madVR - high quality video renderer (GPU assisted)

[StinDaWg]

Have you checked your DPC latency? If you get big spikes there, you get choppy playback no matter how much free CPU and GPU you have.
And perhaps XBMC is better at handling big spikes in DPC than your madVR setup is.
You can check your DPC latency with this program (http://www.thesycon.de/deu/latency_check.shtml).

Can't use it.
Windows 8 Compatibility: The DPC latency utility runs on Windows 8 but does not show correct values. The output suggests that the Windows 8 kernel performs badly and introduces a constant latency of one millisecond, which is not the case in practice. DPCs in the Windows 8 kernel behave identical to Windows 7. The utility produces incorrect results because the implementation of kernel timers has changed in Windows 8, which causes a side effect with the measuring algorithm used by the utility. Thesycon is working on a new version of the DPC latency utility and will make it available on this site as soon as it is finished.

[Qaq]

StinDaWg, read this: http://forum.doom9.org/showpost.php?p=1673653&postcount=25003

[BetA13]

i dont know if this is teh right place here, but i think its madvr, or maybe not... not shure really.

ok, so my MPCHC+MadVR (LAV filters) the video fps in windowed or fullscreen is allways what the HZ of the monitor is.. 60fps@60hz for my tv and 75hz/75fps for my monitor..

now, i think thats kinda wasting gpu power, or is it?
i cant find teh option top turn this behavior off..i cant even remember if i aktivated it somewhere, somehow, or if it was allready there form teh start..

so i tested that MPDN player and there the fps gets limited to what the movie was recorded, in order 23.something , 30 fps etc...

so, my question i guess, is, is there any benefit of using high FPS rate? and how the hell do i change this behavior??

like i sayd, not shure if this is the right thread fo rthis..But hey, asking doesnt cost ya anything...

greetz BetA

[upyzl]

Hi all,
sorry to interrupt, but I want to know
is there anything news about madVR x64 ?

[namaiki]

i dont know if this is teh right place here, but i think its madvr, or maybe not... not shure really.

ok, so my MPCHC+MadVR (LAV filters) the video fps in windowed or fullscreen is allways what the HZ of the monitor is.. 60fps@60hz for my tv and 75hz/75fps for my monitor..

now, i think thats kinda wasting gpu power, or is it?
i cant find teh option top turn this behavior off..i cant even remember if i aktivated it somewhere, somehow, or if it was allready there form teh start..

so i tested that MPDN player and there the fps gets limited to what the movie was recorded, in order 23.something , 30 fps etc...

so, my question i guess, is, is there any benefit of using high FPS rate? and how the hell do i change this behavior??

like i sayd, not shure if this is the right thread fo rthis..But hey, asking doesnt cost ya anything...

greetz BetA
It's a feature of the new fullscreen exclusive mode. Also, it would be even better if you enable MadVR's smooth motion mode to take full advantage of it.

[ryrynz]

Hi all,
sorry to interrupt, but I want to know
is there anything news about madVR x64 ?

A search of the forum would have provided you with an answer..

Not until a 1.0 release.

[upyzl]

Not until a 1.0 release.
ah... thx
A search of the forum would have provided you with an answer..
searched results either months ago or irrelated too much...

[huhn]

i dont know if this is teh right place here, but i think its madvr, or maybe not... not shure really.

ok, so my MPCHC+MadVR (LAV filters) the video fps in windowed or fullscreen is allways what the HZ of the monitor is.. 60fps@60hz for my tv and 75hz/75fps for my monitor..

now, i think thats kinda wasting gpu power, or is it?
i cant find teh option top turn this behavior off..i cant even remember if i aktivated it somewhere, somehow, or if it was allready there form teh start..

so i tested that MPDN player and there the fps gets limited to what the movie was recorded, in order 23.something , 30 fps etc...

so, my question i guess, is, is there any benefit of using high FPS rate? and how the hell do i change this behavior??

like i sayd, not shure if this is the right thread fo rthis..But hey, asking doesnt cost ya anything...

greetz BetA

the first line in madVR OSD shoes your monitor refresh rate.
with default settings this is not changed what so ever usually 60 FPS. soa normal movie with 23 (24000/1001) judders liek crazy at 60 hz. that's why you should always match your monitor refresh rate with the movie refresh rate.

you can set it up under devices -> "device name" -> display modes.

