Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
Please, support PV!
It allows to keep PV going, with more focus towards AI, but keeping be one of the few truly independent places.
The APEFOS Settings
  • 45 Replies sorted by
  • This is not related to patches, it is the way the GH2 interpret colors.

    in low light, the 3100 K (kelvin) with magenta/green and ambar/blue in the middle default will avoid chroma noise, than you do color adjusts in the post production, some green reduction and some white balance. more work for post production...

    if you prefer to adjust the white balance to get good colors from camera and avoid post production, you need to first set the K (kelvin), than you set the other adjust more to magenta a little.

    after each adjust you need to start recording to perceive changes, because GH2 shows different colors when you press the record button, and this is what the camera is recording.

    also the kelvin number sometimes is different from the charts we see. example: charts say that sunlight is 5200K but a better result in GH2 can be found with another number...

  • Here they are, all the apefos patches, 6 versions: 3 luma matrix design, and 2 datarates:

    txt file in the zip has all informations, please read.

    recomended version = 2C = setc

    important information: in 720p60 SH with OIS and iDynamic enabled, using manual exposure, whit a native lens, when you change the fstop or the shutter speed, the information in the LCD display can take some time to update, because the camera processor is working a lot to compress the image. this can happens in other recording modes also. turning off LCD information display can helps stability.

    apefos_patches.zip
    65K
  • how is 2C different from Apefos Patch v2? Thx!

  • Versions with very high datarates in 720p, 1080i and hbr for people who likes to be on the extreme edge

    Please read the txt file inside the zip

    apefos_high.zip
    5K
  • in my last two uploads now there is set a, b, c, d, e, f, h, i, j

    three matrix designs, three datarates = 9 patches

    there is no set G yet...

    my choice to be the set G is the 2C, setc

    what will be your choice among them? please read the txt inside the two zips, do your tests and vote.

    Thanks

  • @apefos If it can be useful, I tried the apefos_patches. For me I would choose the 2d. There is something with colors which seems to make it better, while the noise reduction is always very good. But the difference is so little. Now I'm downloading these last three settings.

  • developed version G with this luma matrix version 4 and three datarates options:

    006 009 012 018

    009 015 021 036

    012 021 045 054

    018 036 054 072

    can be a good option for a balance between lower isos and higher isos in same patch

    all patches are here in this zip with the txt. read the txt and you will understand

    apefos_patches_high_g.zip
    130K
  • @frullaccia the difference is subtle, more noticeable in low light, in light brown surfaces, and light gradients.

    the patches changes texture, noise and artifacts, it does not change colors.

    in PAL lands the 2D can work ok, but in NTSC lands the 2C will be more stable in 720p and HBR.

    best comparison to do now is: 2C vs. 3E vs. 4G-low (all in iso 1600 in low light) this will show wich one has better textures, noise and avoids artifacts the most. They are the more stable versions, same datarates, and subtle differences in luma matrix (see the txt file inside the zip).

  • I have used the teleconverter on, full zoom, at 800 iso. Sunset and morning.

    The difference is little. Of course I need to test it again, but I choose 4high; 2d and 2c.

  • I did some tests right now.

    Camera on tripod, same wb, same iso, same lens, same focus, same fstop, same shutter, same picture profile, same exposure, same framing, same light, same everything.

    Second test handheld with movements and zoom in low and high shutter speed, low light and daylight.

    Also did slow down from 60p to 24p to perceive noise and texture.

    The matrix 3 wins all the way, the texture and noise are just perfect: "I don't even see the code", and it also does this with the datarate in the lowest setting. Great! It is the version 3E.

    Got it working flawless on the formated 45Mbps card and also in the 95MBps card in 720p60 SH NTSC with OIS and iDynamic enabled in daylight with lots of details in iso 2000!!!

    For my eyes it is the Arri Alexa inside the GH2. Amazing image quality!

    Interesting thing: the on screen display does not show significant delay, maybe this is the result from a good maths in the matrix, the matrix 3, wich uses same frequency increase for all the lines.

    Remember that it can show artifacts in iso 2500 and above, and also in iso 1250 and above if iDynamic is enabled, keep iso 800 and below if using iDynamic. Iso 2000 is working pretty good in low light without iDynamic.

    If you do not care for auto electronic features and/or native lenses, the Version 3F, setf, increases datarate, and version 3J, setj increases datarate even more, both with same matrix 3.

