On May 31, 2018, at 2:55 PM, Patrick Cusack <[email protected]> wrote:
>
> I definately couldn’t understand how endianness would be related to this. If
> there were an endianness issue, the picture would be different and the issue
> wouldn’t evidence itself with a flip.
Yeah, that was my thought, too.
I would not have easily guessed orientation was an issue without your example
image and code. It's actually fairly rare to see TIFF files with non-default
orientation.
When I saw that you were making a fresh ImageSpec instead of copying the one
from the input, that's when I got suspicious that maybe it was related to the
uncopied metadata.
The one circumstance when I *DO* see orientation metadata used typically and
correctly is for JPEG or RAW images from digital cameras. When you turn the
camera sideways, it's got a sensor that detects this. But rather than
re-ordering the data on the camera (it wants to be fast and power-efficient),
it writes the image in the usual sensor order but just sets the orientation
metadata in the file so that your display program (hopefully) rotates it
properly upon display.
-- lg
--
Larry Gritz
[email protected]
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
