Things should always work the way users expect, but sometimes implementation
gets in the way.
The output side of the operator is calculated by the algorithm, it is the
true size that the effect will produce given its input parameters. The
Resize node is just a different UI on the transform 3D, so when you put an
exact pixel value what it does is in fact figure out the right numbers to
poke in a transform matrix to produce the output. The image size produced
is the result of that matrix transform, not the number you entered you
entered in the PPG. It must work this way because I need to stamp out any
time that the results are not correct, and not hide the bugs by forcing the
output to exact pixel. Which would give an image the right size, but with
some aliasing garbage or missing lines.
I'd like to have clear repro steps on how to have it not work right, so we
can address any remain issue.
Note that you need to try this with a scene created at least in XSI 4.2,
because the resize node checks the scene version and emulates old bugs in
order to maintain compatibility. If you worked around a bug, for example by
adding 1 to the desired size, I must maintain that behaviour in existing
scenes. Cuz I'm that nice of a guy.
------------------
Luc-Eric Rousseau
Team Leader, User Interface
Softimage|XSI
---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi
