AVCHD maximum image quality settings and testing

cbrandin

I'm using Sony Vegas. I also looked with Elecard StreamEye.

Can you send me a screen shot? I'll see if I can find the same frame and dump it from Vegas.

Chris

cbrandin

I did notice that the player in StreamParser (which is just Microsoft's player) looses track of the correct frame number when stepping through - maybe some players' codecs just don't like these high framerates. Everything seemed OK in the editor, though - so maybe it's not really a problem - except with playing unedited clips.

Chris

Ralph_B

I played the clip in Vegas and it's OK. However there is definitely a problem playing it in Media Player Classic. (see screenshot) Not sure if high data rate is the issue - MPC plays 132M clips without any hiccups.

It just one of those things.

driftwood

OSX Movist will play back GOP1 and 3 ok. VLC if I remember should be ok.

Note: Interesting that VLC reports a 23.97 short GOP frame rate as double (2x23.97) = 47.952048

http://code.google.com/p/movist/

cbrandin

@Ralph_B

That's extreme alright! It's definitely not the clip - it's the player.

Chris

peterosinski

no matter what I do I keep getting glitches in my footage like the ones here

that is using cbrandin 44m patch with a 8GB sandisk extreme 30MB/s card in ETC mode.

what could this be?

Ralph_B

@cbrandin

When you say the player loses track of the frame number - is that something that's fixable at our end?

cbrandin

No, it's a limitation of the media player plug-in

cbrandin

@peteroskini

Something else is wrong, then. I never get anything like that. I just tested ETC mode again, and everything is fine. These increased bitrates aren't, shall we say "strictly legal" for AVCHD level 4. Some codecs might just not be able to handle them.

Chris

stoney

Among all settings I tested with AQ3 or AQ4, all showed weird behavior except @cbrandin's 44Mbps AQ4. Often I-frame size became very small when panning slowly. All I, B, P frames became flat. Usually happens within 1st minute. I concluded that AQ3 or AQ4 is not stable for higher than 44Mbps with current PTool.

cbrandin

@stonebat

That makes sense. Realize that AQ is not an absolute - it's relative to bitrate. 66M at AQ2 shows the same individual frame IQ improvement as 44M at AQ4 - with 66M AQ2 there is just more bandwidth that can be applied to motion. Also, un-tuned 66M settings will exhibit a lot of odd behavior.

Chris

B3Guy

Ok, so to sum up, this is completely an issue of video players mishandling straight-from-camera files in playback? So the strobing is not present in the actual file, and once converted to ProRes or some such, the strobing will be gone, a non-issue?

cbrandin

Well... I don't know if it is always the issue - but it was in this case, it appears. I assume you meant the @Ralph_B clip? The issues with some high AQ 66M settings is a different issue

Chris

chauncy

I think so. I just saw it inside the source window in adobe premeire pro cs 5(full quality).
This was with several high bitrate samples, 56, 66,132. not 44mb.

Upon transcoding it to mpg at the same high bitrate, there was no sign of this jumpiness.
And no quality loss I could see.

I think we're beyond what a lot of media players are prepared to render in avchd.
So my guess is that's the issue.

peterosinski

@cbrandin

what could be causing those glitches...i just set my camera to kae's 65mbit settings from ptool 3.61 and i still am getting glitches

what is the best way for me to revert back to the original firmware of the camera to try and fix this?

cbrandin

Just load the original firmware into PTool, don't change anything, save it, and load it into the camera. By the way, there is a newer version of PTool.

Chris

peterosinski

yeah i know,

ive been using it but went back to 3.61 to see if it would make a difference

B3Guy

Ok, here is one clip (straight from cam) from me messing around last night that I was seeing this "double image" stuff in. Check out from 0:30 seconds in and after, focus on the strong vertical of the wall, IDK, but I personally see double-image blinky weird stuff going on.

B3Guy

Download the .mts if you like, but I was seeing it some even watching on Vimeo.

cbrandin

Guys, this stuff is completely off topic. Besides these decoder issues really deserve their own topic. Can we move this?

Chris

B3Guy

Apologies. Move, please. There are also posts on this over in the Low GOP topic. A new compiled thread would be ideal. Thank you.

cosimo_bullo

So, to summarize, Chris, do we need new settings for 66AQ3? Or is this just a player thing and we can just change the AQ setting from your last 66/44 batch?

Also, any chance you can create a tuned 88M file? Pretty please!!!!!!!!

As always, Ralph, THANK YOU for continuing to really be the ringleader here in the Max Image and Quality Settings thread. My lord you most have loaded a lot of firmware the last few days!!!

cbrandin

You can just change the 66M settings AQ parameter - but there may be resulting stability/in camera playback issues. The player issue is completely separate.

I have not been able to come up with an 88M setting that is better than 66M - in fact they tend to come out worse and less stable.

Chris

shrigg

Here's an ungraded clip of some live music I shot with cbrandin's 66 AQ2 settings using ETC with my Nikon 17-35mm f2.8ED lens. ASA 160, exposure was 1/50 at around f 4.0. Smooth -2 -2 -2

I'm very happy with the motion and detail. Well done Chris!

Ralph_B

@cosimo_bullo

I concur with Chris. I tried an 88M setting and it was slightly worse than 66M. But rejoice, if you want to go beyond 66M, Driftwood's 132M setting is so close to the HDMI picture for both static and motion, that there's very little room for any further improvement. Just stock up on your SD cards!

Howdy, Stranger!

Categories

Tags in Topic

Top Posters