Checker: FORWARD_NULL (help)
File: base/src/source/blender/render/intern/source/texture.c
Function: do_lamp_tex
Description: Variable "co" tracked as NULL was dereferenced.
co was set to NULL at the beginning of the function and it could
possibly slip through all the logic above so lets test it before
we use it blindly.
Kent
* Missing changes to release/windows/installer
* Sequencer fixes in source/blender/src/seqaudio.c dont apply to 2.5
* brechts fix for #18855 r20763 wasnt merged, does this apply to 2.5?
*Octree works nicely on hierarchic trees
*the old code was quite destructive at the Isect variable
changing isec->start, end, vec..now it only changes isec->labda (and hit results)
Currently rendering a BVH of all objects, where each object has it own octree.
Start of planned DerivedMesh refactoring. The mface
interfaces in DerivedMesh have been renamed to reflect
their new status as tesselated face interfaces (rather
then the primary ones, which are now stored in mpolys).
short review: mpolys store "primary" face data, while
mfaces store the tesselated form of the mesh (generally
as triangles). mpolys are defined by mloops, and each
mpoly defines a range of loops it "owns" in the main
mloop array.
I've also added basic read-only face iterators, which
are implemented for CDDM, ccgsubsurf, and the bmeditmesh
derivedmesh. Since faces are now variable-length things,
trying to implement the same interface as mfaces would not
have worked well (especially since faces are stored as
an mpoly + a range of mloops).
I figure first we can evaluate these simple read-only
face iterators, then decide if a) we like using iterators
in DerivedMesh, b) how much of it should use them, and c)
if we want write-capable iterators.
I plan to write official docs on this design after I get
it more stable; I'm committing now because there's a rather
lot of changes, and I might do a merge soon.
svn merge https://svn.blender.org/svnroot/bf-blender/trunk/blender -r19820:HEAD
Notes:
* Game and sequencer RNA, and sequencer header are now out of date
a bit after changes in trunk.
* I didn't know how to port these bugfixes, most likely they are
not needed anymore.
* Fix "duplicate strip" always increase the user count for ipo.
* IPO pinning on sequencer strips was lost during Undo.
Preview renders:
- Added proper button type (BUT_EXTRA) for preview buttons, to
handle drawing better. It now first draws an alpha mask, to
ensure the preview is correctly fitting inside the widget style.
It then draws the outline.
- Added protection for executing preview renders while regular
rendering, that's not going to work...
Nice goodie: Preview renders!
- Added new preview.blend, allowing super wide cinemascope previews
- Draws nicely blended inside widget type, rounded corners
- Preview now renders using all available cpus/cores.
- Uses - hopefully rock stable - method, which doesn't copy or
allocate anything for previews, but just uses render API calls.
- Multiple previews are possible! But, added provision in Jobs
manager to only render one preview job at a time. If you start
more preview jobs, they're suspended until it's their turn.
Bugfix: new buttons context code crashed when going full-window.
Tweaks are still needed for notifiers. I have to figure out still
how to retrieve SpaceButs button view types...
-Many things not yet working properly :)
-Experimental new (hopefully more logical) categorization of buttons, feedback on this is very welcome.
-Separate render & draw types, for example now there is actually a render option "halo" instead of all the different point draw types.
-Particles get recalculated only from buttons that actually change something that has to be recalculated, for example changing visualization doesn't reset particles any more.
-Boid physics buttons are still missing as I'm currently redoing the whole boids code.
-Point caching is still very wip, so baking is not possible for example, but I added a few cache baking flags for rna that were/will be needed for particle buttons logic.
Cleanup commit:
- Yafray removed. Also did cmake/scons files, but didn't compile with it,
so test would be appreciated :)
- Removed old crap from Windows release dir, should be checked on
further by windows release builder later.
My bug fix to support AO with "Amb" texture channel changed code too that calls
AO as a pre-shade process, when texture isn't calculated yet. This caused very
first pixel in a tile to show wrong AO.
Especially myself deserves a kick in butt for not testing the regression files
for release binaries again! Error shows clearly... in the cornelius_passes
.blend file.
But I still wonder about its correction because it looks the
oposite of whats documented and of what was used before :S
I also took a long time to find it because I tought blender was feeding the raytrace structure
with quads and triangles but it looks the quads get triangulated before reaching makeraytree (on rayshade.c).
Third transparent shadow bug... this time it's a Material Node, which
has mirror + transp-shadow on, and when it traces its own material it
enters an eternal loop...
Raytracing + shading + materialnode combo really needs work!
