View Full Version : MPC-HC GothSync tryouts
Pages :
[
1]
2
3
4
5
6
7
8
9
10
11
12
13
14
15
ar-jar
6th July 2009, 15:34
Here's the first prototype version of MPC with frame lock features added from my (likewise experimental) GothPlayer movie player (see this thread: http://forum.doom9.org/showthread.php?t=146012). The objective is to eliminate all judder caused by the small mismatches between the video frame rate and the display refresh rate that are always present when playing video on a PC.
The prototype should be compatible with the existing vsync features of MPC-HC except Alternative VSync.
If you wish to try it out, please read more and download the exe file from my blog here: http://www.ostrogothia.com/video/?page_id=1050. Post any comments or bugs on this thread for the time being.
tetsuo55
6th July 2009, 16:01
Hey ar-jar, glad you finally got things working, even better to see the first build out!.
If i understand correctly it only work with VMR9 right? Do you have any plans to expand to the other renderers(Especially EVR-CP)?
EDIT:
I see your blog is full of the strugle trying to understand MPC-HC, cool stuff!
I hope you can pour all those compiler woes into a step by step guide to compiling from scratch?
ar-jar
6th July 2009, 16:55
If i understand correctly it only work with VMR9 right? Do you have any plans to expand to the other renderers(Especially EVR-CP)?
Yes, it's VMR9 only so far. I don't own anything newer than Windows XP as of now (I guess EVR came with Vista). But it shouldn't be too hard to plug it into other renderers. The algorithm is rather "non-intrusive" and thus easy to wire to pretty much anything that displays frames through DirectX. My plan is to get VMR9 to work well first. Then I'll look at the other renderers.
-A
tetsuo55
6th July 2009, 16:57
Some basic testing on an old system using SD content, the numbers are changing so it appears to be working, no noticable difference though. (this system uses windows 7, so EVR-CP results should be better)
tetsuo55
6th July 2009, 16:59
I don't own anything newer than Windows XP as of now (I guess EVR came with Vista). EVR can be made to work with XP too, You can get windows 7 for free(legally) here http://www.microsoft.com/windows/windows-7/download.aspx
The licence will work until march 2010 iirc
ar-jar
6th July 2009, 17:15
Some basic testing on an old system using SD content, the numbers are changing so it appears to be working, no noticable difference though. (this system uses windows 7, so EVR-CP results should be better)
If the Delay values in the brackets (in the on-screen stats) stay between say 0.35 and 0.65 then the frames come into PresentImage with the proper timing in relation to vsync. As long as the # of adjustments (on-screen stats too) doesn't change, you are already in sync and nothing needs to be done so you won't see any difference. It's when the video and the display refresh get too far out of sync (Delay gets too close to 0.0 or 1.0) that the algorithm kicks in.
I've only done some basic testing with some avi files, a DVD and a blu-ray movie but those work fine here on two different computers.
VMR9 is really not a bad renderer if you handle it with care :-)
-A
rack04
6th July 2009, 17:45
See to it that your display refresh rate is an exact multiple of the frame rate of the video.
What and how do I set the display refresh rate for 23.976 fps video?
ar-jar
6th July 2009, 17:51
What and how do I set the display refresh rate for 23.976 fps video?
It depends on your display. My test setup has a HDTV display that accepts a 47.95 Hz refresh rate which works fine with 23.976 video. Some newer displays and projectors accept straight 23.976 Hz.
I use Powerstrip to set the refresh rate. Or you can use the graphics board's control panel to create a custom resolution. Not all displays can handle this refresh rate though. -A
clsid
6th July 2009, 18:43
To get EVR on Windows XP all you need to do is install Microsoft .NET 3.5.
tetsuo55
6th July 2009, 18:47
If the Delay values in the brackets (in the on-screen stats) stay between say 0.35 and 0.65 then the frames come into PresentImage with the proper timing in relation to vsync. As long as the # of adjustments (on-screen stats too) doesn't change, you are already in sync and nothing needs to be done so you won't see any difference. It's when the video and the display refresh get too far out of sync (Delay gets too close to 0.0 or 1.0) that the algorithm kicks in.