MPDN has the same "problem" it simple shows the video frame rate of the movie in HZ not in ms like madVR but it is totally the same.

to anser your question it is best to match the refresh rate so high FPS is not better but if this is not possible you should use madVR smooth motion and the highest possible refresh rate.

usually no monitor can do 23p so smooth motion should be used.

[Asmodian]

usually no monitor can do 23p so smooth motion should be used.

Lots of monitors will do 48 or 72 Hz if forced to, these work well without smooth motion too. There is very little performance benefit to running the monitor at a lower refresh rate unless smooth motion is on and smooth motion looks better at higher refresh rates.

Personally, on a 60 Hz monitor, I leave the screen's refresh rate alone and use smooth motion.

[BetA13]

hmmm, ok... i still dont find an option to turn it off..Is it the exclusive fullscreen? so, if i turn that off its normal fps?

the thing is, its also in windowed mode..
heres a pic with afterburner and and madvr info.

so there you see its 75fps actually inwindowed and fullscreen mode.. i have SMOOTH motion enabled since it came out, since it really makes a difference...

anyway.. pics:

http://abload.de/img/dfgdfgdgnts7k.jpg
http://abload.de/img/tzrrjtrj8csum.jpg

if i wanna load up an real blueray with 40gb, i cant archive the 75fps, gpu usage is at 80 - 99% and the frames keep droping..soo, if the movie was made in., lets say 30fps, isnt it better then to watch it in 30fps?
so the gpu isnt loaded as much and i DONT get dropped frames etc etc..

can someone please make a screenshot of the option? so i can disable this "allways" full fps behavior..

Thanks...

[namaiki]

The frame has to change at the same time as the refresh rate or how does smooth motion work otherwise? Don't disable fullscreen exclusive mode. Probably lower your scaler for chroma or disable DXVA if you are using it. Also try unhook that GPU statistical utility from MadVR/MPC-HC.

[huhn]

hmmm, ok... i still dont find an option to turn it off..Is it the exclusive fullscreen? so, if i turn that off its normal fps?

the thing is, its also in windowed mode..
heres a pic with afterburner and and madvr info.

so there you see its 75fps actually inwindowed and fullscreen mode.. i have SMOOTH motion enabled since it came out, since it really makes a difference...

anyway.. pics:

http://abload.de/img/dfgdfgdgnts7k.jpg
http://abload.de/img/tzrrjtrj8csum.jpg

if i wanna load up an real blueray with 40gb, i cant archive the 75fps, gpu usage is at 80 - 99% and the frames keep droping..soo, if the movie was made in., lets say 30fps, isnt it better then to watch it in 30fps?
so the gpu isnt loaded as much and i DONT get dropped frames etc etc..

can someone please make a screenshot of the option? so i can disable this "allways" full fps behavior..

Thanks...

madVR is only scaling 24 fps in these screens so everything is working as intended. SM takes these frames 24 frames and "makes" 75 fps out of it this task is every easy for a GPU so this shouldn't be the problem so there is no always full FPS option.
and smooth motion shouldn't be the problem for dropped frames.
in the screen the GPU usage is low do you use nnedi3? this may trigger only in fullscreen with a 720p source.

[GCRaistlin]

Is it worth to upgrade from NVIDIA 8800 GT to GTX 260/280/275 for madVR? What will give the best results?

[Soukyuu]

I'd say no, or maybe it depends on the content. I have a 260 GTX and it's not really doing more than my 8600m GT. Error diffusion or NEEDI is impossible, and both cards handle jinc3+smooth motion well enough. Don't remember about debanding.
I mostly watch 720/1080p@23,976fps stuff on a 1080p screen.

[GCRaistlin]

Hm, that sounds strange 'cause the comparison (http://www.hwcompare.com/1835/geforce-8800-gt-512mb-vs-geforce-gtx-260/) shows a relative big advantage of GTX 260 over 8800 GT. What is the good candidate to upgrade to then? What should I pay attention at - memory bandwidth, texel rate, pixel rate, core speed, shader speed, number of shaders, etc?

