Media Player Classic Home Cinema (MPC-HC) - DXVA!

[clsid]

It works! but instead disabled I set it for YUV otherwise videos won't play in SubtitleWorkshop (I remember to enable this and set it for all supported because of SW)
Thanks a lot for your help namaiki!The original problem with wrong colors was probably due to the buggy conversion of BGR24 to RGB24 in ffdshow. So you could also try unchecking RGB24 on the Output page in ffdshow settings. That will trigger conversion to RGB32 and hopefully give proper colors.

[boyumeow]

@Hera,
Have U try using directx web installer rather than standalone installer? Not sure this will work, just checking. Thanks.

[Hera]

I determined that MPC disallows subtitles if I enable CoreAVC, this worked just fine before,
http://img87.imageshack.us/i/coreavchaali.jpg/


EDIT: Renderer settings magically appeared after re-installing DX from standalone installer. Odd - I never had a limbo state where DirectX allowed some features to work but not others - it either worked or told me that I forgot to update DirectX...
Subtitles are still non-functioning.

And here is the VMR9 greyed out options
http://img199.imageshack.us/i/vmr9renderless.jpg/
The *only* settings that work: reset optimal - reset default - tearing test. Everything else greyed out.

Odd.

[namaiki]

subtitles

You seem to have missed the bottom part of my previous post:
Also, what is listed in the menu: Play-> Filters, when you have the video open? (with subtitles)


In MPC-HC: is View-> Options-> Playback-> Auto-load subtitles ticked? Also, are the subtitles in the MKV file or external(for example, .ssa or .srt)

Finally, what build number of MPC-HC are you using? The model of video card matters as MPC-HC's internal subtitle renderer has certain requirements which I can't recall of right now.

[Hera]

Build number is in the screenshots provided - 1.5.1.2959.
Same exact issue with latest stable build.
GPU: NV 6600 non-GT 256MB AGP8x - 2-hundred series drivers.

Issue is Haali Renderer + CoreAVC reproducible. Which I don't get at all.
Subtitle menu was not greyed out when CoreAVC was not set to preferred codec.
Which is odd, or just a coincidence or something... ?

I will do more testing tomorrow.

[ceb]

Hi, I started using MPC-HC a few years ago although only now I'm starting to 'tweak' the settings to have the best quality. I have some questions, I hope someone can answer them even though they come from a newbie. :p
My CPU is an Intel Core i5 2500 and my video card is an ATI 5770. I use Windows 7 64-bit with all the latest drivers, please note that I use the "Only Driver" version of the Catalyst since I don't need the CCC. I use the 64-bit version (with installer) of MPC-HC, currently I have version 1.5.2.2962.
Please don't suggest me using other software (FFDShow, Reclock, etc.) because I want to use onyl MPC-HC.
I mainly watch anime with subtitles, the formats are:
- .mkv - 720p with soft and hard subs
- .mp4 - 720p with hard subs
- .avi - 480p (640x480) (not sure if it is correct, probably DivX quality) with hard subs

1) This is what I do after installing the 64-bit version (with installer) of MPC-HC, correct me if I'm doing something wrong or if I can do anything else to increase video quality. I've only written the changes that affect the video, everything else is at default (except some UI settings).
- change the output resizer to "Bicubic A=-1.00 (PS 2.0)
- in the internal filters, disable the DXVA on MPEG-2, H264/AVC and VC1
- change the maimun texture resolution of subtitles to "Desktop"
- enable the "Sharpen Complex 2" shader - I noticed that this one makes the hard subs look a little bit uglier, but from a certain distance it's not noticable anymore

2) I use the 64-bit version, does it have any downside against the 32-bit one? I don't use any other sfotware (like FFDShow and such), only MPC-HC, so there isn't any problem in that way.

3) When the camera moves in a video, it seems to "stutter" a bit. The audio and subtitles do not go out of sync. Not a big deal, but I would prefer to resolve it.

4) As I've written in the first question, I disable the DXVA. I do this because with .mp4 videos, if the DXVA is enabled, I can only hear the audio with a black video. Maybe it's an incompatibility with the Sandy Bridge processors?
Anyway, since my CPU is powerful enough (I think), should I disable it always?

5) What does the "Round up to power of two" option in the subtitles do?

I know they are a lot of questions, I hope someone has the patience to help me.
Thanks. :)

[namaiki]

Issue is Haali Renderer + CoreAVC reproducible. Which I don't get at all.
Subtitle menu was not greyed out when CoreAVC was not set to preferred codec.


