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

[BelowSky]

One of missing feature is also the lack of an amr splitter in mpc.
Currently mpc does have an amr decoder but without a splitter.
is there any chance of porting amr splitter from ffmpeg?

http://samples.mplayerhq.hu/A-codecs/amr/sound.amr

[Reino]

1) About MP4 - give a sample.
2) About RV2,cook - give a sample.

And - show a filters list when you play or crash.

P.S. RV2,cook - play fine, no crash.It's all in my post :confused:

This is the case with the internal MP4 Splitter, Video- and Audio decoder. With FFDShow instead of the internal Audio decoder, the video plays "normal" after a seek, but cpu is still at 100% for 30 seconds. No problems at all with LAV Filters.

- If you don't have the Firefox plugin to download the full-length Youtube movie, this link (http://o-o.preferred.ams03s09.v16.lscache7.c.youtube.com/videoplayback?upn=WwhZO1YDSao&sparams=cp%2Cid%2Cip%2Cipbits%2Citag%2Cratebypass%2Csource%2Cupn%2Cexpire&itag=18&ip=84.0.0.0&signature=0130FA2C15C0C4EFBF57D9E082A6242C2E973D67.36AB2CC587BA75274555DA76A122A6A0FAAC327C&sver=3&ratebypass=yes&source=youtube&expire=1334003591&key=yt1&ipbits=8&cp=U0hSSVBLUV9OS0NOMl9IRVVGOlhVRmJ2d3lsM0Uz&id=3265d4d57461001d) should help.

- RM[rv2,cook] IS the link!
On second thoughts I think it's a ffmpeg issue, because LAV Filters has the same issue.

MPC-HC with the internal RealSplitter constantly crashes with:
The instruction at "0x102117c7" referenced memory at "0x00000000". The memory could not be "written".

MPC-HC with LAV Filters constantly crashes with:
The instruction at "0x02a017c7" referenced memory at "0x00000000". The memory could not be "written".

[Aleksoid1978]

It's all in my post :confused:

- RM[rv2,cook] IS the link!
On second thoughts I think it's a ffmpeg issue, because LAV Filters has the same issue.

MPC-HC with the internal RealSplitter constantly crashes with:
The instruction at "0x102117c7" referenced memory at "0x00000000". The memory could not be "written".

MPC-HC with LAV Filters constantly crashes with:
The instruction at "0x02a017c7" referenced memory at "0x00000000". The memory could not be "written".

For me play fine this file http://samples.mplayerhq.hu/real/RLG2_19.rm, no crash, all ok. What filter you use ???

About .MP4 - i download .mp4 file from here - http://www.youtube.com/watch?v=MmXU1XRhAB0 (Crop Circles - Hyperspace Gateways - Feature Length.mp4 - 311Mb). And seeking is perfect.

Always use only internal splitter and decoder, EVR Custom output.

[Reino]

MPC-HC with the internal RealSplitter constantly crashes with:
The instruction at "0x102117c7" referenced memory at "0x00000000". The memory could not be "written".

MPC-HC with LAV Filters constantly crashes with:
The instruction at "0x02a017c7" referenced memory at "0x00000000". The memory could not be "written".That means:
- RealSplitter + [whatever_decoder]
- LAV Splitter + [whatever_decoder]

Does processing that specific file require a SSE2 cpu perhaps?

[ageback]

S.Chinese update to rev.4316
http://dl.dropbox.com/u/132862/mplayerc.sc.rc.txt

[pancserzso]

Sorry for asking a question which must have been asked 100s of times, but I found really confusing information regarding these settings.

My question is that
1. What's the difference between Normalize vs. Regain Volume
2. Is the Boost slider only useful when regain volume is activated, or it has other functions too?
3. If it's only connected to regain volume, why it not disabled when regain volume is not activated? It is really confusing this way.
4. What is the recommended way of making a quiet recording sound at normal volumes. Normalize works perfectly, but in the past I have recognized some audiable distorsions using normalize, so I stopped using it. Should I use regain with +6 db if I want to make it sound louder?

Sorry for asking something which must have been answered before, but I found really confusing answers.