[huhn]

just get a modern card you can pay for.

if it has 128 bit GDDR5 you good for normal usage and nnedi3 for SD is possible.
usually cheap AMD cards are better than nvidia cards in the same prize pool for nnedi3.

but what do you want to do?

i don't see any reason to buy something as old as a nvidia 2xx card.

[GCRaistlin]

huhn
The reason is the price. I can buy a used GTX 260 for $25. I've been guided by comparisons I gave the link to so I've found that GTX 260 has the best price/speed ratio. If it isn't true for madVR what can you say about GTS 450?

usually cheap AMD cards are better than nvidia cards in the same prize pool for nnedi3
I run into a trouble with AMD HD3650 - when I connect my monitor's HDMI input to the card's DVI output (with DVI-HDMI cable) I got black frame around the picture and the screen resolution isn't really 1920x1080 (it may be noticed) while Windows says it is. When I connect my monitor's DVI input to the card's DVI output (with DVI-HDMI cable + DVI-HDMI converter) all is OK. And it is the card's issue 'cause 8800 GT gives the normal image in both cases. So I have some distrust to AMD but I'd be thankful for any advice. What is the minimum AMD card it is worth to upgrade to from 8800 GT?

[flashmozzg]

huhn
The reason is the price. I can buy a used GTX 260 for $25. I've been guided by comparisons I gave the link to so I've found that GTX 260 has the best price/speed ratio. If it isn't true for madVR what can you say about GTS 450?


I run into a trouble with AMD HD3650 - when I connect my monitor's HDMI input to the card's DVI output (with DVI-HDMI cable) I got black frame around the picture and the screen resolution isn't really 1920x1080 (it may be noticed) while Windows says it is. When I connect my monitor's DVI input to the card's DVI output (with DVI-HDMI cable + DVI-HDMI converter) all is OK. And it is the card's issue 'cause 8800 GT gives the normal image in both cases. So I have some distrust to AMD but I'd be thankful for any advice. What is the minimum AMD card it is worth to upgrade to from 8800 GT?

You should check your scaling options in Catalyst Drivers and set it to 0% so that the picture would take all screen size without any borders.

[huhn]

we can't recommend anything if we don't know what you your goal with a new card is.

about AMD black bars see above easy to fix but yeah bad work from AMD.

[GCRaistlin]

we can't recommend anything if we don't know what you your goal with a new card is.

My goal is just more aggressive madVR settings than 8800 GT allows.

[StinDaWg]

Is it worth to upgrade from NVIDIA 8800 GT to GTX 260/280/275 for madVR? What will give the best results?
The minimum to do NNEDI3 upscaling for 720p60->1080p60 is AMD 7850. I'm not sure what the Nvidia equivalent is, but it's probably around a 750ti. These are $100 cards. You don't need to go any higher than this if you don't need to do 4K.

I can do debanding, smooth motion, film deinterlacing, ect as well with 7850.

[Stereodude]

Samples, please. Lots of samples, please. And please cut them generously, so that there's enough smooth playback before the problem starts, and enough playback time with the problem visible. Thanks!
I finally got motivated and was able to create a short sample so you can see the problem without a 3.5GB download and watching 20+ minutes of programming from the beginning without seeking to see the problem. As a result, I created a bug tracker entry (http://bugs.madshi.net/view.php?id=239) (I didn't realize there already was an entry (http://bugs.madshi.net/view.php?id=213) on it when I made it). Most importantly included a 2 minute sample which shows the issue twice. I can create more samples, but I'm not sure they'd help shed any more light on the problem.

[Schwartz]

The 2XX are very performant, but they're old tech and have quite a hefty power drain and heat buildup compared to more recent midline cards.

I ran a 275 for years and it worked fine for madVR. It could run high quality options, smooth motion, error diffusion, debanding and Spline up- and downscaling just fine, but not much in the way of NNEDI.

[Soukyuu]

Is the difference between an (overclocked) 260GTX and 275GTX that high? Error diffusion is definitely impossible for me, debanding + smooth motion work fine though.
Or is the 260GTX just barely underpowered to handle debanding+smooth motion+error diffusion?

[aufkrawall]

ErrorDiffusion uses DirectCompute, which is part of DirectX 11.
You need a DirectX 11 capable card (Fermi and Cypress or newer) for this.

