This problem was in both new bump and new new bump. The derivatives used for
bump mapping did not include the mapping, which gave mismatched offset for the
taps, now just do the mapping beforehand.
Many thanks to them!
For comparison, see here:
http://kishalmi.servus.at/3D/bumpcode/
Based on algorithm in: Mikkelsen M. S.: Simulation of Wrinkled Surfaces Revisited.
http://jbit.net/~sparky/sfgrad_bump/mm_sfgrad_bump.pdf
This fixes bugs:
#24591: Artefacts/strange normal mapping when anti-aliasing is on
#24735: Error at the Normal function.
#24962: Normals are not calculated correctly if anti-aliasing is off
#25103: Weird artefacts in Normal
This will break render compatibility a bit, but fixing this bugs would have also
done that, so in this case it should be acceptable.
Patch committed with these modifications:
* Bump method Old/3-Tap/5-Tap option in UI, 3-Tap is default
* Only compute normal perturbation vectors when needed
* Fix some middle of block variable definitions for MSVC
* Renamed children to "simple" and "interpolated" as this is
easier to explain and more descriptive than "from particles"
and "from faces".
* Also shuffled the child ui around a bit to make it clearer.
* Child seed parameter allows to change the seed for children
independent of the main seed value.
* Long hair mode for interpolated children:
- Making even haircuts was impossible before as the child
strand lengths were even, but their root coordinates were
not similar in relation to the parent strands.
- The "long hair" option uses the tips of the parent strands
to calculate the child strand tips.
* Hair parting options:
- Hair parting can now be calculated dynamically on the fly
when in 2.49 there was a cumbersome way of using emitter mesh
seams to define parting lines.
- For long hair parting can be created by a tip distance/root
distance threshold. For example setting the minimum threshold
to 2.0 creates partings between children belonging to parents
with tip distance of three times the root distance
((1+2)*root distance).
- For short hair the parting thresholds are used as angles
between the root directions.
* New kink parameters:
- Kink flatness calculates kink into a shape that would have
been achieved with an actual curling iron.
- Kink amplitude clump determines how much the main clump value
effects the kink amplitude.
- The beginning of kink is now smoothed to make the hair look
more natural close to the roots.
* Some bugs fixed along the way too:
- Child parent's were not determined correctly in some cases.
- Children didn't always look correct in particle mode.
- Changing child parameters caused actual particles to be
recalculated.
* Also cleaned up some deprecated code.
All in all there should be no real changes to how old files look
(except perhaps a bit better!), but the new options should make
hair/fur creation a bit more enjoyable. I'll try to make a video
demonstrating the new stuff shortly.
* Two separate bugs, with very similar symptoms.
* The distribution binary search didn't work correctly in cases where there were a lot of faces with 0 weights.
* Maximum distribution sum should have been exactly 1, but due to the wonderful nature of floats this wasn't the case at all.
from Dan Eicher (dna)
From the tracker (with minor edits)
========================
cube = bpy.data.objects['Cube']
foo = cube.vertex_groups.new('foo')
foo.add([1,3,5,7], 1.0, 'ADD')
for i in range(len(cube.data.vertices)):
try:
weight = foo.weight(i)
print('vert: %i weight: %f' % (i, weight))
except:
pass
foo.remove([1,3])
cube.vertex_groups.remove(foo)
from Alexander Kuznetsov (alexk) with edits.
From the report:
Blender assumed that all files are .blend as retval = 0;
Now retval is initialized as file cannot be open (-1) for gzopen fail and directory case
retval = -2; is defined for not supported formats
This must be assigned before #ifdef WITH_PYTHON because this part can be missing
Finally retval = 0; if it is a .blend file
---
also made other edits.
- exotic.c's blend header checking was sloppy, didn't check data was actually read, only checked first 4 bytes and had a check for "blend.gz" extension which is unnecessary.
- use defines to help readability for BKE_read_exotic & BKE_read_file return values.
- no need to check for a NULL pointer before calling BKE_reportf(). (will just print to the console)
- print better reports when the file fails to load.
- avoid looping over the entire collection unless a negative index is used.
- dont use the get index function for building the slice list, instead loop over the collection until the stop value.
don't draw the image if the size is 0.
Crash was actually an assert() so debug builds only, replace assert() with BKE_assert() so crash is opt in build option.
- Toobar views were reset on hide/unhide. Now they keep the view
and zoom level.