What is listed in the menu: Play-> Filters, when you have a video open with subtitles and CoreAVC is not being used?

In MPC-HC: is View-> Options-> Playback-> Auto-load subtitles ticked? Also, are the subtitles in the MKV file or external(for example, .ssa or .srt)

[Momber]

It has always bugged me that the subtitle renderer of MPC-HC causes dropped frames with DXVA (tried all XP, Vista and 7) so I upgraded to a more powerful computer and am using software decoding now.
Now the subtitle renderer causes tearing! (Using EVR-CP under XP SP3 at the moment).
Please advise how to resolve this, or alternatively, if it's a bug of MPC (which I believe it is because it has been reported by many people for at least two years), provide an ETA for a fix.
Thank you.

[namaiki]

What about: View-> Options-> Renderer-> Reset-> Optimal?
(good luck if you're on ATi)

Also, is this on the bug tracker?

[madshi]

Hey Momber, the best tearing fix is to use fullscreen exclusive mode. Of course that comes at the cost of not having the full GUI (with the stock renderers). EVR-Custom has some tearing/VSync related options, you can try playing with them to get rid of tearing even in non-exclusive mode. Maybe you will succeed. Of course you can also give madVR a try. It automatically switches between windowed and exclusive mode, so you get the best of both worlds... :) However, madVR doesn't support the MPC-HC subtitle renderer yet.

[JanWillem32]

I've patched an over-filtering bug of the subtitle renderer, but it's otherwise a badly maintained, single-threaded resource hog. It even renders on a A4R4G4B4 surface...
I tried setting up a transport of screen space overlay surfaces, just like with the OSD screen (similar renderer) on vanilla EVR. It's very difficult. The lack of developers to help develop the subtitle renderer is a problem, it should have been edited long ago to use some more recent rendering methods.

[cca]

I've patched an over-filtering bug of the subtitle renderer, but it's otherwise a badly maintained, single-threaded resource hog. It even renders on a A4R4G4B4 surface...
I tried setting up a transport of screen space overlay surfaces, just like with the OSD screen (similar renderer) on vanilla EVR. It's very difficult. The lack of developers to help develop the subtitle renderer is a problem, it should have been edited long ago to use some more recent rendering methods.
That's my experience with it too, depends a lot on how fast the CPU is per single core. Makes sense if you realize it was written years ago, before the multicore CPU era.

[JanWillem32]

I'm happy to report succes on replacing the trilinear interpolation for the LUT of the color management, without making it much heavier to process.
Because of a few optimizations, it's now a lot faster to switch the dithering test mode, dithering, and color management off and on again. The initialization of the main renderer should be slightly faster, too.
I'm definitely satisfied with the improvements I made today.

[madshi]

I'm happy to report succes on replacing the trilinear interpolation for the LUT of the color management
What did you replace it with? Isn't trilinear interpolation exactly what is needed? At least that's what I'm using in madVR for 3dlut math.

[janos666]

What did you replace it with? Isn't trilinear interpolation exactly what is needed? At least that's what I'm using in madVR for 3dlut math.

I think tri-linear is the safest method (and should be good enough) but it should be some kind of polynomial interpolation (in theory) because these RGB values are gamma weighted (or not... but mostly...).
And it applies for the dithering noise as well. That should be gamma corrected too (at least in theory...).

You can find his current method in final.psh (in his source code package).

@JanWillem32
Yes, it feels much faster now.

[madshi]

So it's tri-cubic interpolation now or something like that? Ok, you're using smaller 3dlut sizes than yCMS does, right? I guess with small 3dlut sizes using tri-cubic might be worth it.

[janos666]

So it's tri-cubic interpolation now or something like that? Ok, you're using smaller 3dlut sizes than yCMS does, right? I guess with small 3dlut sizes using tri-cubic might be worth it.

64x64x64, 128x128x128 or 192x192x192, according to the settings (Low, Medium, High).
But the values are float16 or float32 numbers, according to the surface settings.

I tried to hit the 256^3 with the High option but it fails with 32-bit numbers. (227^3 is the highest working number but it works with widowed mode only. At least with any VGA cards which it was was tested with... ; 256x256x256@16bit works fine but it's slow to create.)

