or Connect
AVS › AVS Forum › Video Components › Home Theater Computers › Let's set this straight - No one can do 24p consistently well
New Posts  All Forums:Forum Nav:

Let's set this straight - No one can do 24p consistently well - Page 39

post #1141 of 1281
Quote:
Originally Posted by Sammy2 View Post

That is one frame out of every 6000 frames. Not very many on the grand scheme of things but with madVR and nVidia with very little time spent tweaking the custom resolution (one iteration) it is more like one frame drop per hour or 2 for an entire 2-hour movie.

Then something must have changed in hardware or drivers. Even after they had supposedly fixed the timing problem for 23.976 source I was seeing quite a few frame drops when running an i3 setup about 12-15 months ago and using the onboard Ivy Bridge video.... and that was after creating a custom refresh rate for 23.976.
post #1142 of 1281
Quote:
Originally Posted by Sammy2 View Post

My contention is that if you cannot see the drops while watching the film why do you want to go looking for them with a tool?

I understand that and agree, I'm just curious is all.
post #1143 of 1281
How is it that people are still perpetuating the myth that you need exactly 24/1.001 out of the video card?

Install Reclock or JRiver and you will never have frame drops.
post #1144 of 1281
Personally I agree with you but:

- This means you cannot bitstream HD audio for decoding in your AVR
- Assumes you are happy to resample audio on the PC - not acceptable to purists
- Assumes you have the TMT DTS audio decoder if you want HD audio on the vast majority of blurays
post #1145 of 1281
Quote:
Originally Posted by Chronoptimist View Post

How is it that people are still perpetuating the myth that you need exactly 24/1.001 out of the video card?

Install Reclock or JRiver and you will never have frame drops.

And never have HD audio either. That must be another thing that only anal retentive people care about.
post #1146 of 1281
"HD audio" is very possible with ReClock or JRivers VideoClock functionality. You just have to switch to lossless decoding instead of bitstreaming, which results in the exact same audio (except in some cases of very cheap HDMI Receivers, which handle Multi-Channel PCM differently than Bitstreaming)

Its a very common misconception that only bitstreaming will give you HD audio, while in reality you can get the full lossless audio just fine when decoding the audio as well.
post #1147 of 1281
Quote:
Originally Posted by jong1 View Post

- This means you cannot bitstream HD audio for decoding in your AVR
Pointless.
Quote:
Originally Posted by jong1 View Post

- Assumes you are happy to resample audio on the PC - not acceptable to purists
Seems like these are the same people that want all audio supplied in formats that extend well beyond the limits of human hearing. (24/96 and beyond)
Quote:
Originally Posted by jong1 View Post

- Assumes you have the TMT DTS audio decoder if you want HD audio on the vast majority of blurays
That's true.
Quote:
Originally Posted by jmpage2 View Post

And never have HD audio either. That must be another thing that only anal retentive people care about.
There is no difference decoding in your PC vs decoding in your AVR.
post #1148 of 1281
Quote:
Originally Posted by Nevcairiel View Post

"HD audio" is very possible with ReClock or JRivers VideoClock functionality. You just have to switch to lossless decoding instead of bitstreaming, which results in the exact same audio (except in some cases of very cheap HDMI Receivers, which handle Multi-Channel PCM differently than Bitstreaming)

Its a very common misconception that only bitstreaming will give you HD audio, while in reality you can get the full lossless audio just fine when decoding the audio as well.

I agree that bitstream and PCM decode are identical performance-wise, however, the only way I am aware of to get PCM output of HD audio material from XBMC when re-sampling the audio is through the use of an external player.

The "easiest" way, for me personally is to simply get a video card that comes as close as possible to 23.976 output.
post #1149 of 1281
Quote:
Originally Posted by Chronoptimist View Post

Pointless.
Seems like these are the same people that want all audio supplied in formats that extend well beyond the limits of human hearing. (24/96 and beyond)
That's true.
There is no difference decoding in your PC vs decoding in your AVR.
As I said, I agree it's the way to go, provided the user has the TMT decoder. But, for sure there are those who won't want to for the reasons mentioned.