[mindz]

Is there an update coming soon? Anyone know the status of this project?

[Soukyuu]

ErrorDiffusion uses DirectCompute, which is part of DirectX 11.
You need a DirectX 11 capable card (Fermi and Cypress or newer) for this.In that case, Schwartz's statement about their 275GTX running error diffusion is false, then.

[Thunderbolt8]

Is there an update coming soon? Anyone know the status of this project?probably soon™

[StinDaWg]

Is there an update coming soon? Anyone know the status of this project?
madshi is on permanent vacation. lol.

It's just us here shooting the breeze for now. :)

[aufkrawall]

In that case, Schwartz's statement about their 275GTX running error diffusion is false, then.
There's likely a fallback to more simple or none dithering.
You can check this by setting display output to a low bit depth.

I'm on a 9600 GT currently, and when I choose Error Diffusion, rendering time doesn't change.
Since under some conditions even my former GTX 780 Ti@1.24Ghz couldn't keep pace with NNEDI + Error Diffusion, I think it's highly unlikely that Error Diffusion is working on a 9600 GT. ;)

[Soukyuu]

There certainly is a fallback, I just wish madvr displayed some sort of warning when it does fall back due to hardware incompatibility. I forgot ED needs DX11 to be honest. Coming back to the "is 260GTX worth it over 8800GT", I guess the "no" is even clearer for me now.

[James Freeman]

You really think madshi has dropped this project?

[StinDaWg]

You really think madshi has dropped this project?
Who said that?

[Hprd]

I thought he (madshi) said that he had other projects that are a higher priority atm, and that he'd get back to this when he has time again.

[burfadel]

He will need to add DirectX 12 support ;).

[ryrynz]

I thought he (madshi) said that he had other projects that are a higher priority atm, and that he'd get back to this when he has time again.

madVR will continue to have it's quiet times, it's been like that since development started, it's just how madshi plans his time.

I'm most looking forward is an alternative to NNEDI (Shiandow's SuperRes) and 64 bit support so I can finally move the whole chain along, along with a deblocking function too would be nice for lower quality videos.

Anyone else want a new release of MadVR for xmas? xD

[James Freeman]

Anyone else want a new release of MadVR for xmas? xD

Unlikely.
Madshi's last activity was in May.

[romulous]

Unlikely.
Madshi's last activity was in May.

Nonsense - his last post was literally only a week ago:
http://forum.doom9.org/showpost.php?p=1697372&postcount=203

I think he has the patience of a saint frankly. With all the demanding users here - for a product for which they pay nothing - I would have thrown in the towel long ago if I was him (though nev seems to attract just as many of these people - though at least he is clearly now starting to ignore them now, unlike madshi who continued to answer their posts when he was last generally active).

[madshi]

That said having a configurable delay after a resolution change would still be nice. In my new setup the audio receiver is already outputting sound before the UHD TV has fully connected to the receiver at the new resolution. Sometimes there is around two seconds of audio with no picture, other times its instant.
The problem is that most of the delay is happening inside of the display. That means the HTPC doesn't know when it's safe to restart playback again. Some displays synchronize to a new display mode very quickly, others are very slow. Because of that I don't think it makes sense for madVR to delay playback after a display mode change, because I just don't know how long playback should be delayed. Yeah, of course I could add an option to adjust the delay. But I don't want to add hundreds of new weird options. The settings dialog is already complicated enough as it is...

I recently upgraded to Windows 8.1 and now i have a problem with madvr. I always used exlusive mode because to my eyes it gives me better PQ than without it, dont know why. But now with windows 8.1 when i go FSE i feel something wrong, while is not shown on CTRL-J, everything is fine there, but when people move or faces move quick i feel its not smooth like it was in windows 7 with madvr FSE and aero disabled... In windows 8.1 i cant disable aero or wtv desktop vsync in use and i think its competing with madvr own vsync...
But the OSD says you're in FSE mode? Can you post a screenshot of your OSD? You may have to use a digicam, if PrintScreen doesn't work.

Does madVR read the embedded icc profile in Jpeg image with adobeRGB profile so that I can use madVR as a color managed image viewer?
madVR does not have direct access to the source file. madVR is just a renderer, not a source filter or decoder. It's the job of the source filter / decoder to read the metadata of the source file and make it available to the renderer. If the source filter does that, madVR will properly respond.