    I believe that the versions 3J will be just amazing for very high isos, or to handle perfect texture when there is lots of light gradients and shadows in the face. the perfect matrix with enough datarate to handle all the noise!!!

    If you are working for DCP (digital cinema package) output for theather projection, the noise and texture from the matrix 3 will amaze you, no need to do denoise, no need to add digital film noise simulation.

    The conversion to apple hq mov using the 5dtorgb works ok.

    This is the end of patches research for me. Period.

  • Apefos 3G

    an attempt to put the GH2 in the extreme limit for image quality..

    using the luma matrix 3 and highest possible datarate for 720p and hbr.

    avoid native lenses and auto electronic features.

    apefos_3g.zip
    10K
  • I think this is the final work...

    apefos_settings.zip
    31K
  • APEFOS SUBTLE VERSIONS G H I J:

    subtle datarate increases, this thing is more stable than I was thinking, so I made four datarate versions:

    apefos_settings_ghij.zip
    41K
  • I got the Version seti from the last upload working in 720p60 NTSC SH, in manual exposure, OIS enabled, iDynamic enabled, AFS autofocus (focus pressing the shutter button), 14-42mm native lens, in iso 320 in the 95MBps card in a detailed outdoor scene.

    Now it is a matter of you to find the best version from last upload which works ok for your camera/card/lens/features/scene combo.

  • My final decision is:

    To keep the version G from last upload on the camera.

    And use the version J from last upload when I need best possible texture noise.

    The 95MBps card is a little more stable than the 45MBps card, but not enough to completely avoid stop recording with iDynamic enabled in version H, I and J. If the camera is steady the 95MBps card makes possible to do the shoot. The version G works in the 45MBps card.

    So better thing to do is to use the version G for most situations, and version J just if you are a pixel pepper for a DCP output for theather projection, using a powerful intermediate codec. For the WEB and BluRay, version G from last upload is stable and more than enough.

    The version G from last upload works ok in the Death Chart Movie test with iDynamic enabled (but not active), with OIS enabled, in iso 2000, shutter 60, f11, with the 14-42 native lens in 720p60 SH NTSC which is the most difficult recording mode. In the real world in iso 320 the iDynamic get active and works ok, it is stable and the texture in the raised areas are ok. Also, good thing is that the image recorded from the Death Chart movie shows excelent texture in 720p SH. In low light it is amazing noise quality.

    The version J will be safe with OIS and iDynamic disabled, and also with manual lenses, not with native lenses, but maybe it will not worth the effort of firmware update and the drawback of not use the auto electronic features.

    The only situation I saw an image improvement using version J was with the face with light in just one side and other side under shadow. In this situation the set J higher datarate shows a little better texture on the shadow side of the face in iso 2000, but almost unoticeable. in iso 160 it is hard to see any difference. But if you recompress to bluray or web this difference will never be useful, just for a theather presentation in DCP output using a strong intermediate codec for editing.

  • @apefos "A theather presentation in DCP output using a strong intermediate codec for editing" is what I often need. Making my tests with the previous set of patches I noticed that in 24P, Pal mode, the 4th setting seems reliable. What I did was: stressing the camera with high contrast in low evening lights, pointing at a group of dark trees under a bright sunset sky, with full zoom and teleconverter on. This was the worst condition for the camera. The result was a very noisy pattern crossed in dark areas by horizontal stripes. I also tried the patches at daytime. Well, as I said before, differences are very little. My impression was that the 4th setting could be the finest. But I could have been wrong.

  • @frullaccia there is nothing else to do. the last uploaded patches reach the camera limit. Get my advice: buy the Alexa Mini.

    Now I will keep away again...

    Thanks!

  • @apefos I know there's a limit. I am ok with these patches. As for the Alexa Mini I don't think that camera is worthy to pay 36.000 euro.

  • It seems I found a way to solve the diagonal rain pattern problem.

    and also to make the G version with stable datarate to be similar good texture to the J version which is high datarate.

    I am doing tests.

  • Diagonal Rain Pulsing Pattern is Solved!

    Also, the stable datarate version is showing great textured in shadows and no need to use higher datarate versions anymore!

    Careful tweak in Deblocking Tables did the final magic!

    I am calling this "The Patch"

    the_patch_by_apefos.zip
    10K