And, to be precise about it, if you are using Reclock/Videoclock and so resampling, there IS a difference between decoding on the PC and AVR, even if, depending on the correction being applied, it is most likely inaudible and still, undoubtably HD (for an HD source).
Edited by jong1 - 3/1/13 at 7:53am
post #1150 of 1281
Quote:
Originally Posted by sievers View Post

So I guess cntl+j in mpc-hc reports both types of drops? Are there other tools to see this with?

MadVR reports the refresh rate and the number of dropped frames (both types combined) precisely. Dropped frames by internal glitches are very rare if you configure the player/codec properly. Dropped/repeated frames by a mismatch between the frame rate and the refresh rate do occur periodically and the period can be calculated easily:

Period = 1 / | frame rate* - refresh rate |,

where

frame rate* = frame rate with audio clock/system clock deviation accounted.

For example,

AMD: 23.976 +/- 0.001; the period >= 17min
NVIDIA: 23.971 +/- 0.001, 23.976 +/- 0.001 with custom resolution; the period = 3min-4min, >= 17min with custom resolution
Intel: 23.972 +/- 0.001; the period = 3min-6min.
Edited by renethx - 3/2/13 at 8:53pm
post #1151 of 1281
Quote:
Originally Posted by Sammy2 View Post

That is one frame out of every 6000 frames. Not very many on the grand scheme of things

Well, this does not seem to be a good interpretation. A dropped/repeated frame is recognized as a stutter. There is a stutter every 4 minutes and some people (not everyone of course) are actually seeing it and think it is bothering.
Edited by renethx - 3/1/13 at 7:04pm
post #1152 of 1281
One frame drop every four minutes is not perceivable due to the way the human brain interprets what it is seeing.

As far as using re-clock goes, I'm really not wanting to add another level of complexity to my set up.
post #1153 of 1281
What's the theory (source, please)? Each stutter is surely perceivable. Suddenly it becomes unperceivable when it comes in succession with an interval of a few minutes, according to that theory? That contradicts my experience.

If your brain interprets a stutter every 4 minutes as smooth, that's OK. smile.gif
Edited by renethx - 3/1/13 at 8:43pm
post #1154 of 1281
So far madVR SMFRC seems to be a simple, good solution to this problem. Just set the desktop refresh rate to 59.94Hz (59.94Hz because of video compatibility) and SMFRC removes pulldown judder perfectly, whatever movie is played back (23.976fps, 24.000fps, 25.000fps, telecined or not). Zero dropped frame (ok, this may not be a correct expression), no artifact, no ReClock required. Supporting HD audio bitstreaming, of course.

The only remaining issue is flickering with DXVA2 upscaling (DXVA2 up/downscaling is the best and practically only option for Intel HD Graphics).
Edited by renethx - 3/2/13 at 12:58am
post #1155 of 1281
Quote:
Originally Posted by renethx View Post

What's the theory (source, please)? Each stutter is surely perceivable. Suddenly it becomes unperceivable when it comes in succession with an interval of a few minutes, according to that theory? That contradicts my experience.

If your brain interprets a stutter every 4 minutes as smooth, that's OK. smile.gif
+1. I can easily spot a single dropped frame. A dropped frame is a dropped frame.
post #1156 of 1281
There's one other thing worth mentioning. With most renderers when refresh rate and frame rate do not quite match there is a point when they are "lapping each other" when you get not one dropped frame but a burst of judder, like 2 racers on a track jostling for position. The closer the rates are to being the same the longer this burst goes on for ( but the bigger the gap between). This burst can be VERY noticeable. MadVR, even without Reclock/Videoclock is aware of this risk and does what is needed to limit the drop/repeat to one clean frame. This makes the judder caused much less noticeable and maybe acceptable to some (although not us purists!).
Edited by jong1 - 3/2/13 at 2:20am
post #1157 of 1281
Quote:
Originally Posted by jmpage2 View Post

PCM output of HD audio material from XBMC

Well sure, XBMC isn't the most advanced when it comes to HD audio, even HD bitstreaming is still a new thing for them. tongue.gif
All of the things primarily born in the Linux world seem to suffer from this, VLC does still not do HD bitstreaming iirc, for example.
post #1158 of 1281
Quote:
Originally Posted by Sammy2 View Post

One frame drop every four minutes is not perceivable by me because I don't notice such things due to the way the human brain interprets what it is seeing.

