On 2/11/22 14:42, Kenneth wrote:
> Sorry for the long delay, I needed to wrap my mind around your very useful
> comments and to run some more tests.
> 
> This camera-normal stuff is indeed weird, and certainly not as simple to fix as
> I thought.

No worries - this our hobby - we play when able. ...And, I cannot count 
the number of times I've mumbled to myself, "...not as simple to fix as 
I thought." :-)

Aside: I find your images here appealing.

---
I'll start by saying I'm sure I don't see or understand all that can 
happen when apply a normal block to the camera. That said, I think your 
general rules more or less correct.

(2) - Whether the pattern is centered or not will depend on the pattern. 
Internally the coordinates any pattern sees from the camera are 
-0.5,-0.5,0 to +0.5,+0.5,0.0.

(3) - Yes. The camera direction vector is effectively the raw normal 
into the normal pattern. (I've wondered whether there are any AA induced 
distortions when perturb camera rays with a normal{}. We, there, are 
twiddling with ray directions)

(4) - Yes. Similarly, beyond the simplest user defined functional camera 
set ups, figuring out how to set up a camera in a way which behaves in 
an understandable across all possible transformations is difficult.

Aside: Unsure I mentioned it. There is an odd check in the normal.cpp 
code whether a valid intersection exists. I initially took it to be an 
unneeded conditional and added an assert to see if it ever tripped while 
rendering. Camera normal{} and kaboom. Then had a reason for the 
conditional and also served notice the perturbed normal information is 
getting back to the camera direction rays by a somewhat different means 
than usual.

Bill P.

Post a reply to this message