I've only done some basic testing with some avi files, a DVD and a blu-ray movie but those work fine here on two different computers.
VMR9 is really not a bad renderer if you handle it with care :-)
-AYeah once every so often adjustments increases. (using 25p content on a 85hz screen)
ar-jar
6th July 2009, 19:44
Yeah once every so often adjustments increases. (using 25p content on a 85hz screen)
That's never going to work I'm afraid. 25p on a 75 Hz would be fine; screen refresh rate must be an even multiple of video fps to start with. My patch can only do fine-tuning. Sorry.
-A
Casshern
6th July 2009, 19:48
Thanks for this great player,
i will try it tonight but could you elaborate a little on how it works? Before you implemented 2 variantions in gothplayer: 1) adjust powerstrip timings or 2) adjust audio clock. I guess its the latter you implemented here. Does this have implications for spdif output at a near exact multiple of the source frame rate? Do the shaders still work?
regards,
casshern
ar-jar
6th July 2009, 19:53
To get EVR on Windows XP all you need to do is install Microsoft .NET 3.5.
I think I must have installed .NET 3.5 when I upgraded to VS 2008 because now my filter manager actually reports that I have EVR installed! So it goes! Thanks for that clue clsid!
-A
ar-jar
6th July 2009, 20:17
Thanks for this great player,
i will try it tonight but could you elaborate a little on how it works? Before you implemented 2 variantions in gothplayer: 1) adjust powerstrip timings or 2) adjust audio clock. I guess its the latter you implemented here. Does this have implications for spdif output at a near exact multiple of the source frame rate? Do the shaders still work?
regards,
casshern
I will update the instructions on my blog when I get more time. Anyway, right now I only do number 2 kind of (number 1 is actually implemented too but as I haven't implemented any settings yet so I just hard-coded number 2 for the time being as it is more universal). It actually inserts a new reference clock into the graph that takes over from the audio renderer so it doesn't use the audio renderer's clock at all. Then it gently tweaks the new reference clock up and down as needed to stay in sync. Like a "Reclock lite". I had to go and check but yes, I'm using SPDIF from the computer to my HarmanKardon receiver in my home cinema setup. It seems to work fine as far as my ears can discern. The audio renderer syncs up to the rate of the reference clock and so does my receiver as it seems. I have watched quite a few movies with this sync on with GothPlayer and never heard any glitches.
I have not touched the actual image presentation code so shaders and most other stuff should work just as before.
-A
ashlar42
7th July 2009, 08:48
Tried it yesterday night, with Reclock but disabling vsync correction. Playback seemed smooth.
I honestly don't completely understand the point of this when Reclock exists and works well. Maybe the point is this being open sourced?
In any case, it worked well but I did not do extensive testing, to be honest. One thing I noticed: if I selected DirectSound default renderer (in my case it's an X-Fi) the channel placing in DTS came out wrong. Not so by using Reclock as audio renderer.
ar-jar
7th July 2009, 13:23
Tried it yesterday night, with Reclock but disabling vsync correction. Playback seemed smooth.
I honestly don't completely understand the point of this when Reclock exists and works well. Maybe the point is this being open sourced?
In any case, it worked well but I did not do extensive testing, to be honest. One thing I noticed: if I selected DirectSound default renderer (in my case it's an X-Fi) the channel placing in DTS came out wrong. Not so by using Reclock as audio renderer.
I guess the main point is that I absolutely enjoy the intellectual challenge of this. I'm not a professional programmer so this is my sudoku or on-line chess if you wish. If it works for somebody else too, then so much better. If not, I will do this just for fun anyway. And yes, it's open source which is kind of nice.
The development of this solution has been a winding road. I started with a solution that I still consider as the optimal one from a theoretical point of view, one that controls the refresh rate of the display and leaves the filter graph alone. Just like an old fashioned TV. This solution depends on Powerstrip fully supporting the graphics board which is not the case e.g. for the newest NVidia boards. There may also be displays that don't tolerate the tweak. All my four 720p displays work well with it but I have never tested it with a 1080 display (one of those is on the shopping list...). This solution is in the code but not exposed as I haven't implemented any GUI components yet.