There is one exception to what I wrote above: Since there are so many different video decoders for MPEG2, VC-1 and h264, most of which don't inform madVR about the proper source gamut etc, madVR is hacking into the source filter so that it's able to extract that sort of information. But that's really an extra feature I implemented and it's not what a renderer normally is supposed to do. And I don't plan on doing this for anything other than the main video codecs.

The 12bpp deep color mode is available since 14.x version. For this card with driver 13.12 it automatically selects 1080p 10 bpp mode to send signals to my 65" Sony TV. But I wonder it makes any noticeable difference with madVR since the dithering is done by GPU --- not madVR itself to 10 / 12 bpp color information.
It should not make any difference, unless maybe you're using the GPU gamma ramps, or unless you've told the GPU to stretch the Windows desktop from 0-255 to 16-235 behind the back of the OS (and madVR).

I know this has been mentioned before in this thread. But to be clear, even if the changelog states so, madVR still uses Intel GMA4600 instead of NVidia GTX 850M (Maxwell).
You can force madVR to use the NVidia GPU for *OpenCL* related features, but only for that. Direct3D does not give me the option to choose the GPU. So most of madVR's work is done on the GPU that the OS assigned to madVR. I don't think there's anything I can do about that.

This video, in particular, is sometimes getting me 40 dropped frames per second with a rendering time under 9ms.
https://www.youtube.com/watch?v=csmwgui5BMk
It plays fluidly for a period, then lags for a period of time, then play fluidly again, etc. Here's a screenshot of the stats
decoder queues are not full your decoder/filter can't handle the file.

split is a copyback split is normally not used on amd so it looks like it is using your igpu.
huhn is exactly right. This looks like a decoder problem to me. If the decoder isn't delivering the frames to madVR fast enough, stuttering is the only possible outcome. Which decoder are you using? Try a different one. Also if you're using hardware decoding, try software decoding instead.

What would cause (5s) max render times to jump up to the 200+ range?
Usually it's either some other process accessing the GPU. Even simple GPU helper tools (e.g. those which record the GPU usage etc) can sometimes have that effect. Some browsers can do that. Or other tools which you might not expect.

Or it could be caused by power switching. If your GPU switches between high and low power/clock speeds, this can cause problems. Maybe in high power mode your GPU has nothing to do (plenty fast) so it thinks "why using high-power mode/clocks"? So it clocks down. But suddenly in clocked down mode it's not fast enough for madVR, anymore, and max render times spike.

Try closing/terminating all processes which are not totally essential. And try forcing your GPU to use high-power mode all the time, to see if either of these fixes the issue.

For a long time I've seen (reported in OSD as opposed to visual) occasional presentation glitches (1 every few seconds) in madVR when playing video, and I'd tried a lot of what was mentioned on the board with no resolution. [...] I found simply changing Jinc Image Upscaling from 3 to 4 taps completely eliminated the presentation glitches.
There's no specific reason why that should help. It's probably simply "luck" that the render times for Jinc 4taps somehow work better in your specific case than Jinc 3taps. This will probably be different for most other users.

Are we talking about FSE mode or windowed mode? If you have presentation glitches in FSE mode, is it possible that your GPU is switching between different power states / clock speeds? It's possible that using Jinc 4taps demands your GPU enough so that it isn't tempted to clock down, while maybe using Jinc 3taps could be so easy to handle for your GPU that it clocks down once in a while, and maybe clocking could produce such presentation glitches. Just guessing here, of course...

To be clear I'm not getting constant stuttering. Just frame drops as often as every couple of minutes, to a few times an hour.
And you're sure that it's not just your refresh rate which is a tiny bit too "slow", so that madVR has to drop a frame once in a while to keep audio and video in sync?

ok, so my MPCHC+MadVR (LAV filters) the video fps in windowed or fullscreen is allways what the HZ of the monitor is.. 60fps@60hz for my tv and 75hz/75fps for my monitor..

now, i think thats kinda wasting gpu power, or is it?
It does "waste" a bit of GPU power, but not as much as you might think. Suppose we have a video with 1fps. If you display that with madVR at 60Hz with the madVR default options, madVR will "present" this frame 60 times per second. However, the frame is still *rendered* only once. Rendering involves things like deinterlacing, chroma upsampling, scaling, artifact removal etc. All of that is only done once for every video frame. However, the fully rendered video frame is kept in memory by madVR and sent to Direct3D for presentation once per VSync interrupt. This presentation at 60Hz does cost a little bit of extra GPU performance, but on the positive side, it allows madVR to present multiple video frames in advance, which makes the whole presentation more stable (less risk of stuttering).