[LigH]

1.: "Normalize" does like you might have learned in maths, vector calculus: Ensure that the maximum and minimum sample values do not exceed the value range (in floating point: -1.0 .. +1.0; for 16 bit integer: -32768 .. +32767), to avoid clipping while optimizing the volume. For an optimal result, this would in fact require two steps: first a scan to find the extrema and to calculate the optimal attenuation factor, then a second for the attenuated playback. But who would like to wait several minutes for the first pass? Therefore another approach is used which continouosly adapts the attenuation factor while playing the audio (See elsewhere: AGC = Automatic Gain Control). The "Regain volume" option resets the factor each time you seek in the media while the 1-pass adaptive normalization is active.

2.: The Boost factor is a manual factor to multiply the audio. But if normalization is active, it will be scaled back to the same range again. "Regain volume" is an option which depends on the "Normalize" option. If you don't normalize, there is no attenuation being adapted which could be reset while seeking, only the fixed boost factor.

3.: Depending on the selected boost factor, you may notice more or less how much the gain is reset while seeking. But after a little while, the normalization will adapt the attenuation to the same range again.

4.: The perfect way is to record already with a well balanced volume. The second best way is to normalize the audio stream separately in two passes; with or without a dynamic compression is a matter of your personal taste. Several years ago, BeSweet was the preferred tool to convert audio streams with applied normalization (and I once added the "Boost" per-sample dynamic compressor). Today you can do similar things in many other audio convertors, even in MeGUI via AviSynth functions (using the BeHappy engine).

[fastplayer]

"Regain volume" is an option which depends on the "Normalize" option.
In other words: "Regain volume" should be greyed-out when "Normalize" is disabled. Correct? If so, can somebody fix this in the UI?

[LigH]

Well, provided that I understood its purpose correctly. I didn't make it...

[aufkrawall]

What about H.264 I444 output?
MPC HC can play it but it looks like 4:2:0 CSS.
The current builds of ffdshow can deal with it with no problems.
It'd be really nice if MPC HC would support real H.264 I444 output.

Edit: Maybe simply converting AYUV to RGB would be the best fix?
Since EVR can't get along correctly with AYUV.
Nev will implement such a thing for LAV.

[JanWillem32]

AYUV for 8-bit 4:4:4 isn't hard at all to mix. It's bit-compatible with the well-supported X8R8G8B8 surface format, but the GPU drivers currently don't include mixing support for it (you can use DXVA checker to see the video processor options).
A custom mixer is in production. I've already slapped on the ATi chroma fix filters in the experimental builds (technically a mixer part), but I lack experience to further implement the DirectShow and MediaFoundation interfaces required to write a full mixer. I'm not so worried about keeping DXVA compatibility, but a custom mixer does complicate things in terms of supporting deinterlacing. A weave or bob deinterlacer is easy to add natively, but I'm probably not going support the adaptive types that are currently handled in the external EVR and VMR-9 mixers.
I might also just reserve the custom mixer for only the (early experimental) DirectX 11 renderer, which is pretty incompatible with EVR and VMR-9 anyway, so that I don't have to juggle with the shared rendering paths of EVR CP, VMR-9 r., the DirectX 9 handler of Quicktime and the DirectX 9 handler of RealMedia.

Anyway, simply adding a sub-filter for the mixer handler in EVR CP is impossible in its current state. If I would add a custom mixer for the DirectX 9 components, I very much doubt that I'll tolerate the plethora of support functions and code branches in the presenter part to support the four other mixers. (A similar issue exists for fixing/merging EVR Sync. I really don't know what to do with it. My predecessor's work also botched up some functions, which doesn't help either.)

[aufkrawall]

Huf, your posts exceeds my comprehension level. :eek: :D
But I think it's rather a decoder issue since it works with ffdshow and MPC's EVR CP renderer.

Btw: H.264 RGB is now a bit better, the colors aren't messed up totally anymore.
It just still converts to 4:2:0 output just like with I444 input.

I'm just a layman, but I suppose it really can't be so hard to fix.

[dukey]

AYUV for 8-bit 4:4:4 isn't hard at all to mix. It's bit-compatible with the well-supported X8R8G8B8 surface format, but the GPU drivers currently don't include mixing support for it (you can use DXVA checker to see the video processor options).


Sure they do, just not on all graphics cards. My nvidia 9800 works with AYUV. I know cause I tested it myself :p

DirectX 11 renderer, which is pretty incompatible with EVR

See this:
http://msdn.microsoft.com/en-us/library/windows/desktop/ms704630%28v=vs.85%29.aspx

Not sure h/w accel will work with DX11 though. I can't see the benefit of using DX11 over D3D9EX. You need geometry shaders for rendering video ?

[ryrynz]

What's the chance of getting the /noicons switch on the installer like with ffdshow?

