Float-Point?

Avalon
Avalon Posts: 95 Just Starting Out*
edited March 2013 in Post-production techniques
Wondering what 'float-point processing' means to HitFilm projects.
When using float-point video processing there should be no irreversible under- or overexposure done by any internal processing like color correction filters. E. g. if you apply a levels filter to a linear grey-scale (ranging from 0=black to 1=white), set input black and input white to 0.5 the grey-scale will turn into one half being black, the other half being white. Apply a second levels filter and set output black and output white to 0.5 this will restore the grey-scale.
In Vegas Pro and Nuke this works perfectly. In After Effects it works rather good, but not perfect (I assume the math they use isn't quite correct). In HitFilm this does not seem to work. Also importing float-point based video sources does not seem to work in float-point mode in HitFilm.
Which processes in HitFilm are able to use float-point processing? Filter (like color corrections) don't, do they?

Comments

  • SimonKJones
    SimonKJones Posts: 4,448 Enthusiast
    Hi Avalon. An end-to-end, less-destructive workflow similar to what you describe is something we aim to implement in a future version.
    I talked to one of our devs about your questions and received some feedback that may be of interest:
    HitFilm does use floating point precision but sometimes it clamps between the range 0-1. HitFilm provides the option for 8bit and 16bit but all input/output is clamped in the range 0-255 or 0-1 depending on bit depth. This clamping occurs on some filters.
  • Avalon
    Avalon Posts: 95 Just Starting Out*
    Thanks for that infos!