@cosimo bullo
>Wir sehen eine groß genug Unterschied zwischen 44 und 88 in 3,62 der höheren Bitrate (Ich weiß, das ist etwas subjektiv, und das ist gut so) zu rechtfertigen<<br />
??? yes very subjective: I think that's wishful thinking or a kind of self-fulfilling prophecy

Hi there!
That's my first post but I've read and tested a lot already, actually scince the GH13 developments. So, let me contribute my settings. Rather extreme at first glance. Lots of trial and error testing before I decided to post something here. Strange enough for those settings - it's stable, plays back in camera and produces the best image of everything I tried out so far (motion, shadow details, unwanted noise/macro blocking...) I use sandisc class 10 cards, the in-camera settings are noise reduction +2 and sharpness -2 !
I found that buffer sizes/limits along with doubling the original gop (to 24) does unexpected (positive) things to the otherwise extreme settings that crashed before. Also setting noise reduction and sharpnes in camera to those values helps not only visually - also stability wise. So if you like, give those settings a try and look at all the weired patches I've checked.
1 thing I should add - it's aimed at 24p recording only, so it affects 24H in AVCHD mode.

film_noir, does this allow files to span correctly?

Also, have you tried shooting the death chart?

@film_noir >"So far I'm not planning on a feature film scene, incorporating death charts."

LOL!
I hear ya bro. But, it certainly adds that sense of knowing you're probably safe from errors.
If you're interested, the "death chart" is in one of these posts. Maybe someone can point to it.

@Butt - no idea what that means.

Just keep it clean, fella, we're all here to help make a good thing better.

@Angry_C
What are you talking about?

If you need to blow it up 400% to see image improvement, you don't need the hack, you need glasses.

I see the improvement in overall detail and low light detail (shadows).
When using higher iso, the higher detail ability now ables the noise to successfully be displayed which now can make software programs such as Neat perform a better job of removing the noise. I'm no longer seeing noise being smeared by macro blocks dancing in the shadows. This is a BIG improvement for me for overall clean image quality and for color grading in post.

It's too quiet here - compared to the "extreme settings" topic. Scince we are talking about quality, here is a still - of course taken from a MTS sequence. That's my personal "bush of (codec) death" and so far - no problems. EX tele is fine too, using lots of c-mount lenses with great results. The scene was shot in the evening with ISO 160, with the in camera settings and the hack-settings I posted above.

The max bitrate is set to 96 but streamparser tells me an average of 52 for this sequence. As I said - my settings are rather try and error than theroretical mathematical studies...
Sorry, "only" 50 average. So far I'm getting no takes lower than 48 and nothing higher than 62 depending on details in the scene. Ex-tele shots tend to have a higher bitrate

"The goal fo this thread is to get the highest quality image output with no crashes, write speed issues, etc. Could someone just post a .ini file that does this?"

The settings I've posted above work for me. Limited, though.

"sete" contains settings for Encoder Frame Limit, Frame Buffer size and top/bottom Bitrate seetings.
"setf" contains no EFL and FBS settings, but top/bottom Bitrate settings.
Both 88Mbps / 48Mbps / 32Mbps. GOP=12. AQ4 + AQ3.

24H shows an even stream cadence with both settings at all my tests so far. My camera does not crash. No card speed errors. "sete" produces somewhat larger I-Frames (naturally).
However in-camera playback is an issue with "sete". "setf" seems to do fine.
"sete" requires class 10 cards (I use SanDisk), "setf" also works fine with class 6 cards.

24L set at 48Mbps (default x3) works okay in both settings.
It may occasionally show an uneven cadence with "setf".

24L reduced to 32Mbps (default x2) actually ALWAYS shows an uneven cadence in both "sete" and "setf".
FSH (which is 50i for me ...) shows no issues at all.

I've also tried 44Mbps for 24H ... but it always shows an uneven stream cadence.

I am not up to reach a super high bitrate -actually I don't care ... but interesstingly my 24H/88Mbps setting is the only one that did NOT exhibit the uneven cadence issue. I can't promise the setting is stable under all circumstances, though.

Here are the settings.
In short: "sete" for larger I-Frames (and class 10 cards) / "setf" for stable in-camera playback (and class 6 cards).

I've enlarged comparable shots of both settings to 400% and my feeling is "sete" looks somewhat better. But probably negligible ...
I've seen one shot (but only one) that showed a kind of pulsing in shadow details with "setf". Don't know why ...
Other than that the camera "feels" more stable with "setf" (as far as in-camera playback goes, again, it IS more stable).

edit: oh, of course also the files don't span.

Happy testing ;-)

@cbrandin
sometimes right after ~10'' ... sometimes somewhere in the stream (although nothing dramatically changes in the actual scene)...

would you post your settings?

edit: first post on page 3 of this thread I've posted a streamparser screenshot of a scene shot with your settings (actually not your settings-file, but a setting based on the values you've posted somewhere...):
http://www.personal-view.com/talks/uploads/FileUpload/cc/19dbc3e80681b40d2d0056e120831c.jpg

what is 10 seconds?

@towi

I mean when the uneven cadence starts to show up in your streams.

Chris

@cbrandin

"I mean when the uneven cadence starts to show up in your streams."
I don't have the files anymore... I don't know at which timecode the cadence started to become uneven.
I've seen the uneven cadence with your settings starting at 10'' in some shots, though.