[Hera]

@Jan
I get corruption (video extremely super contrasted) with this new build when using "Bilinear PS (2.0)" resizer.
This is the Radeon 4250 rig.

EDIT: It might be more appropriate to say Colors become extremely over saturated bright when video is scaled using P2.0 resizer.

[madshi]

64x64x64, 128x128x128 or 192x192x192, according to the settings (Low, Medium, High).
But the values are float16 or float32 numbers, according to the surface settings.

I tried to hit the 256^3 with the High option but it fails with 32-bit numbers. (227^3 is the highest working number but it works with widowed mode only. At least with any VGA cards which it was was tested with... ; 256x256x256@16bit works fine but it's slow to create.)
Ah, interesting. As you (probably) know, madVR/yCMS uses 256x256x256@16bit. If you find any test patterns where you can see or measure a difference between trilinear and tricubic interpolation, please let me know, thanks.

[janos666]

I think the current interpolation is actually worse. :( (But it's only my first impression.)
EDIT: And my second impression is that I should recalibrate my display. :)

[Eliminateur]

Jan,
umfortunately, screencap function is still broken even without FullFP.
i've disabled fullFP (left 10 bit rgb) and look at the result attached

if i disable 10bit then it works as expected.

(this is with RGB32 being fed to the renderer)

[Momber]

Thanks for the recommendations re. renderer settings / subtitle renderer.
And "hi" to madshi in particular. Long time no see :)

I guess I should add that I have already optimized the Renderer / Vsync settings and I'm getting perfectly smooth playback at all appropriate refresh rates just as long as I don't try to display srt subtitles

I use EVR-CP mainly because of the color management it supports. My pj is calibrated and the ICC profile makes quite a noticeable difference when it's getting used by the player.

I'm using D3D fullscreen w/GUI support. Without the use of a GUI, I would probably shoot myself.

Yes, my video card is a Radeon.

My CPUs is an Intel dualcore running at 3.3 GHz per core and Windows task manager shows that it's mostly rather bored, even when playing back BluRay video. No, I'm definitely not using DXVA.

[gngn]

@Jan
I get corruption (video extremely super contrasted) with this new build when using "Bilinear PS (2.0)" resizer.
This is the Radeon 4250 rig.

EDIT: It might be more appropriate to say Colors become extremely over saturated bright when video is scaled using P2.0 resizer.

result is even worser on Nvidia (doesnt matter what decodingfilter is used. doesnt happen with Bicubic or if i check "Video Frame=Normal Size"):
http://i266.photobucket.com/albums/ii259/gngnmumu/dxva/th_bilinearps20.png (http://i266.photobucket.com/albums/ii259/gngnmumu/dxva/bilinearps20.png)

[Hera]

result is even worser on Nvidia (doesnt matter what decodingfilter is used. doesnt happen with Bicubic):
http://i266.photobucket.com/albums/ii259/gngnmumu/dxva/th_bilinearps20.png (http://i266.photobucket.com/albums/ii259/gngnmumu/dxva/bilinearps20.png)

Looks 'bout the same.

[janos666]

@madshi (mainly, but anybody else)
Which one do you think is better: 256^3@float16 or 192^3@float32 (with the current cubic interpolation...)?

Because it seems that we won't hit the 265^3@float32 any time soon, so may be it would be advisable to use float16 textures exclusively and set the High quality preset to 256^3.

The interpolation theoretically works better with higher precision numbers but I am not sure if the sample point quantity is not more important than the precision of the individual samples (given that these are gamma weighted numbers).

But I think these questions are far away from the precision of the spectrophoto-, and colorimeters which we can afford and use at home. (Not to mention our eyes.)

[dawar49]

this latest build is destroying my graphics card. albeit it's only a 8600m gt. tester 2 based off of 2963 worked fine. iirc gpu-z reported gpu usage below 20%. with this build i can only get a few frames per second and gpu usage gets maxed out. eventually mpc crashes. disabling lcms fixes the problem.

[namaiki]

Is full floating point precision on? My 9600M GT(pretty much the same card) can't handle that, only half.

[madshi]

Which one do you think is better: 256^3@float16 or 192^3@float32 (with the current cubic interpolation...)?
Well, I don't have the ultimate truth for that. I can only speculate. For one thing 256^3@16 = 96MB. 192^3@32 = 81MB. So 256^3@16 has a higher data depth... :p Not sure if that means anything, though. I like it that 256^3@16 exactly matches the bitdepth of the source data (8bit). However, interpolation is still necessary, anyway, so I'm not sure how much worth that has. Having higher "output" bitdepth sounds worthwhile, but then 16bit per component = 48bit per pixel already sounds like plenty to me.