- Added operator to delete all unused 'space data', this to make
clean startup.blend files, remove unused editors, and to be
able to test starting defaults for editors.
No hotkey, use search for "Clean-up space-data"
Compositor: Texture Node now behaves like an image.
- Image always in render output size
- Buffer outputs RGBA and Value both supported
- Works for filter and blur and scaling too.
- Mixing 2 textures works
Implementation note:
The texture node was meant to be 'procedural', not a buffer
but a color-sample method. Unfortunately the node editor
didn't support this well, blur/filter/scale ignored it too.
For now, its better to drop this procedural concept, then
things work at least as expected. :)
Auto-texture space now is more responsive and correct.
- on transforming it, the buttons get redrawn to show option was reset
- on enabling option, texture space is recalculated/reset
Issues:
- It looks a bit crowded though, so perhaps we could do without, and
just leave it for the TimeLine only?
- Due to the way the invoke() testing works, the rename operator
called from the menu currently fails. Will need to experiment more
with execution options to find a workaround.
Speedup for getting collection indices, avoid getting the collection length unless a negative index is given. This avoids a loop over the entire collection in many cases.
Speedup for getting collection slices by detecting collection[:] and internally calling collection.values(), this gives a big speedup with some collections because each slice item would loop over the list until that index was found.
Rough test with 336 objects.
- getting index of listbase collection ~ 5.0x faster
- getting index of array collection ~ 1.15x faster
- getting slices of listbase collections ~ 34.0x faster
- getting slices of array collections ~ 1.5x faster
animation editors (DopeSheet, Graph Editor, NLA, Sequencer) ==
=== Usage Notes ===
In animation editors, marker operators will only be considered while
the mouse is hovering near/over the horizontal scrollbar (i.e. where
the markers usually appear). That means, in order to do something to
the markers, just position your cursor in line with the row of
markers, and then use the same hotkeys you'd use in the TimeLine (so,
unlike in 2.4x, no more need to hold down extra modifier keys for this
case). In the TimeLine, nothing changes, so you don't need to worry
about mouse placement there :)
=== Technical Details ===
Since early 2.5 versions, this functionality has been disabled, as the
markers were always getting evaluated first, and hence "swallowing"
all the events before the editor's own keymaps could access them.
In order to get this working again, I've had to give every marker
operator a "wrapper" invoke callback which performs some checking to
ensure that the mouse is close to the markers (vertically) before the
operator will try to be run. This wrapper also makes sure that once
the operator has finished running, that if it didn't manage to do
anything, then the editor's own keymaps get to have a go.
The vertical tolerance used is currently 30 pixels (as was used for
the borderselect operator).
=== Other Assorted Changes ===
* Gave marker operators dependent on having selected markers to
operate on suitable poll() callbacks. These new poll callbacks ensure
that there are selected markers for the operator to operate on,
further cutting down the number of places where markers may override
standard hotkeys (and avoiding calls to the wrappers too)
* Simplified some of the selection code
* Made some formatting tweaks for consistency, and in one case so that
my text editor's function-list display doesn't get confused
- convertblender.c, remove assignments to unused vars.
- readfile.c, fix 2 possible crashes. null pointers were being checked for then used later without checking.
- space_graph.c, use switch statement for automatic color assignment rather then a float array.
I've been asked by Giuseppe Ghibò (joeghi) for more icons for Mandriva Linux distro.
I've noticed that some folders had a blender.svg different from the one in scalable/, so made them all the same derived from
scalable/blender.svg
Now each svg file has "document properties" of the right dimensions according to the folder it's in, and the svg is saved so that people
just have to press "export bitmap" in Inkscape in case they need it.
Outliner: using the object-data icons to enter editmode, didn't
reset the toolbar Operator redo. It was not using an operator
call here. Note to self: more outliner tools don't...
Error was that operator-redo then would repeat an action incorrect.
Code changes are minimal, re-using the code as already was there.
Options are in pulldown menu, or SHIFT+T and SHIFT+ALT+T
Might be that Martin likes to see it different... do we need
a special operatortype for it?
Object color option now is incorporated in preview render, and
in display in 3D window.
Note that Object-color is "modulating" the render result, it just
multiplies... which is very limited for practical use. This was
added for Apricot game project. The original meaning however was
to replace diffuse only... so specular and light work still correct.
ALso made header for properties editor start on top, to reflect
default (report in tracker).