[ceb]

I have a video file with 2 subtitles, is it possible to display both of them at the same time?

[nevcairiel]

I'm not so worried about keeping DXVA compatibility, but a custom mixer does complicate things in terms of supporting deinterlacing. A weave or bob deinterlacer is easy to add natively, but I'm probably not going support the adaptive types that are currently handled in the external EVR and VMR-9 mixers.

If deinterlacing (and DXVA) is not going to be supported anymore, you can throw away your work now and stop trying.
EVR becomes pretty much useless when you rip out some of its key features.

Honestly, every time i read one of your posts, the chances of your work being merged back into MPC-HC seem so much smaller yet again.
If you want to write a completely new renderer (based on EVR or not), then feel free to do so, but know that the only chance of being remotely useful is that it needs to support at least the basic features the other renderers offer.
If you still want to modify the existing EVRs, then its 100% clear that you cannot remove any existing features (especially important ones like Deinterlacing)

[JanWillem32]

@aufkrawall: Sorry for that, it's just annoying that I haven't been able to get a mixer to be able to universally accept AYUV, 10- and 16-bit input types. I can probably get a decent set of basic mixer sub-filters working, but getting all mixer functions to work properly is biting off more than I can chew on right now.

@dukey: So Nvidia added it natively (not the software AYUV to X8R8G8B8 converter in EVR)? Good to know, thank you.

A DirectX 11 device will probably be accepted, but needs a full set of SharedResource queue functions to provide synchronization between multiple devices. That usually means a lot of trouble.
A quick search reveals that that's indeed the case: http://social.msdn.microsoft.com/Forums/en/mediafoundationdevelopment/thread/b2aafe7a-aff8-44c2-b1ba-03de663e02db . When using it with a native DirectX 11 device it requires outfitting with a full mixer and presenter. That still disables all video processor devices of the EVR of course.

Geometry shading is quite a nice feature for dynamically creating vertices. The vertex contents in a video renderer are generally very static. Although I had some fun with the standard teapot model in an earlier version of the renderer, I don't even include vertex shading for the video renderer at the moment (all vertices are pre-transformed and used raw).
Nearly three years ago, I started to specialize in DirectX 10 rendering. I learned to use the DirectX 9 counterparts later on, mainly for this project.
I generally create DirectX 11 devices, at a DirectX 10.0 support level. (Naturally, if the hardware supports higher, I'll use the native supported feature level for for example the pixel shader compiler.)
The DirectX 10+ features I mainly like over the DirectX 9 counterparts are:
-arrays of textures, and other resource type features (many high- and low-level memory operations are absent in DirectX 9)
-using the capabilities of a modern shadercore: http://en.wikipedia.org/wiki/HLSL (A bit old and incomplete, but the difference between base DirectX 9 level 2.0 and base DirectX 10 level 4.0 is big. A different view of the support levels: http://msdn.microsoft.com/en-us/library/ff476876%28v=vs.85%29.aspx#Overview .)
-asynchronous rendering queue handling (it's pretty weak under DirectX 9)
-the debug layer interfaces
-no implicit swapchain creation (DirectX 9 devices always have an implicit swap chain)
-output to multiple backbuffers: http://msdn.microsoft.com/en-us/library/bb174570%28v=vs.85%29.aspx (This is partially implemented in DirectX 9, and some more parts in 9Ex, but the interfaces are messy and don't support fullscreen backbuffer queues at all.)
-resizing of backbuffers when an associated window resizes (DirectX 9 requires a new swap chain, Reset() or ResetEx() for that)
-native 3D presentations (this is only available on DirectX 9 through a few plugins)
-a decent amount of rendertargets (The maximum allowed under DirectX 9 is 4, which limits techniques like those used for constant frame interpolation. I initially wrote that function to handle 8, and it didn't work because of that.)
-Hmm, if I dig deeper into the matter, I'll probably find some more things.

@nevcairiel: I don't expect that sacrificing DXVA compatibility is required. The reference device created by the DXVA2CreateDirect3DDeviceManager9 functions seems simple enough to handle with a basic synchronized queue (it outputs to normal DirectX 9 NV12 surfaces).
Deinterlacing in EVR is a combined sub-filter together with color conversion and strongly binds to only a few mixer input types, typically NV12, YUY2 and UYVY. It behaves like a fixed-function pipeline.
Setting up a mixer that only has to do chroma up-sampling and color conversion would be a good 'step 1' in terms of development. Adding weave and bob deinterlacing should be easy too (as I did mention). Those only require simple line-by-line pixel copies. Integrating adaptive types of deinterlacing would require extensive (and heavy) pixel shading when not using the EVR mixer.
Trying to use deinterlacing from the EVR mixer using a custom mixer is perhaps a nice idea for a 'step 4' or something (if it can work at all). It's a lot more complicated to get external sub-filters to work (and to get familiar with the limitations of those).