[JanWillem32]

A lot of responses this time.:)
The bilinear shader was subject to an experiment. I forgot to paste the correct code back in. I'll integrate it with the next release. Thanks for the input.
I re-used the code from "Catmull-Rom spline4×4 sphere" in the final pass shader for the 3DLUT, so it's a spline4×4×4 interpolation. I didn't try cubic or Lanczos interpolation for it yet. I just wanted to get gradients behave better in 10-bit mode with debanding. Trilinear filtering is not accurate for interpolating (gamma) curves, so I just tried something better suited.

@dawar49: I guess with a video card that has weak memory management and memory clock, sampling 64 items from a volume texture is a bit too much compared to the 8 items that it used to be. I'll add an option to the settings. I didn't notice slower playback or anything at all, so I wonder what I should set as default.

@janos666: I guess we could also try a 16-bit integer format, as the sign and exponent bits aren't really used with both floating-point 3DLUT textures. The problem of that format is that it has worse support than 32-bit floating point, still has less minimum precision and I don't know how fast the integer to floating point conversion on loading pixels from the 3DLUT is with this type.

@Momber: D3D fullscreen lost GUI support ages ago. The bar at the bottom of the screen is generated by the OSD renderer.

@Eliminateur: I can't replicate this problem, but I guess your video card refuses to convert the 10-bit input to 8-bit when transferring into an off-screen plain surface. I haven't found the right function to generate errors with yet, but it should at least refuse to output a file when conversion fails. Does it?

[burfadel]

My HD4850OC 1gb can, just. There is times when it just looks like its below framerate (compared to normal) i noticed last night. Mind you, I'm also using the v22 shader scripts:
4÷2÷0 to 4÷2÷2 intermediate Catmull-Rom spline5 chroma up-sampling for SD&HD video input
4÷2÷2 Catmull-Rom spline5 chroma up-sampling for SD&HD video input
sharpen complex v3 + deband + mild denoise
16-235 to 0-256 for SD&HD video input

[madshi]

I just wanted to get gradients behave better in 10-bit mode with debanding. Trilinear filtering is not accurate for interpolating (gamma) curves, so I just tried something better suited.
Is there a sample I can check to see the non-optimal gradients with trilinear interpolation? Thanks!

[JanWillem32]

@burfadel: Do you actually have the ranges expansion bug? It's rather rare.
The "sharpen complex v3 + deband + denoise" shader is the very heaviest of all. I added the comment to use full 3D clocks with it. I can't use it on 30/1.001 Hz or higher input frame rates with mild rendering settings on 2D clocks at all. With subtitles, it's of course even worse.

@madshi: I'll take a look if I can generate a synthetic test, or find a sample without copyrights. I do have a few samples of the worst banding gradients ever. Even 10-bit output with the "blur" shader in FFPP and single-level dithering added fails on those.
Red to brown top-to-bottom and to black right-to-left is my current "favorite" banding picture. It's nearly as horrible as the "nails on a chalkboard" audio sample I edited for a matrix mixing and clarity test.

[burfadel]

I'm pretty sure I don't have that bug, but strangely I think it does look better with it added! (on both the computer screen and TV). I guess basically its like adding a bit of dynamic contrast enhancement?...

Before someone asks - I have calibrated both screens with the Windows calibration tool (accessible from 'display' or 'color management' in the control panel.

Out of curiosity, how many shaders does each shader script use, such as the sharpen complex v3+deband+denoise?

[madshi]

@madshi: I'll take a look if I can generate a synthetic test, or find a sample without copyrights. I do have a few samples of the worst banding gradients ever. Even 10-bit output with the "blur" shader in FFPP and single-level dithering added fails on those.
Red to brown top-to-bottom and to black right-to-left is my current "favorite" banding picture. It's nearly as horrible as the "nails on a chalkboard" audio sample I edited for a matrix mixing and clarity test.
A sample or synthetic test would be awesome - thanks! :)

[JanWillem32]

