Personal View site logo
Panasonic GH1 firmware research, testing
  • 277 Replies sorted by
  • @apefos ipx8
    death chart image iso 3200, low light imageimage

    ip8x_death_chart.jpg
    1293 x 681 - 262K
    ip8x_3200_iso.jpg
    1303 x 691 - 286K
    00027.MTS_snapshot_00.01_[2015.06.10_20.31.01].jpg
    1920 x 1080 - 177K
  • did not do deital test till I'm sure its stable. Spanning test: stapes after ascend file

  • @humpman in gh2 there are some tables called "deblocking tables" it helps to avoid the diagonal rain problem and also minimizes macroblocking, did you see these tables in gh1 firmware? thanks.

  • @konjow many thanks for the test. images looks good! did you like it in low light? (I repeated the images in the topic where I developed it for reference) also, does 720p works good? thanks!!!

  • ip8x_correct. patch seems to be ok 2 pics attached. biterate 34 Span test I switched off too soon 3.9gb in 30 minutes.

    f1.7.jpg
    1920 x 1080 - 709K
    f8.jpg
    1920 x 1080 - 896K
  • the patches and user file below are revised and correct. if they do not work, so I will need the GH1 body to do tests...

    @humpman, please, can you confirm if the user ini file below is correct? are the offset values correct for both interlaced and progressive? are all typo correct? thanks.

    ip8x_abcde_correct.zip
    3K
  • I did simple spanning test at night. Don't know about quality because its not stable.:)

  • @konjow thanks for the test. I found some typo error in the matrix design and I corrected them. Now there is no more typo error and should work ok.

    ip8x_correct.zip
    1K
  • @apefos just some another couple tables, seems that not affected on anything.

  • @konjow then try to increase Min qp, 10, 15 etc. No time for testing yet. What about quality compare to previous settings?

  • @humpman @apefos I did quick test for spanning. And the results are:

    setd: motion error quite often unless camera steady like rock on sandisk extrim 45mbs 32gb pictures attached

    ip8x atrefacts over the place unless camera steady like rock 2 pictures attached -files can't be play by QT(black screen)

    @humpman little lower please :)

    8.jpg
    1920 x 1080 - 491K
    8.png
    1565 x 881 - 2M
    d.jpg
    1920 x 1080 - 748K
  • @konjow keep user.ini file in same folder of ptool, it will enable the user patches for matrix and qp. If you can, please try the patch inside the "i-p_8x.zip" in my previous post, it should be stable and good quality.

    @humpman I saw the patch and user, in the user there are new things after scalling tables for p frames (E2, E3, E4, E5, E6, E7, with new arrays) what are these for?

    your screenshot shows the I frames very big and P frames very low size. decrease I frames size is important for stability, but just the GOP TABLES can do this, is there a chance for you to find it?

    thanks

  • I will do span test as its essential for me

  • sorry but what to do with user uni file?

  • Screen above i get with these settings, @apefos thx for organization of user.ini

    setd.ini
    581B
    user.ini
    5K
  • ok I can do quick test tomorrow. Which one you like me to check guys?

  • @humpman thanks for the streamparser screenshot. it shows and confirms what I was wondering to see: the average I frames size is about 8x the P frames size. This is useful information to create a patch which keeps the frames size under camera limits.

    I developed a patch which has the gop and datarate tweaked to keep the I frames size under the constant values considering this 8x ratio between I and P. This patch is tweaked for 720p and 1080p, pal and ntsc, with custom matrix.

    please give it a try, and remember to record more than 20 seconds in each take to get a better streamparser file reading:

    i-p_8x.zip
    1K
  • @apefos like this, that is on death chart.

    stream.jpg
    1297 x 687 - 288K
  • In my researches about GH1 previous patches, seeing streamparser screenshots, and comparing to the GH2 behavior, I perceived that in GH1 the I frames size can be very big and P frames size can be very small in size.

    So, when increasing the datarate in GH1, the I frames size can become a problem for the frame buffer and frame limits, because they become too big. Also the GH1 options in PTool do not provide complete settings to tweak the frame limits and frame buffer as we can do in GH2.

    The solution for this would be to tweak the GOP TABLES to keep the I frames size within something about 5:1 compression ratio which is pretty good quality for an I frame and this would keep the I frames size to be not too much big and the camera limits and buffer could work ok even with higher datarates. The GOP TABLES shares the data among the I and P frames and can make the I frames size bigger or smaller, when it increase I frame size it decrease P frame size, and when it decrease the I frame size it increase P frames size.

    In my math calculations I perceived that 32Mbps with Overall 34 can probably be the maximum stable for the camera, after I found some informations about the maximum frame size which is possible for the GH1 to handle.

    So finding the GOP TABLES in GH1 firmware and allow to tweak it in PTool can be very useful to achieve higher stable datarates. Also useful to get better noise and texture avoiding mud because bigger P frames are welcome to do this.

  • What I like about this setting is that I got only was speed error with our freezing and that was Sandisc ultimat - so slaw card. Before I could not film in forest. Sorry but currently I bant do any tests for a while. Will be back to it in about 10 days

  • I don't like this ones, it appears even on static i-frames.

    d26f43634e8be5ba7b082766123756.jpg
    1920 x 1080 - 783K
  • @humpman you need to find the "GOP TABLES" It is what shares the data between the I and P frames and this calibration helps to avoid mud and preserve noise, (see GH2 firmware in PTool).

  • @yskunto personally i don't like mud in shadows, on trees etc. You can see it on screenshots above compare to GM1. So i trying to solve it.

  • Keep the good work!

    Can you describe, what actually you do with the setting and what do you expect from the new setting (eg. reliability, perceived noise performance, etc)? So anyone that would like to test can focus their effort on that particular aspect.
  • Also try this settings, tell me if you can see differences compare to previous settings.

    seta.ini
    858B
    user.ini
    3K