 2011, 23:15
So the complaint you get now is different from what you got before? (since you're saying that there's no real network activity)
Is there a free/demo version of that firewall you're using available?
Well, there are two parts of firewall: HIPS and network filter. Netfilter is ok now (no packets sent by madVR). But it still triggers HIPS by trying to use some kind of network related functions. You can get trial version here (http://www.jetico.com/jpf2setup.exe). But it should be same with any other HIPS.
robpdotcom
10th April 2011, 00:30
Madshi, what improvements/differences are we likely to see with the final FSE rendering path?
Regarding the OSD (not the "red" OSD), I much prefer madVR's over the normal MPC-HC one - it just looks more modern. Combined with the FSE seekbar, it feels like a more modern player all around.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.