[GM-help] DPX woes with gm1.2 unreleased Q16

[Jos'h F. <Jos'h.Fuller@Starz.com>]

Hi,

I'm trying to convert sixteen bit tiff files to 10 bit log DPX's with
GraphicsMagick 1.2.020061102 and running into a spot of bother.

I'm starting with one of the standard EFilm calibration images. I load
it into Eyeon Digital Fusion and convert it to a 16 bit tiff, which
matches our actual composite output. At this point, I would like to
convert it from tiff to dpx using GraphicsMagick to load on our Scratch
DI system. However, using this command line:

gm32.exe convert -define tiff:alpha=3Dunassociated +matte  =
efilm_cal.tiff
-define dpx:bits-per-sample=3D10 -colorspace CineonLog =
efilm_cal_gm32.dpx

However, the image I get is very blown out and "quantized" -- i.e. what
should be a smooth grey ramp shows "stairstep" aliasing on the waveform
graph. It is also clipped, as in the top bar of the chart which should
read 1023 is instead cut to 685 (white point), blacks are also clipped
to 95. It's like the view conversion is being saved instead of passing
through the original pixel data. By the way, the "32" in gm32 means that
this one has been compiled with 32 bit quantum depth.

All the data is in the tiff file -- if I load it into Fusion and save it
out as dpx then I've returned to the original. However, I need to be
able to close the loop with GM instead!

Thanks in advance!

Jos'h Fuller
Starz Animation
416-682-5200 x 5395=20