In D3D11 it's possible to achieve a similar effect without re-presenting the frame all the time, but in D3D9 it's not possible.

i have SMOOTH motion enabled since it came out, since it really makes a difference...
Smooth motion FRC (when actually active) *has* to present more frames than the source video has. Technically it wouldn't always have to be the exact refresh rate of the display. But currently it's implemented this way (reason see above). There's no way to do smooth motion FRC with e.g. a 24fps movie and a 60Hz display, without presenting more than 24 frames per second.

[StinDaWg]

Usually it's either some other process accessing the GPU. Even simple GPU helper tools (e.g. those which record the GPU usage etc) can sometimes have that effect. Some browsers can do that. Or other tools which you might not expect.

Or it could be caused by power switching. If your GPU switches between high and low power/clock speeds, this can cause problems. Maybe in high power mode your GPU has nothing to do (plenty fast) so it thinks "why using high-power mode/clocks"? So it clocks down. But suddenly in clocked down mode it's not fast enough for madVR, anymore, and max render times spike.

Try closing/terminating all processes which are not totally essential. And try forcing your GPU to use high-power mode all the time, to see if either of these fixes the issue.

And you're sure that it's not just your refresh rate which is a tiny bit too "slow", so that madVR has to drop a frame once in a while to keep audio and video in sync?
As an update to my situation, I tried windowed mode and I get much better performance than FSE mode. I don't get any dropped frames in windowed mode. This hasn't always been the case though, so I don't know what changed. It could be a Windows 8.1 issue, or an AMD issue. I also get better menu performance in XBMC windowed mode vs FSE, which is the opposite of what I would expect. Present times are ridiculously low in windowed mode, almost 0. Deinterlacing times are lower too (using gpu deinterlacing).

I use ReClock so audio/video sync isn't an issue.

I've checked MSI Afterburner and gpu isn't clocking down.

Is there any downside to using windowed mode over FSE? I can't deal with the constant frame drops, so I'm going to keep using it for now.

[madshi]

The new windowed presentation path added in v0.87.10 does work quite well in Windows 8.1, at least it does for me. That said, FSE mode is still a tiny bit better for me. Not sure where the problems you're experiencing are coming from. But if you have a perfect experience in windowed mode, there's no need to use FSE mode at the moment. Windows 8 desktop composition works fairly well, that does help madVR's new windowed mode.

[leeperry]

I don't think it makes sense for madVR to delay playback after a display mode change, because I just don't know how long playback should be delayed. Yeah, of course I could add an option to adjust the delay. But I don't want to add hundreds of new weird options.
More or less on that same matter, captain obvious to the rescue but I just realized that A/V sync was far more spot-on once I delayed audio by 28ms in ffdshow, which is the input lag figure that was measured in a review of my TV.......I don't think other audio decoders allow for delay and even Reclock's coder refused the idea, wouldn't it make sense to have a delay option in mVR so you could make it positive if needed or negative in order to keep perfect lipsync? I use headphones, there's no delay due to loudspeakers in a room.

[madshi]

Audio delay is usually offered by the AV receiver, or by the media player (e.g. Audio Switcher in MPC-HC). It's not usual practise to offer video delay. Although, the concept does make some sense since the delay is usually introduced by the video side of things (video processing in the display). Anyway, the usual practise is to fix this by doing audio delay in the AV receiver or somewhere else in the audio chain. So I'm not crazy about adding another option to madVR which could be solved through different means...