As far as using re-clock goes, I'm really not wanting to add another level of complexity to my set up.

There, fixed that for ya.
post #1159 of 1281
Quote:
Originally Posted by renethx View Post

The only remaining issue is flickering with DXVA2 upscaling (DXVA2 up/downscaling is the best and practically only option for Intel HD Graphics).

HD4000 w/ dual channel RAM @1600 can handle many of madVR's non-DXVA2 scaling options.
post #1160 of 1281
Quote:
Originally Posted by rahzel View Post

+1. I can easily spot a single dropped frame. A dropped frame is a dropped frame.
Yep. Unless I happen to blink when it occurs, the stutter from even a single dropped frame is very obvious to me.
post #1161 of 1281
I agree. Obviously there are scenes where its not obvious because motion is either too fast or not existing at all. But there are many scenes with slow or medium paced motion where such a dropped frame gives a really obvious hicup in the otherwise smooth motion, and i see it. The problem for me is not necessarily that one dropped frame, but the whole concept of "once seen, you cannot unsee it", because after that i'll pay even more attention, and actually look for it, possibly even distracting me from the movie.

Luckily, i never get this anymore, fine-tuned refresh rates to as close as NVIDIA lets me and JRiver MCs VideoClock to ensure it stays perfect.
post #1162 of 1281
Quote:
Originally Posted by renethx View Post

What's the theory (source, please)? Each stutter is surely perceivable. Suddenly it becomes unperceivable when it comes in succession with an interval of a few minutes, according to that theory? That contradicts my experience.

If your brain interprets a stutter every 4 minutes as smooth, that's OK. smile.gif

My theory is that it isn't perceived by some because their brain fills in the gaps.. I have no source but the thing is is that I don't perceive the skips even if the statistics screen says they are there. A lot of study has been done about pseudo-acoustics but not of the same type of effects in what is seen; I am pretty sure it is similar, though. I'm not trying to argue. I am more than happy with using madVR with my el cheopo GT 430 that you even recommended last summer as the PQ is better than the HD 2000 in my i3-550 but it is in areas other than frame skips. It is in color space and depth that I see the differences.
post #1163 of 1281
Quote:
Originally Posted by Nevcairiel View Post

I agree. Obviously there are scenes where its not obvious because motion is either too fast or not existing at all. But there are many scenes with slow or medium paced motion where such a dropped frame gives a really obvious hicup in the otherwise smooth motion, and i see it. The problem for me is not necessarily that one dropped frame, but the whole concept of "once seen, you cannot unsee it", because after that i'll pay even more attention, and actually look for it, possibly even distracting me from the movie.

Luckily, i never get this anymore, fine-tuned refresh rates to as close as NVIDIA lets me and JRiver MCs VideoClock to ensure it stays perfect.

I'm not using VideoClock or ReClock but with the fine-tuning of nVidia it is really not happening for me either so maybe that is why I don't see it. It has been a while since I've used HD2000 and my memory might be slipping too.
post #1164 of 1281
Quote:
Originally Posted by Nevcairiel View Post

I agree. Obviously there are scenes where its not obvious because motion is either too fast or not existing at all. But there are many scenes with slow or medium paced motion where such a dropped frame gives a really obvious hicup in the otherwise smooth motion, and i see it. The problem for me is not necessarily that one dropped frame, but the whole concept of "once seen, you cannot unsee it", because after that i'll pay even more attention, and actually look for it, possibly even distracting me from the movie.

Luckily, i never get this anymore, fine-tuned refresh rates to as close as NVIDIA lets me and JRiver MCs VideoClock to ensure it stays perfect.

I agree. During fast action or during a static shot, like dialogue, a skipped or dropped frame can go unnoticed. During a slow pan, such as seen at the beginning of many films or during "slower" action sequences it can be VERY noticeable.... and once you start seeing that annoying tear in the picture you are just tuning your brain to notice it even more.

