@cbrandin

"It seems to me that if the image looks the same even when the bitrate drops in some frames, then the bitrates while they are high aren't contributing to image quality. Why not just drop the overall bitrate so they are all as low as the bitrate during those frames when it is lower?"
not necessarily. In my case the uneven cadence goes hand in hand with larger I-Frames, see here:
http://www.personal-view.com/talks/uploads/FileUpload/ec/adfb424748f2cdcffbf5611780d098.jpg
So the actual bitrate does not seem to go down.. just the distribution of I-, P- and B-frames is chaning.
So maybe somehow things get leveld out ...

@cbrandin
"Does it just change cadence by itself, or did something in the shot change at that time?"
Exactly my thoughts when I saw the stream.
I've looked at the footage and at the timcode in question absolutley nothing happens. It's more or less a still scene (also still a few seconds before and after the cadence change). I've just filmed passing traffic and at the time of the cadence change no car or pedestrian is passing. It's actually only an architectur-still :-)

Or maybe:

@LukeV ... or anyone


Video Bitrate 24H ADD=66000000
Video buffer=0x3000000
Video buffer 24p=0x3600000

what do these setting do?

http://www.personal-view.com/talks/uploads/FileUpload/3f/c77656fff5ea67dbe5e9ff1c9bd3c5.zip

@LukeV

Why are you setting these?

Video Bitrate 24H ADD=66000000
Video buffer=0x3000000
Video buffer 24p=0x3600000
1080p24 High Top Setting=42418
1080p24 High Bottom Setting=29692
1080p24 Low Top Setting=31178
1080p24 Low Bottom Setting=21824
Other Modes Top Setting=31878
Other Modes Bottom Setting=22304

These are automatically set when you set bitrate. Have you found any particular advantage in setting those parameters?

Chris

Hmmm. Raising FB1 and FB2 and Frame limit also raise maximum I frame size (you have to set all three) - how about adding those? Maybe the I frame size will get even bigger - which would be good.

Chris

The camera crashes if you haven't set all three in the same proportion - frame limit maybe very slightly less. Given that, I haven't had problems with it.

Chris

@cbrandin
I've set them to the same proportions.
With your settings I always get an uneven stream cadence. Only with really high values the cadence gets even (88mbps, FBS + EFL x4). But then the values are so high that the price to pay is a possible codec crash, playback and transcoding issues.
My latest high settings (based on raised top/bottom BR settings) record and playback fine on both class 6 and class 10 cards. I Frames are not quite as large, but IQ looks really good so far.

okay :-)
http://www.personal-view.com/talks/uploads/FileUpload/b9/e5fc5d240ef05cdc03d037da270a6e.zip
24H x4 = 88Mbits ... Top/Bottom Bitrate settings mutliplied by 4.
24L x3 = 48Mbits ... Top/Bottom Bitrate settings mutliplied by 3.
FSH x2 = 32Mbits ... Top/Bottom Bitrate settings mutliplied by 2.
GOP12. AQ 4.

Yes, the MJPEG is from lpowell

@cbrandin Just to confirm what @towi said as I'm seeing the same things, the uneven cadence does occur on just end user patches too. Also yep, it occurs even if nothing changes in the shot, most of my later tests when I was tweaking to try and remove it were completely static shots indoors. I have not seen the cadence problem with AQ1 yet, I did see it briefly at AQ2 during a shot with a lot of motion, however the stream did go back to normal cadence. In all the tests I've done I've had frame buffers doubled. Except when using settings from @towi (which I just lowered the 24H to 66Mb on) where I also had the top/bottoms multiplied (as per bitrate mulitiple) but as per Vitaliys advice I have stopped doing that and went back to FB x2.

"Apparently the uneven cadences do not destroy IQ. On the other hand it's certainly not what the stream should look like … so in fact it makes me a bit nervous."

Ha, exactly it makes me nervous too, as does the effect of it dropping to a lower but even cadence instead if you replace T4 with T1 after about 20 seconds. I get what you're saying that it may be a buffer fragmentation issue @cbrandin, especially as it only seems to happen during longer shots. Are you saying that at a higher bitrate (such as 88MB) the buffer is having to flush more regularly as it doesn't have the space to store references for as long ? If thats the case, would changing QP also force this flushing at lower bitrates ? Probably not I'm guessing. The buffer fragmentation makes sense to me in that it also sort of explains the drop off after 20 seconds if you change T4 to T1, as it seems to be basically starting over somehow (my tests there were all static shots in regard the T4 changes, @kron IIRC found the same behaviour too).

Edit : Alternatively, is it worth testing with a smaller video buffer, but leaving frame buffer x 2 ? Weather is crappy here, so I will probably have some testing time today.

I wanted to ask if the low GOP assertion that it was the cause of cadence problem still standing. Because what we are seeing now is that the uneven cadence problem affect all GOP. Am I asking because this belief has ended the short GOP testing because it was said that we had to wait for a bug fix in the next ptool version.

The cadence problem _may_ happen with GOP12 occasionally under certain (unpredictable) circumstances.
But it seems to happen all the time with lower GOP (6 or 3).