@burfadel: The expansion shader does nothing but expanding and unbalancing the 3 RGB saturation ranges and expanding the luma range (more for white than black). The source colors that fall out of this range, are clipped to minimum or maximum intensity, so they're lost.
Both "RGB brightness, contrast and gamma control" and "hue, saturation, luma gamma and chroma gamma control for SD&HD video input" have balanced and accurate color controls. If used appropriately, these won't damage the picture as much.
It tends to depend on the source video, but sometimes I de-saturate with the chroma gamma control and I often shift the general RGB gamma. I hardly ever have to change the other controls.
I use a CRT with the analog controls set to near 2.6 gamma and additional digital calibration by a i1 Display. It's pretty hard to get all source videos lined up to look right, but my settings proved to be very rewarding yesterday. (I also have a projector, but it needs some work at the moment.)
For your other question: the scripts all contain a single shader. That's also why the scalers (muti-pass) are in 2 separate scripts.

I've added audio samples to my folder, I don't know if the video samples are going to work out. The synthetic test I tried can't generate any proper gradients, so I'll have to write a gradient shader. (Maybe it's even useful as an integrated banding and gamma bars test.)

[vBm]

This is offtopic but it's kinda important.
Before you go and install SP1 for VS2010 better read THIS (http://msdn.microsoft.com/en-us/visualc/gg697159) first. Pro and Express versions are exposed to this flaw.

[Eliminateur]

I'm using a Radeon 5870 with cat 10.12 (december ones iirc)
It isn't refusing anything, it saves the file as shown(it isn't a printscrn->save the one i attached, it's a result from the save image function)


@Eliminateur: I can't replicate this problem, but I guess your video card refuses to convert the 10-bit input to 8-bit when transferring into an off-screen plain surface. I haven't found the right function to generate errors with yet, but it should at least refuse to output a file when conversion fails. Does it?

[hoborg]

DTS stuttering problem

Hi.
I found some MKV with DTS track that is causing stuttering and droped frames if EVR custom or EVR sync renderer is in use.
Simple EVR have no problem.

Sample video is here (http://rapidshare.com/files/452072726/Sample_PA2.mkv).
I tryed LAVF splitter, FFDshow audio/video decoder, with the same result. Problem seems to be in custom renderer.

[G_M_C]

Janwillem, Janos, i have (another) i hope you have the opportunity to answer;

Currently my setup is as follows;
PC, Ati HD 5770, YUV 4:4:4 full range --> Onkyo TX-SR 875, with REON VX enabled.

As you can see, my receiver has a S.O. Reon video processor build in. It has full 10/12 bit processing. My receiver accepts, among others, YUV 4:4:4 and RGB 4:4:4 (both studio as full range) and some 10-bit color mode I guess.

Through the CTRL-J screen I see that MPC's mixer outputs A32R32G32B32F, I also read that 10 bit color is set for the output of D3DFS.

My question is as follows; In relation to what my video processor accepts, what output color space do i set my HD 5770. I can select YUV 4:4:4 (studio/full) or RGB 4:4:4 (studio/full). My goal is to have as less conversions as possible.

The reason i came to my question: I see the MPC's mixer outputs A32R32G32B32F, but I have set output on my HD5770 as YUV 4:4:4 (full). Isnt there a conversion needed between the two, the RGB (mixer) -> YUV ( output) ?

[janos666]

@janos666: I guess we could also try a 16-bit integer format, as the sign and exponent bits aren't really used with both floating-point 3DLUT textures. The problem of that format is that it has worse support than 32-bit floating point, still has less minimum precision and I don't know how fast the integer to floating point conversion on loading pixels from the 3DLUT is with this type.


EDIT: Hmm... On a second thought it sounds much better to use unsigned 16-bit integers instead of float16 numbers. And we could use bigger LUTs (like 256^3, even with the 32-bit surface mode), and I think it would worth it because 16-bit should be enough while the quantity of the sample points is important too, given the nonlinear nature of the RGB numbers. (And we could still benefit from the extra precision of the float32 surface during the interpolation which is inevitable anyway...)
And it would allow some more simplifications...
However, my first quick attempt to use a D3DFMT_A16B16G16R16 texture didn't go well.


@G_M_C
You have to choose RGB Full in CCC. (It's not a question with today's hardwares and softwares.)
The source is usually YV12 which is YCC 4:2:0 but you can't send it out through any standard connections. You have to resize the chroma first. The result is usually RGB (as your VGA usually works with RGB data) and your VGA should send out it without converting it back to YCC, just to let your external VP to convert it back to RGB again (because your display has RGB pixels ; there is no YCC display).
And the precision of this conversion depends on your Presentation settings. It can be even float32 which is much higher than the precision of your external VP anyway. (More precisely, the surface is float32 at max, but the calculation precision of your GPU can be even higher...)

