All areas are toy use "local" number generator, in order to keep
behavior predictable and threadable. This is what BLI_rng_() API
is for.
There are still lots of usages of BLI_frand(), which are to be
ported to BLI_rng_get_float(). but that is somewhat involved.
For the time being, remove unsafe API, so new areas have zero
chance using it.
Two fixes here (only the second one is strictly needed to fix the issue,
but both make the system better).
First is introduction of a random generator array for use with threaded
systems where each thread needs to access its own number generator.
The random texture now uses this so it should not be influenced by other
random generator reseedings of the main random generator like it did
before.
Second, I reshuffled the texture code to resample the upper bits of the
random number first. According to Numerical Recipes, this is where the
most variance can be found, so by sampling those we avoid correlation
issues. Also, multiplying here is not ideal because if a pair of bits
are zero, then the whole result will also be zero.
Overall this is much more random (tm) than before, however result will
also be brighter, since we now have less black spots. Tweaking the
brightness/contrast should somewhat fix that, generally having the same
result as before is not possible anyway if we are to really fix this.
Also, seems like exposing procedural depth might be nice here since it
influences the precision of the texture lookup.
generator with a local one. It's not thread safe and will not give repeatable
results, so in most cases it should not be used.
Also fixes#34992 where the noise texture of a displacement modifier was not
properly random in opengl animation render, because the seed got reset to a
fixed value by an unrelated function while for final render it changed each
frame.
That whole "element re-arrange" area could use some love, though (e.g. make a single vert operator with more options, as faces; and make faces work in edit mode, probably no more reasons to switch to object mode?). Post-release TODO, anyway.
Also spotted a glitch in BLI_rand code, imho, which makes first element of an array not guarantied to be shuffled by BLI_array_randomize()... No correction though, as this might/would affect other parts of the code!
=========
- Fix crash in particle transform with the particle system not editable.
- Particle child distribution and caching is now multithreaded.
- Child particles now have a separate Render Amount next to the existing
Amount. The render amount particles are now only distributed and cached
at render time, which should make editing with child particles faster.
- Two new options for diffuse strand shading:
- Surface Diffuse: computes the strand normal taking the normal at
the surface into account.
- Blending Distance: the distance in Blender units over which to
blend in the normal at the surface.
- Special strand rendering for more memory efficient and faster hair and
grass. This is a work in progress, and has a number of known issues,
don't report bugs to me for this feature yet.
More info:
http://www.blender.org/development/current-projects/changes-since-244/particles/
Removed all limitations from render code for maximum threads. The only
define for this now is in BLI_threads.h, and currently set to 8.
Note that each thread renders an entire tile, and also allocates the
buffers for the tiles, so; more threads might work better with smaller
tiles.
IMPORTANT: node system won't work yet with more than 2 threads! So, don't
try material nodes or compositing with over 2 threads. That I'll commit
later today.
What does work (should work :) is AO and soft shadow now.
Random seeding is still not perfect in render, especially lack of good
thread support still.
- VectorBlur node was calling seed for each exec, causing other nodes to
get fixed random too.
- added seed in non-OSA main loop for render
- use BLI_srandom, is better than BLI_srand
Error in BLI_array_randomize(), an unused call in Blender until the
Build Modifier was added. It did a weird modulus I don't really grasp...
somehow that caused the bug (create a corrupt mesh). But why? Dunno...
- AO and soft shadow AreaLight tables were generated without fixed seed,
causing animations to give unwanted amounts of noise.
- Made sure these tables now are calculated before render, with fixed seed
- Then found out the BLI_rand() has very bad seeding... it showed up as
patterns. After some experimenting, found a nice method using noise.c
hash tables. For compatibility with old code, named it BLI_srandom() to
use this next to the BLI_srand(). This follows libc rand() and random()
naming convention.
- Then of course threading should work... so made a BLI_thread_rand version
of the calls. Now supports up to 16 threads, comments added in .h and .c
Result is stable animation render with AO and soft shadow. But, please
test and feedback!