A versatile custom mixer and presenter will always be heavier than the standard (but mostly fixed function) EVR. Heavy pixel shading to handle deinterlacing will probably pose a problem for low-end GPUs, and offloading it to the CPU isn't any better.
The point is, when setting a custom presenter, you will have to handle the output from the mixer and put it on screen yourself. The EVR won't do that for you when you set a custom presenter.
The same thing goes for the mixer. When setting a custom one, the EVR isn't going to do the mixing for you anymore. Development on a full custom mixer has to start from the bottom up then.
That's mostly why I mentioned to only add a mixer to the DirectX 11 version, the renderer is heavier to begin with. Using heavy shading is less of an obstacle if the GPU provided is at a base DirectX 10 level (i.d. November 30, 2006) rather than at a base DirectX 9 level (i.d. December 19, 2002).

Anyway, I'm still at 'step 0' right now: getting a properly working allocator-presenter working for the shared renderer and EVR Sync. That's the main focus and also the reach to what the tester builds I made include. Any other things will have to be developed, tested and integrated on a later iteration. I think a similar approach like the one used for developing EVR Sync would be best for that; making a working copy of the shared renderer and basic EVR mixer handler first, and then editing it as a separate renderer. I didn't imply removing the current DirectX 9 renderers (with mixer handlers).

[CruNcher]

@ Jan
you have big expertise in Directx related things did you ever considered to bring in your knowledge into Firefox too Bas Shouten http://www.basschouten.com/blog1.php i guess could really need someone like you helping him improve the Direct2D Performance part of Azure http://blog.mozilla.org/joe/2011/04/26/introducing-the-azure-project/ :)
There are certain things unclear of how Microsoft handles Performance in IE9/10s Direct2D implementation and he basically tries to get this all done by himself, though sure Firefox itself is much much more complex then MPC-HC primary target you have much more different components to handle here. :)


A versatile custom mixer and presenter will always be heavier than the standard (but mostly fixed function) EVR. Heavy pixel shading to handle deinterlacing will probably pose a problem for low-end GPUs, and offloading it to the CPU isn't any better.

Not only the fact that no Player yet except commercial ones can make use of Intels Hardware Scaling is a big drawback especially if you can guess that Scaling will get better in different Hardware over time concentrating to much on Shader based Scaling for these new Generation is imho a backturn in a time where Scaling was handled badly by Vendors on Hardware (Intel Showed that this doesn't have to be with their Lanczos implementation in Hardware away from the old Billinear to the Bicubic Scaler (heavy Shader load) we get used too and trying to compensate since then with Shader based Scaling), Players need a more flexible system now to allow Hardware Scaling and Shader based Scaling together depending on the Hardware in use :).
And especially Intel is on the trip to exclusively support only NT 6 later on they even killed of many NT 5.1 support in their SDKs as they also came in late and fully concentrate on NT 6 support these days.

[dukey]

Deinterlacing in EVR is a combined sub-filter together with color conversion and strongly binds to only a few mixer input types, typically NV12, YUY2 and UYVY. It behaves like a fixed-function pipeline.

It is really. It's a thin wrapper around the hardware, and it's designed like that for maximum efficiency. You need to simply things a bit. I mean the fact you support so many renderers must be a support nightmare.

[madshi]

A DirectX 11 device will probably be accepted, but needs a full set of SharedResource queue functions to provide synchronization between multiple devices. That usually means a lot of trouble.
Indeed it does. madVR is currently broken on the latest NVidia driver revision, because there appears to be a bug in those drivers, so basically resource sharing doesn't work properly at the moment.

native 3D presentations
You mean frame packed HDMI 1.4 output? AFAIK this is planned for D3D11.1 / win8, but not yet available for D3D11 / win7.

[CruNcher]

dukey yeah i guess it would be time to cleanup what would be really needed and what is really no target anymore these days to get code complexity down somewhat :)

@madshi
If i understood this bug would also mean some scenario like using Quicksync for Decoding (Intel IGP) and Nvidia Discrete for Rendering (Madvr) also fail currently with Nvidias new Driver ?

