Panasonic Cinelike-D to Aces

Josh Hendrix Jan 5, 2017

  1. This happens on every picture. I'll struggle with 15, 20, 25 nodes, then create a new version, blow it all out and start over, then create a different look from scratch, using simpler tools. I then A/B the look to the point I forget which is which and look only at the hero display, and I stop flipping back and forth when I like what I see. If it's the simpler grade, I go with that and consider redoing the entire scene. If the better grade was the original, I stick with that and keep moving.

    It's rare that I can do it in one node, but I've definitely had cases where I go from 20 nodes down to 4 and wind up with a better picture. It may still need a little tweak, but better is better: I don't question why or how, I just want the results.
  2. True but the sensor density is higher (20 vs 16 Mpixel) and despite better AD circuitry the DR more or less evens out according to Uematsu Michiharu, at least that is how I understood it.

    But I take 2/3 of a stop for I will have one myself in March/April. :)
  3. Specifically in GH5 video mode, 4K (~8MP) is alleged to be oversampled from the entire 20MP sensor which would be the first for Panasonic, theoretically when doing oversampling, you gain maximum 0.5EV of shadow SNR each time the pixel count is reduced by half.
  4. i once calculated ACES ITDs for the pictoire profiles of the GH4. you can get it here:

    or take a look into the source code:;a=summary

    but i wasn't very satisfied with the results and lost the interested in this approach.
    cine-d footage shows some complex color shifts related to gamut compression, which can not be reversed / linearized later.
    v-log is much more suitable in this respect, but it has other drawbacks (more noise, less tonality).
    for traditional rec709 oriented grading, cine-d may be seen as the better choice, but for more demanding color reproduction and match of different cameras, v-log has its advantages.
    Last edited: Jan 11, 2017

Share This Page