The solution that you've tested was just a quick and very simple fix that worked surprisingly well. That's why I have left it in the code. I like simple solutions if they work. I will soon add some GUI stuff that exposes both solutions.
As to the mixing of channels, I don't have any idea, sorry. I have not looked into what a DTS bitstream looks like so I don't know how the channels are tagged and how they can possibly be mixed. How does the regular 1165 build of MPC work in this respect? (It's identical to my build minus the sync stuff.)
-A
tetsuo55
7th July 2009, 18:24
That's never going to work I'm afraid. 25p on a 75 Hz would be fine; screen refresh rate must be an even multiple of video fps to start with. My patch can only do fine-tuning. Sorry.
-ABut even at an non-multiple refresh rate the timing would be be better than without the patch?
ar-jar
7th July 2009, 18:58
But even at an non-multiple refresh rate the timing would be be better than without the patch?
I doubt that it will do much good in those cases. At least I don't understand why it should. The sample arrival times will be pretty evenly distributed over the whole vsync cycle with or without my code. I think Beliyaals fixes are much more useful in those cases as they seem to skip or duplicate frames in a controlled fashion. (Although I have to admit that I haven't figured out the EVR code yet as I have never worked with EVR before.) -A
Casshern
7th July 2009, 22:25
Just tried it and it works pretty good for me. Only one thing: It uses and old DirectX SDK (35) instead of the new 41. This leads to problems on some machines with the shaders and the upscaler. Please update so it uses the 41 dll, like the regular builds (i even copied the d3dx9_41.dll to the same directory as the mpc exe, but it still showed version 35 on the stats screen)
Apart from that: Great work... the adjustments (even when the frequency is a little off - by 0.001) did not seem to lead to audible problems with the spdif streams. So it seems that my amp can handle slight variations in clock.
Its also fantastic that something like reclock is finally open source - sadly james from slysoft stated that reclock will not be developed further. So this is great timing, even if not all features of reclock are implemented. For me - i only use it with exact multiples of the source frame rate - its already good enough. Others might miss some of reclock features like pal speeddown, the adaption to non multiple screen refreshrates, the pitch correction etc...
ashlar42
7th July 2009, 23:31
I guess the main point is that I absolutely enjoy the intellectual challenge of this. I'm not a professional programmer so this is my sudoku or on-line chess if you wish. If it works for somebody else too, then so much better. If not, I will do this just for fun anyway. And yes, it's open source which is kind of nice.This is not nice, it's great and I thank you for it. My secret hope is that part of your work finds its way into XBMC. :D
And I totally respect your desire of facing this as an intellectual challenge. I understand the feeling.
This solution depends on Powerstrip fully supporting the graphics board which is not the case e.g. for the newest NVidia boards.This is *bad*. With newest Nvidia cards Powerstrip now means all cards released in the past two years or so. That's a big slice of the market (one of which I'm part...) and it's a shame, considering Nvidia has done many things pretty well, with CUDA, OpenGL support, etc. There may also be displays that don't tolerate the tweak. All my four 720p displays work well with it but I have never tested it with a 1080 display (one of those is on the shopping list...).Tried it on a 1080p screen, seemed not to care about the resolution.As to the mixing of channels, I don't have any idea, sorry. I have not looked into what a DTS bitstream looks like so I don't know how the channels are tagged and how they can possibly be mixed. How does the regular 1165 build of MPC work in this respect? (It's identical to my build minus the sync stuff.)I'm using regular 1164 and there weren't problems. Unless 1165 regressed somehow, I think it depends on your stuff.
ar-jar
7th July 2009, 23:38
Just tried it and it works pretty good for me. Only one thing: It uses and old DirectX SDK (35) instead of the new 41. This leads to problems on some machines with the shaders and the upscaler. Please update so it uses the 41 dll, like the regular builds (i even copied the d3dx9_41.dll to the same directory as the mpc exe, but it still showed version 35 on the stats screen)
Great that it worked for you too! The DS audio renderers seem to accept a speed shift of <= 0.5% or so. And at least your and my receiver too. So it's not good enough for PAL down-shift but it's good enough to fine-tune the graph clock.
