This new option is located in the shadows options in the render settings.
This approach is simple and just randomize the shadow map position (not
the lamp itself) and just let the temporal supersampling do the average of
all the shadowing. The downside is that is needs quite a large number of
samples to give smooth results and individual sample position can remain
visible.
Enabling this option will make the viewport refresh all shadow maps every
redraw so it has a serious performance impact.
This approach is not physicaly based at all and will not match cycles.
----
The sampling for point lamps (spheres) is not
The old name Instance was logic when the modifier created new object instances, but now works equal to mesh Array modifier, so the old name was not logic and must be Array.
Also added a Object to use as offset similar to mesh Array modifier.
Regression introduced on e88e80a6.
This was broken for both single and multi-objects.
It is a typo that apparently slipt through testing, oh well.
I'm glad I caught this, I just wished it would have been faster.
If ob was NULL it would crash in the else part of the if statement.
If we really think we may run into that (which we should not) we can just assert
or add a if (ob == NULL) return; in the top of the function.
As it turned out the issue wasn't that we were drawing only the active,
but that the "object mode" lattice drawing was drawing on top of the
drawing for all the edit mode lattice objects.
We are doing the same original behaviour for curves and even meshes. To be
investigated if it is ok for those other cases.
Committing this for the sake of completionism. I'm going to bring this
up for review, I think we may want to revert it.
Fundamentally I'm changing the behaviour of the operator both in object
mode (acting on all selected lattice objects), as well as the edit mode
(acting on all lattices in edit mode, regardless of them having any
selected vertice).
Adding a workaround in this case: we blit the depth buffer instead of the
stencil buffer and use the copy as the texture. This is slower but at
least it should work.
When calling glBlitFramebuffer on most (if not all) mac that have a GPU
from the Radeon Pro series, the data is not properly copied and only a
subset of the pixels are correctly copied.
This only happens when blitting the depth buffer if the depth buffer is
GL_DEPTH24_STENCIL8.
Changing the depth buffer format to GPU_DEPTH32F_STENCIL8 fixes the issue
but only works if blitting the depth componnent. The stencil componnent
still provoke issues when being copied.
