@sohus Yep, I'm onto it, and the ETC/80% problems. I'm pretty sure I can fix both and tighten it all up, but it will take about a week or so with my other commitments at the moment.

@sohus

The notes say: bkmcwd 154&154MB 3GOP Q=12 FBx3 FLx1 TBx8.6 VB3.6
When I downloaded it, I put the date Oct 9, 2011 on it. Hope that helps.

@stray I took your dense chart, imported it in After Effects, and put a 'Wigglerama' filter on it that zooms, scales, rotates in variable speeds and intervals. I was not sure if it would work but so far I have crashed many patches that survived my regular Death Chart test just fine. I loop it and do 2min - 4min tests. So far, it has resulted in memory card speed errors for many patches. I am pretty confident that it does work as I hoped for that reason. I have an iMac 27" high res screen. I test them using the 14-140mm lens at 25mm, F/5.6 and ISO800 (and 1600).

I think this test is as extreme as it can get. But because it is indoors, lit by my monitor and an AE project, it is very easy to test every new patch this way and compare results.

@Driftwood's GOP1 patches survive it just fine, his GOP3 132M fails (also on static Deathchart)
@Cbrandin's GOP12 patches survive it just fine (MAX VAR as well)
@Sohus' new 154M GOP3 patch (based on foundation of H. Olonga's patch) survives it just fine
@bkmcwd has to wait, as my Transcend doesn't like it, so I ordered a 8GB SanDisk Extreme for test purposes, arriving tuesday. I also hope to improve upon my modification of Olonga's patch to get it stable with that card.

Most tests were not created for the Sandisk but on the Sandisk.

Also, I see future cameras leveraging the UHS-1 cards so I don't think it is a very wise investment to buy heavily into the Extreme Pro cards now. Most patches run fine on the Transcend, only @bkmcwd and @henryolonga's new patch have serious problems on the Transcend card, but they are extreme settings.

I am sure @Stray's 88M was never created with SanDisk Extreme in mind. I will retest the two patches mentioned above once I get my card, all other patches run fine on the Transcend so no need to retest.

After 3days travel, I came back to the house. :-)

@Stray
Probably I also regard my 154M set as the ability of neither ETC nor the 80% mode to be tested under high stress.
However, it is although I do not think that I would like to raise QP in order to be able to use these... ;-)

@Ralph_B
Thank you very much for your wonderful test!
I am satisfied with the result of your test.
Although, as for the set of Q10, frame size should become larger, did you test it?

@sohus
Thank you for testing my patch!
What is QP or AQ in your new 154M GOP3 patch which survives your test?
By the test of my past, if it is Q14 or more than Q16, also not AQ3 nor AQ4, it turns out that the form of cadence at the time of super-high stress does not become a flat line easily, but since I am thinking as important the practical use level, especially lowlight, I am using Q12 and Q10 or AQ3.
I do not feel a problem at all on these practical use.
Although it becomes behavior also with AQ3 same flat cadence under super extream stress, since frame size becomes small from Q10 AQ3 at the time of lowlight, I like Q10 version.

@cosimo_bullo
Thank you always! I respond to your request. :-)

Of course, I have also already tried his patch before travel.
I liked the extreme setting as you knew, but his set was too going too far to completely use FOR ME truly.
Many frames have fallen at the time of the practical highest stress which I assume.
Although I am thinking the character under lowlight as important, I think that balance is important.
If used only under limited conditions, I will also think that his patch is very attractive.

@Ralph_B bkmcwd means the first setting here, he changed the Q setting for higher image quality http://personal-view.com/talks/discussion/comment/20179#Comment_20179

@Ralph_B
I am sorry that my English is strange! ;-)
As Stray followed up, my latest patch is Q=10.
Since it was Q=12 that you tested, I asked a question whether you tried Q=10 version.

"Does "Q setting" refer to both of these parameters?"

Yes! However, I am not sure whether "Initial quantizer=10" is required.

"And does he feel 10 is better than 12?"

Yes!!! It is AT LEAST FOR ME.
Q=10 is because frame size becomes large rather than 12 under lowlight.

@Stray
Thank you for explaining instead of me.

@Ralph_B
Thanks! :-)

BTW, I do not learn how to use ISO6400 with GH2.
Would you teach usage?

@bkmcwd
Here are the results of comparing your 154M Q=10 and Q=12 to lossless HDMI.

[url]http://www.sendspace.com/file/jz657q[/url]
[url]http://www.sendspace.com/file/py5hzb[/url]

IMHO, Q=12 is better. The finest details are better preserved. Examine the noise.

Always check your resuts of Q and IQ on Streameye - check your QP results and scan the macroblocks.

@driftwood
Thank you always!
I try on later. :-)

Major failure on a project with two 64Gb cards Transcend class 10.. with Driftwood 176 mbs hack. A few hickups still, when switched to 132 mbs gop 3 hack.. Not stable in my experience. Had to shoot interlaced.. and go to 44mbs hack..

So far 44 seems to be a best reliable version. there only Transcend cards available in Russia in 64 GB size..
Have to figure out best settings. (kae settings, never had any problems)