I'm trying to figure out where to put the sync code in the EVR presenter but to paraphrase Bowman in 2001, "it's full of threads", and I need to wrap my brain around it some more.
I actually downloaded the old SDK according to the build instructions when I started this tweak. I always use the latest one for my other stuff. i think some files were moved out of the SDK after 35 so I probably need to go include-file hunting when I upgrade. But I'd be happy to delete the old stuff from my HD and use the new version if that's what the MPC folks use already. -A
travolter
8th July 2009, 12:03
I use an avisynth script for interpolate frames (framerate doubler).. and In this way I really notice the difference. (playing 30fps movies at 60fps into my 60hz monitor have a "Live" effect in all these videos)
Could be possible implement a framerate doubler into your proyect?
pirlouy
8th July 2009, 12:18
I use an avisynth script for interpolate frames (framerate doubler)..
Can you give a link to it please (with plugin versions needed) ?
TIA
ar-jar
8th July 2009, 13:02
I use an avisynth script for interpolate frames (framerate doubler).. and In this way I really notice the difference. (playing 30fps movies at 60fps into my 60hz monitor have a "Live" effect in all these videos)
Could be possible implement a framerate doubler into your proyect?
I know what you mean with "live". I have an AVCHD camera that records 1080i material with 50fps. I play it with the Cyberlink h.264 decoder which actually produces 50 *different* frames per second out of those 50 fields which gives a very fluid motion.
As to our question, do you mean a real interpolating framerate doubler like the one you refer to or the ones in some new TVs and projectors and that works for non-interlaced material? (For interlaced material you could use the Cyberlink decoder which gives you pretty much what you want and works with MPC - with some aspect ratio issues though.) The question is where to get the algorithm. Such a function is bound to be non-trivial. Are you aware of any open source code that implements it? Is the avisynth plugin open source? How fast is the plugin? Does it process frames in real-time?
Ideally there could already be a Directshow filter that implements this. But if not, and if I get my hands on a fast enough algorithm, I could give it a try when done with the the renderer stuff. -A
tetsuo55
8th July 2009, 14:52
I know what you mean with "live". I have an AVCHD camera that records 1080i material with 50fps. I play it with the Cyberlink h.264 decoder which actually produces 50 *different* frames per second out of those 50 fields which gives a very fluid motion.
As to our question, do you mean a real interpolating framerate doubler like the one you refer to or the ones in some new TVs and projectors and that works for non-interlaced material? (For interlaced material you could use the Cyberlink decoder which gives you pretty much what you want and works with MPC - with some aspect ratio issues though.) The question is where to get the algorithm. Such a function is bound to be non-trivial. Are you aware of any open source code that implements it? Is the avisynth plugin open source? How fast is the plugin? Does it process frames in real-time?
Ideally there could already be a Directshow filter that implements this. But if not, and if I get my hands on a fast enough algorithm, I could give it a try when done with the the renderer stuff. -AIf such an algorythm is ever added, i hope it will have the option to do Any > 60
Mark_A_W
9th July 2009, 03:54
Ar-Jar
Will this MPC-HC version work with multi monitors? How about a secondary monitor?
ar-jar
9th July 2009, 09:06
Ar-Jar
Will this MPC-HC version work with multi monitors? How about a secondary monitor?
Yes, that is the objective. I believe the clock sync implemented in the first prototype already works on any monitor (with a refresh rate close enough to the video fps). If not, pls let me know. -A
travolter
9th July 2009, 11:31
framerate doubling
Can you give a link to it please (with plugin versions needed) ?
TIA
first tutorial here:
http://forums.guru3d.com/showthread.php?t=288017
then other guys at other forum improved the script
http://www.avsforum.com/avs-vb/showthread.php?t=719041&page=98
follow next pages
you will need ffdshow with avisynth script support installed, also MVtools and avisynth...
also need to replace some files after install...
http://www.megaupload.com/?d=NLHQUV3R
Delete the avisynth.dll and mt.dll and put these versions in their place.
