@PerryWilson

I-frames are intra coded frames, like a jpeg. In simply terms the smaller sizes are more compressed and hold less detail. Therefore, the larger the better, but of course at the expense of required bandwith, device read/write speed, and storage size.

@proaudio4 @sohus Yes, my 88M AQ2 passes the dense deathchart. I'm convinced that the patch is okay, did a good series of shots with it today. I also shot with it at AQ3, and like bkmcwds first release of 66M GOP3 it does break the bitrate limit, but only slightly, under certain conditions. Out of the shots I took with it at AQ3 only one gave me a noticeable problem within a stream. See attached stream at frame 60, visually there is a noticeable jerk at that point. So.. my patch is fine as I posted it here at AQ2, just be aware that there could be problems with detailed shots at higher AQ.

@bkmcwd Was thinking more about the 66M GOP3 and your requirements, being that you want a lower bitrate GOP3 for lowlight (or low detail). Well it does do that as you first released it, and you could make it even more sure that it'll behave and will not break the bitrate by using a soft lens too. However, I'd still use the driftwood 132M GOP3, the bitrate will be higher, but under the situations you want to use it for it won't be very high.

@cosimo_bullo

"though our freshly formatted Transcend 10 cards are having trouble keeping up with the data rate"
Literally? Do you get errors or visual quirks in the stream?

Is this the v3 (with the Q values) or the v2?

buy the Sandisk Extreme cards... im having 0 issues

No problem here with 176 GOP1 Sandisk Extreme..

I've just spent some time with it and I've had no problems with the new sandisk extreme. http://personal-view.com/talks/discussion/comment/18767#Comment_18767

I like it very much, but believe that due to the other settings its actually running at somewhere between 145-152M and not 176M. Has anyone else had higher rates from it ? Don't get me wrong, thats bloody good news as far as I'm concerned, it helps with stability. It does mean though that there is slightly more headroom to tweak up the image quality.

I may have to buy an extreme. How long we talking for recording time say at the Driftwood 176 gop1?

Nigel785,
Since Stay's observation above about 176 running on extreme high detail around 150M, here's what you can expect per minute:

150Mb/s = 18.75MB/s
18.75MB/s x 60 seconds = ~1.125 GB per minute.

Since this number is based on extreme continuous detail, I imagine you will do better than 1.1 GB per minute, but to play it safe. I'd say a 32GB SanDisk Extreme 30MB/s SDHC can offers 30 minutes safely.

I'm getting about 24 minutes on a 32GB card although that's the camera's estimation, not the total running time after filling up 32 gigs.

I'm really torn because I'm in love with how these images look, but it chews through so much disk space. I'm traveling Europe for the next month and I'm debating whether I should go all out with the 176M patch or dial it back to something more manageable like 44M. I'm bringing a blank 500GB external with me which would hold about 6 hours of footage @ 176M.

@nomad, of course :) And it may come to that, but I'm trying to travel light and not completely break myself financially!

I have a free 2TB drive as well I could bring but it's not 'ruggedized' for travel - anyone have experience bringing a standard external on long trips versus 'optimized' drves? Maybe it's just hype.

@proaudio4 I think another 100K on a frame is about the maximum you can get in this bitrate of 176M, but its likely to push it over the edge for a lot of SD cards ability to write it . As it is now, if you look at the test I did in the banding thread it is on a par image quality wise with cbrandins 66M AQ2 (though it is a bit lower than the 66M I think). Sooo.. getting that 100K will definitely put it equal to, to slightly better than the 66M. My comparison however doesn't take into account GOP1 in terms of motion rendering improvements, the absence of smearing, and the far cleaner noise that it has over a longer GOP which does improve image quality a lot on most shots (even if it is actually compressing at a higher ratio). Also, after all, the image quality at that 66M level is ample and definitely good enough for a lot of heavy post work.

Edit : I will have another look at this tomorrow, really want to see how it holds up shadow detail in lowlight, which is the only concern I have about it. I do now after all have enough SD cards to use a GOP1 and it is definitely growing on me.

@proaudio4 Instinctively I'd say yes you would get better results doing NR in post on GOP1. It may just mean that you'll get less medium and low frequency noise (which sometimes results from smearing), which is obviously a good thing. But at the end of the day its really down to the method your NR tool of choice is using (and what its geared to expect or deal with). The only thing to do is try it and see. As you say the compression/bitrate is going to be very key with this issue, and unfortunately at high ISOs the noise is probably going to eat more of the bitrate with this patch than your shots detail is. But as much fun as I've had messing with in camera NR I still won't use it if I genuinely want to do proper noise removal. I'd drop the ISO and grab more lights before doing that if I was worried about losing image detail over the noise.

If I have time in the next couple of days I'll do a proper test with nukes new wavelet based denoise algorithm which can generate very informative profile curves of the noise across the image for a good comparison (as well as outputting noise only images).

Edit : Simply: Hopefully, and logically, the GOP1 will have fine detail noise comparable to the 66M AQ2, and the noise detail should be maintained better than the 66M. This may mean it is easier to clean in post (as in you'll get better results), but it might make no difference.