@driftwood "You can get even bigger i frame sizes than 1M with chris's patches"
Of course, I know. Since the frame size of the limit which can be played back in a camera is about 1.09M(?), I have set the maximum frame size as 1M for it to be less than it. And I know why you are setting frame size of Intra to 800k. It is very rational and I regard you as true. Although I cannot explain well in English ...
It has been raised earlier in this thread, and given the discussion unfolding regarding the current PB shootout, including a GH2 with your magic PTool v3.63d applied, and @driftwood Quantum patch, a valid point raised by @TheNewDeal pertains to adjusting the GH2's gamma curve, to give it a further edge.
GH2 is not 444 or 422, so have be careful not to have a curve adjusted too flat or broad for 420 and post work. Any prospect/progress exploring custom curves or adjusting the existing curves in the GH2?
@driftwood && i waited this kind of response :-) I asked you that because some settings are configured into the patch setting for tester.. So if you tell me to try, i deduct that there is no risk. Ok so, on ze road again!
>>>Trust me the H setting is better, just do a streamparser test and you'll see ;-). After a few seconds the i frame size for the L setting will drop to around 2/3rds of the max of the H setting. But 'L' is great for longer recordings and perfect for most shots especially during daylight. ie It will cover most situations and is far superior to stock. If you're after highly detailed scenes and best quality, stick to H but of course, it costs you in storage size.<<<<br /> @driftwood
I would highly appreciate if you could expand on this a little further - not an elaborate dissertation but it would be nice to know in what kind of general scenarios the H mode would offer obvious advantages and, likewise, when the L mode would offer no noticeable disadvantages. I ask because this is new to me and I don't have enough understanding of the camera (or subject, pardon the pun) to draw conclusions.
@driftwood I'm trying to find the time to get some testing in myself. But in just looking at your Quantum 5 settings, I noticed you have the quantizer set to 4 (which given the 0-51 range is still pretty low). But finding the right quantizer value vs. bitrate can be a tricky thing to balance (and on top of that, it is subject material dependent too). Given the (what I believe to be very high) bitrate of 160 Mbps, I would have immediately started testing with a Quantizer value of 0 against complex and simple subject matter and then increase from there. The higher the bitrate and or ease of subject matter being compressed usually translates to a lower quantizer value yielding better results. I am trying to catch up with what has already been tried and tested and couldn't find any details in regards to quantizer testing for this patch. If the quantizer values have already been tested, I didn't want to re-test what has already been done. Thanks again for all your hard work!
@bkmcwd I just graded your latest test on vimeo. STUNNING quality!! Really looks amazing when graded. How stable is your patch, and what kind of record time are you getting with it?
@Zaven13 GOP size certainly has it's role in all of this, but I am speaking specifically about the Quantizer. @driftwood I just noticed the terminology used, "Auto Quantizer". This choice of words leaves me wondering if the Quantizer values 1-4 are not really QPs (Quantizer Parameters) but rather maybe a choice of different rate quantization models in this encoder's rate controller or maybe a predetermined range of QP to use?? Or maybe it's simply the QP initializer value? But then why limited to values 1-4? It's hard to say with the limited information available. But, nevertheless, given the copious amount of bits we are able to throw at these pixels, I would definitely lean toward lower the quantization values (which is what I think the "Auto Quantizer" value of 0 is trying to communicate)
My sincere apologies if I offended anyone, I realize I am coming in rather late. I am sure at this point the Quant values have been tested quite thoroughly and "4" is the best setting. Thank you again for all the long hours any and all who have been working so long on these patches.
@bkmcwd still trying to get a compressed version that doesn't smear. in ProRes444 it's spectacular. But going to MP4 the compression is smearing the high detail areas.
Without the fact that it doesn't span without a 95Mbps Card is Quantum V.5 the best patch so far for getting the best quality at 24H?
I read that it's close to Seaquake for 1080p 24H but which one is really better ?
Also is there another version of quantum that's more stable than V5? But will it produce the same quality?
I'm reading everything on here but i'm still confused...I only shoot in 1080p 24H mode so i'm just wondering if i should continue using this Great patch and i didn't had any problem with it so far or should i use one of the previous ones for more stability.
@FGCU Looks great! Skin tones are great and the feel of the images are soo nice! I love how the camera renders light.
I just did a bunch of interviews (all shot with a canon c-mount tv lens 50mm f1.4), which I unfortunately cannot show right now as I did it for someone else and there might be rights issues - I used very simple lighting and it turned out great, and graded extremely well also.. Maybe I can link something later when it's edited.