Rebbot after you put these new plugins in their folders:
system32 foldr has the avisynth.dll
avisynth plugins folderr has the mt.dll
PLZ if anyone have updates about these framerate doubling pryects. .let me know.. Im would love see some players with this feature enabled.. like the old crystal player
travolter
9th July 2009, 11:43
As to our question, do you mean a real interpolating framerate doubler like the one you refer to or the ones in some new TVs and projectors and that works for non-interlaced material?
exactly same technology of these Flat TV planels.. "truemotion" and similar names
The question is where to get the algorithm. Such a function is bound to be non-trivial. Are you aware of any open source code that implements it? Is the avisynth plugin open source?
check my post above for links. Free proyect... so GPL or Opensource I suppose
How fast is the plugin? Does it process frames in real-time?
Im playing all my 1080 videos with the script real time.
Oh man! .,. the movies look amazing. The eye can follow the movement without any jump.. and always with a feeling of "live motion"
Im using a quadcore processor (about 50% of cpu usage).. but other guys are doing it in dual cores without problems
the script that Im using currently is this one: gives me enough quality and not so much CPU usage.. (you can change parameters there to impreve quality but consume more CPU)
SetMTmode(2,3)
last=ffdshow_source()
super=MSuper(pel=1, hpad=8, vpad=8, chroma=true)
backward_vec=MAnalyse(super, blksize=16, chroma=false,\
search=3, searchparam=1, isb = true)
forward_vec=MAnalyse(super, blksize=16, chroma=false,\
search=3, searchparam=1)
MFlowFps(super, backward_vec, forward_vec, num=0, mask=0)
distributor()
ar-jar
9th July 2009, 11:56
Im playing all my 1080 videos with the script real time.
Oh man! .,. the movies look amazing. The eye can follow the movement without any jump.. and always with a feeling of "live motion"
Thank you for the pointers. I'm not one of those nostalgic types that cling to 24 fps for that "genuine cinema experience" and I have lamented about that arcane fps standard with its built-in judder many times (see e.g. http://www.ostrogothia.com/video/?p=636, with a link to an interesting article at Projector Central). So this all looks very exciting and I will definitely look at it later on. But first I need to finish the renderer sync. -A
boyumeow
10th July 2009, 07:49
Hihi, I have encounter a minor (I think) problem on ur tryout build(1165.9001). I could not play file extension .rm and mpc shutdown by itself, while rename it to .rmvb plays normally. Revert back to r1164 play on both extensions normally. r1165 on svn shows update on translation language, so shouldn't be a problem if I m not wrong. Btw my file extension for .rm was set to RealMedia, .rmvb was DirectShow. My window environment is Vista32, EVR custom (try out VMR9 renderless, same thing happens). Thanks.
Forget to remind U something, enjoy ur holidays, hehe.
flanger216
11th July 2009, 01:01
Thank you for the pointers. I'm not one of those nostalgic types that cling to 24 fps for that "genuine cinema experience" and I have lamented about that arcane fps standard with its built-in judder many times (see e.g. http://www.ostrogothia.com/video/?p=636, with a link to an interesting article at Projector Central). So this all looks very exciting and I will definitely look at it later on. But first I need to finish the renderer sync. -A
It's not that we're being nostalgic; we just feel that, if a movie were shot in 24fps, it should be displayed at 24fps, just like if a movie were shot in black & white, it shouldn't be colorized. When you upsample 24fps to 60fps, at least 60% of every image that hits your visual cortex is an interpolated guesstimation (with some algorithms, it's actually 100%!), and with most of the current methods, a sizeable chunk of those new frames are just distorted junk. I mean, home-video enthusiasts have been clamoring for 24fps playback for years, and now that we've got it, it's like all-of-sudden we're clamoring to interpolate everything into a smeary mess all over again. Obviously YMMV is in play here, but to my eyes, Truemotion and the like are a lot like oversharpening, or pushing the contrast, or boosting the reds: it gives a great first impression, but if you really focus on the effect, it comes off as hideously artificial.
But yeah, the 24fps standard needs to go. Us cinematographers have been pushing for Maxivision-48 for years (a 48fps film system), and there's really no excuse for 24fps now that the digital revolution is in full swing.