* Added material "type" property, with Surface/Volume/Halo
options, compatible with sim_physics, as requested for
material buttons layout. Obviously the Volume setting
does nothing currently.
* Deprecated MA_HALO flag in favor of this.
fix/workaround - offset by a 500th of a pixel to avoid baking missing pixels that are between 2 faces, its still possible pixels could be between faces but much less likely then it is currently with pixel aligned UVs.
*Adapted octree to a more generic raytrace API
*ray shadow works (other untested stuff disabled atm)
On the scene tested the user-cpu time got from 1:24 to 1:19/20
probably because of removed callbacks or sligtly diferente memory usage
Old bug (2.42): when using node material, transparent shadow did not work.
It was missing to set the proper 'pass flag'.
Do note an important difference with non-node materials for 'transparent shadow'.
If there are no nodes, it uses the color from the unshaded material. When it has
nodes, it uses the color output from the entire node tree, which is typically
from shaded materials. The latter is because node shaders have no support for
shade passes yet (it only outputs rgb + a).
Ray-transparent didn't pass on thread number to shading code, giving
"blothes" in render, when using node materials.
This also rewinds Campbells commit of feb 21, which tackled the error,
but not the cause.
Using "Key Alpha" didn't work when using MBlur render. The accumulation
code was assuming regular alpha then. Now it corrects for it.
Still it's a bit of a weak spot in Blender's render system. I will look
in the future to make this a real post process; converting all RGBA
buffers in the system, including for all passes, to "key alpha".
Combined with that our compositor should become alpha type aware too.
Everything in Blender assumes premul alpha, which still just will work
best in general...
- Removed the gen_utils.c dependency from Mathutils (since gen_utils wont go into 2.5 but mathutils will), repalced with python functions.
- removed Blender.Mathutils.Point, since it was not documented, the C api never used it, none of our scripts used it (and I never saw a script that used it).
Freestyle's pipeline is now fully controllable at the layer level. It can be used:
- in any render layer
- with as many style modules per layer
DETAILS:
Freestyle usage has not changed:
- all the configuration happens in the "Freestyle" render panel, after it is enabled in the "Output" panel with the 'Freestyle' toggle.
- each render layer can choose to render Freestyle strokes by togglingo on 'FrSt' (in the "Render Layers" panel)
- it is fully compatible with compositor nodes
In the "Freestyle" panel, a render layer is selected via the menu list next to the "Render Layer:" label. The options displayed below are those of the currently selected render layer (and are not global to all render layers, as was previously the case).
Style modules are added by pressing the lower button "Add style module". Once added, the following operations are possible:
- deletion (cross)
- reordering (up/down arrows)
- toggling of display (check)
The order of the style modules follows Freestyle's original convention: the modules in the list from top to bottom are respectively the first to the last composited in the render layer. For example, if the module list is "contour" followed by "cartoon", the "cartoon" strokes are rendered on top of the "contour" strokes.
The "Freestyle" panel is constantly synchronized with the "Render Layers" panel: if render layers are added, deleted or toggled off display, Freestyle will take note of the changes.
The current pipeline works as follows:
----------------------------------------------------------------------------------------------
for every scene that is being rendered
if Freestyle is enabled globally
Freestyle is initialized
camera and view settings are transferred from Blender to Freestyle
for every render layer
if: - layer is enabled
- layer enabled Freestyle
- the number of displayed style modules is non-zero
canvas is cleared
geometry is transferred from Blender to Freestyle
settings are fixed for current iteration
view map is calculated
strokes are computed in the canvas (based on view map and style modules)
strokes are rendered in separate Blender scene
scene is composited in current layer
----------------------------------------------------------------------------------------------
A number of changes were made on the codebase:
- the rendering interface between Freestyle and Blender was simplified. The following naming convention was used: functions that are called from within Blender pipeline are prefixed with 'FRS_', while the variables are prefixed with 'freestyle_'
- Freestyle data structures that were put in Blender's render pipeline were removed
- Freestyle cleans up its data structures on Blender exit and shouldn't leak memory
- to ease the configuration in the "Freestyle" panel, a centralized configuration data structure was used and can be easily extended
LIMITATIONS
Even though the current commit is stable and achieves the intended result, it is not as efficient as it could be:
- the canvas and the style modules are at cleared at each layer-level render
- geometry is reloaded at each frame and is duplicated across render layers
This revision clarifies my understanding of the future role of the view map in the compositor. Unfortunately, contrary to what the original proposal said, it is impossible to provide the view map as a render pass because render passes are defined (RE_pipeline.h) as raw floating-point rects. We will have to determine whether or not to extend the notion of render pass to fully integrate the view map in the compositor.