Other people simply don't notice it, or, if you point it out to them they are surprised that such a "minor" thing bothers people.
post #1165 of 1281
Quote:
Originally Posted by Nevcairiel View Post

The problem for me is not necessarily that one dropped frame, but the whole concept of "once seen, you cannot unsee it", because after that i'll pay even more attention, and actually look for it, possibly even distracting me from the movie.
Yep. And you cannot just "sit back and enjoy" until you know (or just believe) its fixed.
Quote:
Originally Posted by Nevcairiel View Post

fine-tuned refresh rates to as close as NVIDIA lets me
Close to what? There are 24 and 23 movies and AFAIK NVIDIA doesn't let you 2 custom resolutions for 23 and 24.

You guys said who cares bitsream or decoded sound, its all same thing at the DSP input. Lets say I have sound glitches with LPCM or TrueHD tracks (everything else is fine, only these two formats are problematic). What can I do? I increase the buffer in ReClock. Is the problem solved? Yes, for glitches that I can note. There are still may occur a little glitches masked by loud sound, they are spoil a sound and I just can't point them for sure (sorry for my poor english). Same thing with dropped/repeated video frames. Oh, my point... My point is that with bitstream I never heard any glitches, so I guess there was no one.
Edited by Qaq - 3/2/13 at 11:11pm
post #1166 of 1281
My library has maybe 2 movies with actual 24.000fps, and i just let ReClock/VideoClock take care of these.

I've never experienced issues with TrueHD or LPCM, however both of these formats are a bit special because they have very short audio frames, so increasing the buffer size can actually solve this issue.
If you don't hear it during normal playback, then glitches won't come back when the audio is louder, this type of glitch is independent of what actually is in the audio.
post #1167 of 1281
Quote:
Originally Posted by Nevcairiel View Post

TrueHD or LPCM, however both of these formats are a bit special because they have very short audio frames, so increasing the buffer size can actually solve this issue
Thanks, good to know.
BTW, in my experience sound glitches may occur very random. I only remember it was somehow related to sound level, thats why I said "loud sound".
post #1168 of 1281
Built my own:

Streacom FC8
PicoPsu 150W
2*4Gb RAM 1600Mhz
Intel DH77-moderkort
Intel i3 3225
Arctic Silver
Samsung 830 128Gb SSD

Windows 7 64-bit ENG

Extremely happy with a 100% passive HTPC. 41 degrees C idle.
23,9728 FPS running "Birds" and MPC HC.
post #1169 of 1281
Just got my new Sony HX850, and for the first time able to attempt 24p playback of my Blurays on HTPC.

AMD Phenom 2 X6
Nvidia GTX460
And all the other parts that aren't important.

Tried using mpc-be with its changing refresh rate to 23hz, went d3d full screen and sync options for renderer, and while the video looked perfect, I could see the audio slowly getting further and further out of sync. Lasts a minute until its out of sync too much to notice. Been reading these posts and many other forums similar about this, but it didn't really click until I could try it myself.

Anyway, I've tried MadVR and Haali, checking all the timings with Ctrl j, and it said 23.970hz. Not good enough.
So I've done some custom timings in nvidia control panel, and I'm on 23.97533hz. MadVR says that it will drop a frame ranging from every 2 hours to 2 days. Playback is smooth, and audio is not going bad after 2 hours into a movie.

This is probably as good as I'm going to get. And I'm happy. Unless someone can tell me how to get closer...
post #1170 of 1281
There is one caveat when decoding HD audio to PCM - there are some titles that make use of DTS-HD's dialog normalization flag. My understanding is that this flag can change during the movie and this can cause excessively quiet audio output (and it may not be lossless anymore). This may be a bug in the Arcsoft audio decoder. I can't remember if Nevcariel mentioned this in a changelog or if it was Madshi (eac3to).

Fortunately the use of the flag seems to be very rare. If it bothers anyone they can always bitstream that title, I guess?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Home Theater Computers
AVS › AVS Forum › Video Components › Home Theater Computers › Let's set this straight - No one can do 24p consistently well