[and, of course, 'art-house' stuff looks particularly preposterous when fed through these interpolators]
tetsuo55
11th July 2009, 07:46
Unless you have a player and a display that can perfectly switch between 24,25,30,50 and 60 hz one needs to find a solution that results in the least jitter and tearing.
I have an equal mix of all 5 formats + for gaming the desktop has to be in 60p mode.
In my case any-to-60 would be the best solution
pirlouy
11th July 2009, 15:58
Personnally, I have chosen 50Hz.
Reclock plays film at 25fps, and game at 50Hz is enough.
@ar-jar: like some others, I use EVR and not VMR9. I have some problems with VMR9 (maybe due to 7), so if you are ready to port your code to EVR, it would be great to test.
Casshern
12th July 2009, 09:53
I actually downloaded the old SDK according to the build instructions when I started this tweak. I always use the latest one for my other stuff. i think some files were moved out of the SDK after 35 so I probably need to go include-file hunting when I upgrade. But I'd be happy to delete the old stuff from my HD and use the new version if that's what the MPC folks use already. -A
Any news on compiling with the new SDK? I would love to finally ditch reclock (as i exclusively use exact frequency multiples), but with the old SDK i cannot use the shaders (Chroma Upsampling) without severe juddering. This was the same with the regular build until the devs used the new SDK....
ar-jar
12th July 2009, 11:13
Any news on compiling with the new SDK? I would love to finally ditch reclock (as i exclusively use exact frequency multiples), but with the old SDK i cannot use the shaders (Chroma Upsampling) without severe juddering. This was the same with the regular build until the devs used the new SDK....
_xxl seems to be working on it. I'll let you know. -A
Casshern
12th July 2009, 19:15
_xxl seems to be working on it. I'll let you know. -A
Cool! How about adding some options? For example to turn the sync code on/off. Maybe even so that reclock is prevented from loading when your code is active - this could be done just by having a second audio renderer selection box when your code is enabled. There one could just use the default audio renderer.....
The best thing about your code is that it - in theory - makes the vsync position slider obsolete in reclock, as you have the renderers knowlege about scanlines/vsync positions. This would mean no more vsync adjustments for different filter chains and smooth playback from the get go (without any pause/play hassels or bad vsync positiions stutter when the vsyncs occur in the "critical" area). Can't wait for the day....
tetsuo55
12th July 2009, 19:17
kierank fixed the problem, the wiki already contains instructions to get the directx SDK problem out of the way.
Using any platform SDK version has also been cracked, but not yet updated on the wiki
Kaotech
14th July 2009, 09:58
Hello, this version work very well, but in your site you say "The Delay parameter should stay close to 0.5." My delay parameter move during playing movie. How can i fix it ?
http://img149.imageshack.us/img149/9602/goth.png
ar-jar
15th July 2009, 13:02
Hello, this version work very well, but in your site you say "The Delay parameter should stay close to 0.5." My delay parameter move during playing movie. How can i fix it ?
If it moves between say 0.35 and 0.65 then you're fine. If it's all over the place, then you may not have a display refresh rate that is an even multiple of your video fps. -A
Mark_A_W
16th July 2009, 00:48
Cool! How about adding some options? For example to turn the sync code on/off. Maybe even so that reclock is prevented from loading when your code is active - this could be done just by having a second audio renderer selection box when your code is enabled. There one could just use the default audio renderer.....
The best thing about your code is that it - in theory - makes the vsync position slider obsolete in reclock, as you have the renderers knowlege about scanlines/vsync positions. This would mean no more vsync adjustments for different filter chains and smooth playback from the get go (without any pause/play hassels or bad vsync positiions stutter when the vsyncs occur in the "critical" area). Can't wait for the day....
As there is no other KS/Wasapi renderer, why would you want to disable reclock completely?
You can just set it to Slave/Original Speed/Locked and resampling is disabled, but you still get the best audio renderer.
And, umm, to tell you the truth guys, I'm getting smoother playback with madVR and Reclock active. But I'm still playing around.
KornX
16th July 2009, 18:56
can the display output (values eg) be dumped in a *.log?