[Aleksoid1978]

DTS stuttering problem

Hi.
I found some MKV with DTS track that is causing stuttering and droped frames if EVR custom or EVR sync renderer is in use.
Simple EVR have no problem.

Sample video is here (http://rapidshare.com/files/452072726/Sample_PA2.mkv).
I tryed LAVF splitter, FFDshow audio/video decoder, with the same result. Problem seems to be in custom renderer.

Play fine - no drop frames. EVR Custom with internal DTS and ffdshow Audio Decoder.

[hoborg]

Play fine - no drop frames. EVR Custom with internal DTS and ffdshow Audio Decoder.

Whole video? It doesn't appear on begin. Win7x64, ATI 5770, stereo output (no passthrough).

For example MediaPortal users have exactly same issues (http://forum.team-mediaportal.com/watch-edit-videos-104/mkv-dts-dropped-frames-94320/).

EVR:
http://www.volny.cz/hoborg/evr.jpg

EVR Custom:
http://www.volny.cz/hoborg/evr_custom.jpg

[Aleksoid1978]

Whole video? It doesn't appear on begin.

Yes - whole video(http://rapidshare.com/files/452072726/Sample_PA2.mkv). drop frames = 0.

[Mercury_22]

Why isn't bobdynlan & Leo1000's patch AKA BE Mod (http://www.xvidvideo.ru/media-player-classic-home-cinema-x86-x64/media-player-classic-homecinema-x86-x64-1-5-2-2969.html) in the svn ?
Does anybody oppose to make their mod the default GUI ?

[hoborg]

Yes - whole video(http://rapidshare.com/files/452072726/Sample_PA2.mkv). drop frames = 0.

No stuttering?
Maybe you have some diferent setting in custom renderer.
Ideas?

EDIT:
Stats added:
http://www.volny.cz/hoborg/evr_custom_stats.jpg

[G_M_C]

[...]

@G_M_C
You have to choose RGB Full in CCC. (It's not a question with today's hardwares and softwares.)
The source is usually YV12 which is YCC 4:2:0 but you can't send it out through any standard connections. You have to resize the chroma first. The result is usually RGB (as your VGA usually works with RGB data) and your VGA should send out it without converting it back to YCC, just to let your external VP to convert it back to RGB again (because your display has RGB pixels ; there is no YCC display).
And the precision of this conversion depends on your Presentation settings. It can be even float32 which is much higher than the precision of your external VP anyway. (More precisely, the surface is float32 at max, but the calculation precision of your GPU can be even higher...)

Thanx Janos for the comprehensive explanation :)

[Aleksoid1978]

No stuttering?
Maybe you have some diferent setting in custom renderer.
Ideas?

i don't use any Sync .

http://s54.radikal.ru/i144/1103/d4/104e20ad6c9at.jpg (http://radikal.ru/F/s54.radikal.ru/i144/1103/d4/104e20ad6c9a.png.html)

http://s61.radikal.ru/i172/1103/f8/1de72492357et.jpg (http://radikal.ru/F/s61.radikal.ru/i172/1103/f8/1de72492357e.png.html)

[namaiki]

No stuttering?
Maybe you have some diferent setting in custom renderer.
Ideas?

EDIT:
Stats added:

Perhaps 24fps material might be more tolerable on a screen set to 60Hz, or a multiple of 24.

[Momber]

@Momber: D3D fullscreen lost GUI support ages ago. The bar at the bottom of the screen is generated by the OSD renderer.
What about the context menu?
Anyhoo, I referred to it by its current name. I did not name it that way.

http://i17.fastpic.ru/big/2011/0312/d0/9e21eb59811c6b2a73064d83641772d0.gif

Still waiting on enlightenment regarding the subtitle renderer issue (tearing and/or dropped frames). TIA.

[Hera]

Back to haali ssue, under Filters I have,

Default Direct Sound Device
Video Renderer
Audio Switcher
CoreAVC Video Decoder
DTS Decoder
And a subtitle menu there as well


Subs are enabled in Maroska Splitter and in "Navigate - Subtitle Language" but the right lcik subtitle menu is greyed out and there is no subtitles.Ideas welcome.