[Razoola]

The problem is that most of the delay is happening inside of the display. That means the HTPC doesn't know when it's safe to restart playback again. Some displays synchronize to a new display mode very quickly, others are very slow. Because of that I don't think it makes sense for madVR to delay playback after a display mode change, because I just don't know how long playback should be delayed. Yeah, of course I could add an option to adjust the delay. But I don't want to add hundreds of new weird options. The settings dialog is already complicated enough as it is...


I understand and yes you are right it is a problem between the receiver and the display. I personally feel its related to the HDMI 2.0 spec and the repeater bit (http://en.wikipedia.org/wiki/HDCP_repeater_bit). The new 2.0 spec forces the receiver to have to disconnect and reconnect everything. Not nice for the user experience.

If you were to change your mind about this in the future the setting would go nicely on the 'display modes' page of a connected device (where you can choose full screen settings).

[leeperry]

Audio delay is usually offered by the AV receiver, or by the media player (e.g. Audio Switcher in MPC-HC)
In an ideal world with a multichannel AVR but if you are using a stereo analog setup or headphones off a USB DAC then delay isn't an option, ideally it should be set in Reclock but James refused invoking more or less reasons as you just did. Some people don't use MPC and like to avoid having useless middle-men in their DS graph but no biggy, as long as ffdshow works I'll be set :cool:

[skydogis]

Hello to everyone and madshi thanks for your time and effort in this amazing renderer.
I have bumped into something curious.
I have a Dell U2311H display and i have calibrated it using icc profiles from the internet and of course the panel options off the display. So assuming that this is calibrated to 2.2 gamma pure power curve i always selected that (GPU gamma ramps always disabled) in the gamma options of madvr.
But i was playing around a little bit and i found that choosing BT.709 curve 2.2 and then choosing a desired gamma of pure power curve 2.2 gave me a lot more contrast and punchy colours, and after calibrating the contrast and brightness in madvr and using quite a few patterns and of course test-calibration videos there doesn't seem to be any loss of information in the high end (whites) or low end (blacks).
I couldn't find anything about it so i decided to ask away here since i consider this curious considering that my display should be calibrated using a pure power curve but i really dont know since its a pack of settings and an icc profile i got from the net and not my own doing.
Am i missing something here? :thanks:

[Shiandow]

It's quite possible that that looks better. Some people recommend using a 2.4 gamma pure power curve to watch video. Since converting 2.2 BT.709 to 2.2 pure power is somewhat similar to converting 2.2 pure power to 2.4 pure power, the result will look a bit like you're using a 2.4 pure power curve.

You may want to try and see what happens when you convert from 2.2 pure power to 2.4 pure power instead.

Although keep in mind that converting from 2.2 BT.709 to 2.2 pure power is not that similar to converting 2.2 pure power to 2.4 pure power, especially for brighter colours, so YMMV.

[skydogis]

Yeah i tried it
Thats what i was doing in the first place
But BT.709 to pure power gives me more contrast because of the brighter whites and the slightly darker blacks so i thinking that if i calibrate it right, say with Spears & Munsil High Definition Benchmark , i am thinking of leaving it that way.
The curious thing is that the contrast difference is so intense to me that it looks like its a problem with rgb output levels, meaning that pure power curve 2.2 to pure power curve 2.4 looks washed out in comparison to bt.709 to pure power curve, even when i calibrate each profile in the same greyscale pattern and tests.

[madshi]

Using a higher gamma value, or moving from a BT.709 curve to a pure power curve can result in a loss of shadow/highlight detail, if you have a display with a rather bad contrast ratio, and also if you have ambient light. Usually it's recommended to use a higher gamma value (e.g. 2.4) if you don't have a lot of ambient light, and to use a lower gamma value (e.g. 2.2) if you do have a good amount of ambient light, because ambient light usually makes shadow detail harder to see.

I'd suggest that you use a gray scale / black/white level test pattern to double check that all steps from 16-235 are still visibly different from each other. If it's the case then everything's fine.

[skydogis]

Thank you madshi
that's what I observed in my tests
but there is also something else
if I go from pure power curve to power curve
I can see every step of the greyscale clearly except 16
if I do that in bt. 709 to power curve then everything is darker, still distinguishable, but whereas 17-16 is total black I can see the 14 strip which was invisible in the brighter settings, power to power.
Perhaps each curve is designed for different rgb levels then?
I guess power curve would be the way to go for a pc monitor