<Offtopic>
@kaotech
mmh think ive seen this movie,
but I can't think of the name...
Sth. with "The ..."?
</Offtopic> ;)
KornX
ar-jar
16th July 2009, 21:39
There is a new version of the patch that claims to synchronize your video to your display (and now also vice versa). It can be found here: http://www.ostrogothia.com/video/?page_id=1050. There is a change log and known issues at the bottom of the page for those of you who like the fine print. Please report any issues on this thread and I'll try to respond and fix (if fixable) asap. Hope it works for you as well as it works for me (of some reason software always works better in the development environment than when it hits the reality :-)).
ar-jar
16th July 2009, 21:43
can the display output (values eg) be dumped in a *.log?
Currently no. What would you like to analyze? Maybe it can be done some other way. Also please check out the new patch version with much better on-screen statistics. -A
pirlouy
17th July 2009, 12:01
I just did a test yesterday evening, with EVR custom, and my TV in 24Hz, and it was the first time I had no jerk in 24Hz.
I'll do more tests in some days but it looks very promising.
Kaotech
17th July 2009, 13:36
<Offtopic>
@kaotech
mmh think ive seen this movie,
but I can't think of the name...
Sth. with "The ..."?
</Offtopic> ;)
KornX
Final Destination 2 ;)
mariner
17th July 2009, 16:45
Greetings ar-far. Many thanks for the brilliant piece of work.
Some observatins after testing the following 1080i H264 file on single core CPU with HD3650 AGP card runnibng XP3.
1. Doesn't work well running in normal full screen (not D3D). Enabling Alternate VSync seems to help.
2. The OSD reports 29.970/Progressive. Why not interlaced? FYI, uisng Arcsoft Decoder in DXVA/VA de-interlacing mode.
3. Can you also display the actual frame rate as well? In my test environment, refresh rate reported by PStrip was 59.996.
Many thansk and best regards.
http://drop.io/hidden/mbrbkguvrfoufa/asset/c2xpY2llcy1udHNjLTEwODBpLWgtMjY0LXRz
Casshern
17th July 2009, 16:51
There is a new version of the patch that claims to synchronize your video to your display (and now also vice versa). It can be found here: http://www.ostrogothia.com/video/?page_id=1050. There is a change log and known issues at the bottom of the page for those of you who like the fine print. Please report any issues on this thread and I'll try to respond and fix (if fixable) asap. Hope it works for you as well as it works for me (of some reason software always works better in the development environment than when it hits the reality :-)).
As i read from the log its also compiled with the new Dx SDK - great - will try it tonight....
mariner
17th July 2009, 17:19
Here's another test of 1080i VC1 file in EVR D3D Full Screen mode, again using Arcsoft Video Decocder.
1. Is the green line supposed to be straight? (The glitch is probably due to pressing Alt Print Screen)
2. While the frame rate is now reported correctly as 59.940 interlaced, isn't it supposed to be speeded up to 59.996 to match the actual refresh rate?
Best regards.
JonasNo
17th July 2009, 19:02
gothsync patch 1.2.1173.9003 results:
-gothsync uses slightly more cpu then the provided version without gothsync
-gothsync stabalizes the video play better, with some more work/tweeks it will be a good addition to mpc-hc
-issue, the 'control sync offset' can sometimes display a very big number, that continues off the screen, see gothsync_9003_stats_bug.JPG
-issue, the green line keeps on moving when paused
-issue, the 'sync offset' keeps on updating when paused
-issue, moving the player from one screen to another:
1. while still playing: sometimes crash, sometimes continue playing but no sound or video, partial hang
2. pausing before moving it to another screen and then unpause, will make it crash
But mpc-hc have always had some issues with moving from one screen to another, like hanging, miss-coloring the video (wmv), video skipping, etc....
One weird thing that always happens when moving the player from one screen to another is this, look at the graph, see gothsync_9003_moving to another screen.JPG
if i move the player back to the first screen doesn't fix it,
the workaround is the reopen the video
this is how the graph normally looks on my computer:
see gothsync_9003_normal_graph.JPG
ar-jar, nice work indeed, i am eagerly awaiting the patch file so i can study your nice work
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.