I unfortunately don't have time to do it myself even though it is an easy
job, but could someone rip out the internal conversions into and out of
"sigmoidal color space" and just use whatever is incoming (linear float
light, if I remember right) both for the EWA Robidoux component (already
uses linear light) and the tensor Mitchell-Netravali component (converts
inputs by passing them through a sigmoid, convolves, undoes the sigmoid)?
If things look weird after this change, just keep an eye for constant
folding. I don't remember doing any that affects the above change, but I've
not looked at this code for a while.
gimp-developer-list mailing list
List address:    gimp-developer-list@gnome.org
List membership: https://mail.gnome.org/mailman/listinfo/gimp-developer-list
List archives:   https://mail.gnome.org/archives/gimp-developer-list

Reply via email to