[madshi]

@CruNcher, nope, no problem with that.

[nevcairiel]

Indeed, QuickSync works just fine, and has no influence on the madVR problem.

[CruNcher]

Madshi so the sharing between different APis isn't affected ? http://i.msdn.microsoft.com/dynimg/IC534123.png so is this more a multimonitor (output) issue ?

[madshi]

The sharing of resources between different devices (even with the same API, e.g. D3D9Ex <-> D3D9Ex) is affected by the NVidia bug, as far as I can say. But Intel QuickSync does not use that feature.

[CruNcher]

ahh ok i see hmm this would mean Nvidias Encoder would be also affected when sharing D3D surffaces of the Decoder i guess when in D3D interoperability mode :(

[ageback]

S.Chinese. Add accelerator "检查更新(&U)".
http://dl.dropbox.com/u/132862/mplayerc.sc.rc.txt

[betaking]

T.Chinese. Add accelerator "檢查更新(&U)".
http://www.mediafire.com/?1783urvh61vaqo1

[betaking]

Japanese. Add accelerator "更新を確認する(&U)".
http://www.mediafire.com/?3aqfe5i34f4puru

[wanezhiling]

http://forum.doom9.org/showpost.php?p=1569925&postcount=10376

http://i.imgur.com/Y0XPw.png

[Aleksoid1978]

http://forum.doom9.org/showpost.php?p=1569925&postcount=10376

http://i.imgur.com/Y0XPw.png

Fixed at rev.4375

[khagaroth]

Updated Czech translation (https://sourceforge.net/apps/trac/mpc-hc/ticket/1505).

[ceb]

I have a video file with 2 subtitles, is it possible to display both of them at the same time?
Any help? :)

[v0lt]

I have a video file with 2 subtitles, is it possible to display both of them at the same time?
There is no such feature.

[jmartinr]

I have a video file with 2 subtitles, is it possible to display both of them at the same time?
You can use internal subtitles and subtitles form an external file, with a filter that displays the internal subtitle (for example ffdshow) and a filter that displays the external subtitle (mpc's own filter).

[JanWillem32]

you have big expertise in Directx related things did you ever considered to bring in your knowledge into Firefox too Bas Shouten http://www.basschouten.com/blog1.php i guess could really need someone like you helping him improve the Direct2D Performance part of Azure http://blog.mozilla.org/joe/2011/04/26/introducing-the-azure-project/ :)
There are certain things unclear of how Microsoft handles Performance in IE9/10s Direct2D implementation and he basically tries to get this all done by himself, though sure Firefox itself is much much more complex then MPC-HC primary target you have much more different components to handle here. :)Alas, Direct2D is something I actually know very little about. My expertise is mostly 3D graphics and art, audiovisual mastering, fundamental math/programming problems and generally tweaking my PC. I can maybe help with some tips involving the floating-point object space transforms that were mentioned, but that's about the reach of what I can contribute. (I'm also quite limited in available free time to develop.)Not only the fact that no Player yet except commercial ones can make use of Intels Hardware Scaling is a big drawback especially if you can guess that Scaling will get better in different Hardware over time concentrating to much on Shader based Scaling for these new Generation is imho a backturn in a time where Scaling was handled badly by Vendors on Hardware (Intel Showed that this doesn't have to be with their Lanczos implementation in Hardware away from the old Billinear to the Bicubic Scaler (heavy Shader load) we get used too and trying to compensate since then with Shader based Scaling), Players need a more flexible system now to allow Hardware Scaling and Shader based Scaling together depending on the Hardware in use :).I can post a modified player that can use the EVR internal resizer. Unfortunately, the system is severely bugged with any other output than X8R8G8B8 on my system and the mixer doesn't allow downsizing pictures as of yet. It looks like yet another feature that needs a set of warnings, error messages and such. I'll see how it turns out.You need to simply things a bit. I mean the fact you support so many renderers must be a support nightmare.True, the code is heavily branched; at compile time (MMX/SSE/SSE2, x86/x64 things), at runtime (4 mixers, DX9+DX9Ex devices, operating system support, GPU features, et cetera) and for the sets of different renderer functions. I've just counted the main shared renderer function CDX9AllocatorPresenter::Paint(), and it's 2191 lines in size. (And yes, I did make all the big constant data blocks, such as the pixel shaders and ditherer external.) Rendering loops are naturally big (lots of debug code, and many sequential sub-functions), but this function is by far the largest I've ever worked with. It's still a lot better organized than the mess that the original renderer contained, but multiple mixers, and having to support Windows XP to 7 (and 8 coming) with the different API support levels is a lot of work to get right. My own system isn't perfect for testing either. For example: the original VSync functions never scheduled properly for me. I added four more scheduling modes, and I'm starting to see improvement (finally).You mean frame packed HDMI 1.4 output? AFAIK this is planned for D3D11.1 / win8, but not yet available for D3D11 / win7.Ah, so you've also been reading up on the SDK beta. It's indeed new, but it should be an improvement over the current plug-in interfaces. (I don't have experience with stereo output rendering yet.) I'm hoping that the 11.1 interfaces will be ported back to Vista and 7, similar to the previous DirectX 11 runtime for Vista. I'm also hoping that there will be samples of how to shift from Metro Desktop to full screen exclusive mode (I haven't found a method yet).

