Stray, Based on my graph tests, the 88M yeilds larger i-frames for a given repeatable scene. It certainly seems that this would support that the 88M will yield better shadow detail. Again, the difference is not great, but measurable.
Also, I can set up a repeatable low light shot under ISO160 and we can compare i-frames and image quality between 88M and 66M for both motion and static.
As Chris mentioned, his 66M max variation is "experimental at this point, so don't rely on this patch yet." So, it's possible Chris may refine his settings.
Also, we still need to come up with a "standard" 88M AQ2 setting (as we spoke about) which we could compare to your max variation 88M AQ2.
Lowlight tests would be great if you have the time to do it.
Yep agreed on the need for a standard 88M AQ2, plus I think it will be easier to make a stable standard 88M AQ2 setting now we have (I think) a better understanding of how FBs, Hi/low and Framelimit settings effect stability.
Stray, the low light tests are done. There in the "AVCHD maximum image quality settings and testing" thread. BTW, thanks for the excellent 88M AQ2 MAX VAR setting. I'm going to try and set it to AQ3 and see what happens.
@proaudio4 I am very thankful to you to having done the polite test.
Although I was imagining in fact that it was a thing with this still insufficient stress to codec in the test environment of my former, your test just proved it.
If I say honestly, I think it quite difficult to stabilize AQ3 in 3GOP. Therefore, first of all, I think that I will stabilize this AQ2 as your advice.
I'm having hard times with cbrandins 66m hack fro august 29. When in 720p, I get horrible failure rates. I'm using transcend cards but with a san disk they are working well. Looks like ill be getting rid of the transcends. All class ten. In 24p, I don't seem to have issues. Everytime I choose etc mode. Goes into immediate failure with the transcend and pretty quickly with the sandisk. Any help on understanding these low gop hacks with a link would be wonderful.
Is there any use I can be to testing (Mac based so no access to streamparser)?
Just got back from two weeks shooting with 2 gh2s as B cams (using Kae's 3.61 GOP3 settings). Would like to contribute instead of just freeloading on other people's hard work.
(Incidentally on the shoot, the gh2 was up against an F3 shooting to nanoflash at 140 mbps I frame, then later an Alexa shooting to Prores 422 HQ. Was incredible how well it stood up -- in many cases it kicked the F3's ass, and there was even a few times it may have bested the alexa at least in terms of "filmic" look and cadence (will post some impression on that in a separate topic elsewhere)).
@proaudio4 Although I improved instantly after that and being said till the good place, since Stray's new STRONG test chart appeared, I gave up 88M3GOPAQ2 despite today... I instead made the strange setting in which both 24H and 24L are 66M,3GOP and AQ3. Although I was testing for making 88M&66M,3GOP,AQ2 from the first, since 24H of this was never stabilized, both were reluctantly set as 66M, and AQ2 was simultaneously set to AQ3.
I am going to change either 24H or 24L based on this setting someday. It seems that however, this is also very stable.
Supposing you are not disagreeable, don't you have this setting tried, either?
Note: The attached screen shots are tested with swaying Stray's new STRONG test chart. This was stronger than Pappas, I think.
Hi guys! Well I was a GH-1 Hacker. And have been away for a bit. Now I finally have a GH-2, seriously, it hasn't even been turned on yet!!!! So, what are the top three current "best" of the low GOP settings? I'm going to start testing for a Tango dancing film. I'll check elsewhere but I'm also going to be shooting some 2x anamorphic with an Sankor, I'll have in a couple days. I know there is a big threat on that. I want to dive in!!! So, where would you recommend that I start? Thank you!
So what is the testing methodology? Do you keep different settings on separate SD cards? Would I have say a 44mb setting card, and a 88mb card?
The 220m GOP1 looks INSANE, but there are some tricks to converting it. That being said, you really should at least try it out and play some back in VLC just for the mind-blowing experience. The 3GOP stuff is plenty stable for low GOP stuff.
Its all over! This is it. This is PURE INTRA heaven (no 2nd P frame). Result: smooth playback in WMP, PP, FCP etc.... drop it on a 24P timeline and see. Slightly tamed (for pappas death chart) till I give it back some bitrate :-) but it is SMOOTH as F@ck... Try it and weep - this kills GOP3. Forget any other settings apart from 1080p24H, the rest are all work in progress, but you have to see this in action. Work out what I did ;-) This is a huge development from my previous Insane GOP1. Time to change my GOP3 and any other 24p setting - rid ourselves of that bloody P frame!!! If you have yourself a revolving chair, sit yourself in it and spin, film yourself, play it back, and you'll know what this means! NO MORE EDITING WOES.
Note: You dont need AQ anything for Intra!
Driftwood PURE Intra - SmoothAsFuck GOP1 Hi Bitrate - NO P Frame - setc.zip
579B
INTRA i frames only - 244M 1 GOP 1080p24H Heaven.png
@driftwood this looks great, can't wait to test it. What about stability? I had some problems with your 220M GOP1 settings. Thanks for your great work!
I was searching youtube. Is there a video that shows step by step how to apply the hacks and settings, and using the software to analyse the videos? Thanks
I was shooting today with the New GOP1 INTRA hack from @driftwood. Didn't crash once and spanning worked fine. A 25min shot became 3 seperate files of 4.29 GB I'm downloading now and will try to post some shots today on Vimeo. Will post my findings.