A workaround could be to bring the precision you needed into the 8bit range:
Use a core image patch to (...)
I tried using a core image patch to preprocess my 16 bit image,
producing a set of two 8 bit images ("most significant" and "least
significant"). Turns out that Core Image patches insist on doing color
correction, which ruins my 16 bit coordinate lookup map (colorspace gets
converted from uncorrected to linear rgb). The approach works when I
turn off all color correction in the hidden preferences, but that is not
an acceptable option for me.
I ended up finding a fix for the issues I was having with the pure Core
Image kernel approach, so I will be using that. If someone wants to see
my approach splitting a 16bit image into two 8bit images, I have updated
the archive to include both the updated (not working) GLSL approach, and
the updated Core Image approach:
http://files.fieldofview.com/temp/qc_uvshadertest.zip
'do
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list (Quartzcomposer-dev@lists.apple.com)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