[dukey]

Alas, Direct2D is something I actually know very little about. My expertise is mostly 3D graphics

Direct2D is just a wrapper around Direct3D. That's basically it.

[JanWillem32]

There are a few extensions that are generally considered more Direct2D than Direct3D. Think of things like the font and vector drawing functions from the older GDI that are hardware accelerated in Direct2D now. I haven't worked with any of those yet.

[ceb]

There is no such feature.
Thanks. :)
You can use internal subtitles and subtitles form an external file, with a filter that displays the internal subtitle (for example ffdshow) and a filter that displays the external subtitle (mpc's own filter).
I'm not sure I understand. My video file is an .mkv and both subtitles are inside it. That said I use only MPC-HC and prefer to not install other stuff. Thanks for the reply though. :)

[ajp_anton]

Two old feature requests that nobody commented on:
1. If video size is within x % of resized fullscreen size, don't resize at all.
2. In fullscreen monitor mode, allow refresh rates that aren't in Windows' "hide modes that this monitor cannot display" list, but actually work anyway. madVR supports these.

[Snowknight26]

r4414:
DXVA on:
http://stfcc.org/pics/i/56fc19a83b676969e2ffc80ca9164d2a_th.jpg (http://stfcc.org/pics/i/56fc19a83b676969e2ffc80ca9164d2a.png)

DXVA off:
http://stfcc.org/pics/i/c28a4f250dd6332b2fd36204e710ce6d_th.jpg (http://stfcc.org/pics/i/c28a4f250dd6332b2fd36204e710ce6d.png)

Only happens on the last frame. Sample: http://stfcc.org/misc/shensavestheday.final.mkv (960x540, 60fps H.264)

[Aleksoid1978]

r4414:


Fix at rev. 4415;

[Snowknight26]

Speedy fix, thanks!

[Polcius]

I just use MPC-HC (revisions that are in xvidvideo.ru) to play my movies; pretty much with default settings (EVR-CP, Bilinear resizing, etc.).

Using Windows 7, PC connected to a Samsung LED through HDMI.

In the ATI CCC I can choose between different "Pixel Formats":

- YCBCR 4:4:4
- YCBCR 4:2:2
- RGB Studio (Limited)
- RGB Full Range

Which should I choose?

Thanks

[FDisk80]

I just use MPC-HC (revisions that are in xvidvideo.ru) to play my movies; pretty much with default settings (EVR-CP, Bilinear resizing, etc.).

Using Windows 7, PC connected to a Samsung LED through HDMI.

In the ATI CCC I can choose between different "Pixel Formats":

- YCBCR 4:4:4
- YCBCR 4:2:2
- RGB Studio (Limited)
- RGB Full Range

Which should I choose?

Thanks

Why people use Bilinear resizing? Bicubic A -0.75 looks much better imho. Sharper image and less aliasing effect.

I use - RGB Studio (Limited) anything else looks just to dark for me and totally kills the shadows.

[droc]

I use YCBCR 4:4:4 and xvYCC on Plasma.Very cool!

[Polcius]

But what's the difference between them? Is there a "right one" for watching movies?

[nevcairiel]

But what's the difference between them? Is there a "right one" for watching movies?

The right one depends on your TV.

[Polcius]

The right one depends on your TV.

What should I look for?

TV is a Samsung UE32D5500 (LED)

[HoP]

how i can "Deselect All" formats?why there is no button for that?i hate clicking and deselecting :D

http://upit.cc/i/085d0b9a.png