@yskunto Hi fellow! I´ll try to make my tests with a faster SDHD one hehehebut anyways , this patch from humpman was holding up ALOT more than any other patch that haven,t seen before!! even with a "theorically" so slow SDHC card!! :) Next time, I´ll do some test on a tripod to see what happens.

@humpman Hello! One question only, Must we to overwrite the existing tables in the original patch? I mean [Patch6] and [Patch7] whith the new ones? Both tables at the same time? Thanks!! Asier.

@Asiertxu Can you retest it using faster SDHC card class 10 like Sandisk Extreme? Is the problem is still occur? You zoom in in the video, how if you did not zoom? Does it still occur? I am curious. At least we could know, is it triggered by certain action.

Here it is!

https://drive.google.com/file/d/0Bz3JTW-9ErvlMGxZMzloMW9kOG8/view?usp=sharing

Cheers! Asier.

This is a good excuse to break out my GH1! Thanks @humpman

Thanks guys, the table from june 4 settings gives more quality on low ISO's in low light scenes and higher bitrate. I think you'll not see differences visually on daylight shots, but i didn't tested it specially.

@frullaccia yes, i just wanted find good balane, but i can't seat two extremes on one chair. If i get nice 3200 iso, then i get bad low ISO's and so vice versa. So i will stay with previous table style. @yskunto there is 91000000 max bitrate number for PAL, and 98000000 for NTSC(24.000), so i would like to separate them.

@humpman. Following our previous conversation.

You are right about the fact that the first numbers of the scaling table determine how detailed the picture is.

Furthermore they are concerned with other parameters and are important for the efficiency of the codec.

Using low numbers will produce a very sharp image. Sharp images are difficult to be managed, because every little problem of the camera appears to be magnified through the detailed scene of the shot.

It is like having a good hifi amplifier. If the whole musical system is not well balanced, every single problem will be clearly audible.

The real problem with low numbers in GOP 1 patches is flickering. If we use GOP 2 or GOP 3, with their P and B frames, then the P frames overall will be critical, because the codec hardly support that level of detail. P and B frames in our GH codec are not conceived for that.

@yskunto thanks for testing, actually 8 june thettings is only for NTSC. By now i'm in work of designing of new tables, just trying to improve quality on high ISO's. Also i find some constants that help stabilize size coefficients between p and i frames.

I have a GH1 as well. I'm on the vacation, but I will try you patch at the end of the week. Thanks for your efforts! Now if I can just remember how to hack the firmware... ; )

Sad, no more GH1 users =(

Here is the settings for NTSC. 68 Mbps. 24.000 fps after deinterlace. I changed a P-scale matrix here to get better image quality on high ISO's. I'm not sure on stability of it, so can't recommend it for day by day usage yet. Also, there is two new patches, "2997i to 2400" and "2997i to 2397". Download third one setb.ini please.

I have a GH1 on the way. I could help to test @humpman patch. About the GH1 60i mode diagonal rain. Does it means to overcome the problem is to go native 24p and sacrifice reliability or stay with a QPmin=19 like what @apefos found and sacrifice detailization?

Last time I mixed your QP setting to @Lpowel peak reliability patch in i60. I perceived no diagonal rain even with QPmin=0 on Death Chart. I will retest.