| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
BKE_scene_update_for_newframe should be called before RE_BlenderFrame in
seq_render_scene_strip. It's not entirely nice, but bot sure about better
solution for now.
|
|
Was own regressions since recursive sequencer commit.
|
|
documented but not actually added to bge.logic.
|
|
isn't beveled.
|
|
as well pass it.
|
|
|
|
unlikely but possible crash if the py-class returns new instances of PyObjects it doesnt own when registering the class.
|
|
but it was being maintained (style, docs, name changes, etc), so I'm removing it.
|
|
KX_BlenderGL.cpp. Use of glIsEnabled() is discouraged since it causes a potential sync with the graphics card. Also, it's faster to just always use glDisable() (even if that feature is already disabled) then to check if it's enabled first.
|
|
this attribute.
|
|
indent in KX_KetsjiEngine.cpp
|
|
checker)
|
|
|
|
Now we hide the LLVM symbols with an ld version script to avoid conflicts.
|
|
rectangular faces after applying rotation, reported by Metalliandi
This issue is caused by floating point precision error. After applying
rotation, the edge lengths change slightly and on rectangular faces the
length comparison can be flipped. Solved by giving a slight offset to
the length calculation for the diagonal during triangulation
calculation. (Same as done during uv unwrapping)
|
|
Was floating point precision error (FLT_EPSILON was too small).
|
|
didn't work (angle check too sensitive). Caused some
(nan,nan,nan) results from offset_in_two_planes.
|
|
Patch by David M (erwin94), thanks!
Also made Vorbis codec using float sample_fmt, otherwise it didn't work
with new FFmpeg.
Perhaps we can make it more clear by explicitly separating audio_input_buffer
for float and integer buffers, but as far as it works i'm not so fussed about
this atm.
|
|
Avoid using IMB_flipy from image save callback. It will use a bit more
memory but wold be thread-safe.
|
|
alternate fix for [#33217] Mirror Modifier not showing "Shadow" mesh in wireframe view
|
|
|
|
RE_Database_FromScene() then the main render loop.
|
|
Early output was not setting image buffer float color space
|
|
rendering), it should still unlink all inputs to avoid debug assert failure that checks for remaining Node connections (debug_check_node_connections).
|
|
using current trunk and previous release.
|
|
There were two issues in scene strip rendering:
- It will skip rendering if scene doesn't have camera but uses compositor
- G.is_break will cancel preview rendering
Also removed Use Sequencer from scene's strip settings, it's not supported.
|
|
redirecting links from the reroute output the input must be completely unlinked, otherwise the debug_check_node_connections will complain (this is a sanity check that ensures all the original Nodes have been fully reconnected to Operations).
|
|
|
|
compare with python)
|
|
|
|
|
|
- PyLong_FromSsize_t --> PyLong_FromLong
- PyLong_AsSsize_t --> PyLong_AsLong
In all places except for those where python api expects PySsize_t (index lookups mainly).
- use PyBool_FromLong in a few areas of the BGE.
- fix incorrect assumption in the BGE that PySequence_Check() means PySequence_Fast_ functions can be used.
|
|
other selection operators. Instead of having only a single "extend"
option which causes the selection to toggle, there are now
three options:
- extend: adds to selection
- deselect: removes from selection
- toggle: toggles the selection
The default keymap is adjusted to have identical behavior as before,
using these new options, so end-users shouldn't experience any
difference.
|
|
available - or may not have UV data if in the wireframe view.
update the function to calculate the derived mesh with bmesh data-mask each time.
resolves bug [#33205] Bmesh "from_object" gives "no usable mesh data".
also found 'mesh_build_data' could run with non mesh objects and fail silently - add an assert.
|
|
Specifically the Rotational Difference and XYZ Rotation Transform Channel
Feature requested during BlenderPRO 2012 (Brazilian Blender Conference)
and coded during my flight back :)
Reviewed by Joshua Leung(Aligorith)
|
|
recent commits)
|
|
properties from operator, now it compares at RNA rather than ID property level,
which is more accurate.
|
|
style edits.
|
|
avoid per vert/edge/face string lookups in BMO_slot_map_* functions --- used in array modifier, subdivide, remove doubles and other tools.
|
|
the type of a PropertyGroup pointer would be fixed to the default RNA type. This is fine in most cases, but it may be necessary to define the property group subtype dynamically in some cases. Now the returned RNA pointer uses the typef callback if it exists.
|
|
|
|
This could break old files a little bit -- clip view could be offset/zoomed
in comparison with previous releases in pixel aspect != 1.
|
|
|
|
Seems ti was here since the very first day this functions were added
and issue was happening for switching sign for negative values.
Wrote a brute-force test locally and seems corrected functions indeed
works the same way as old macroses.
This should fix: #33226: File loading issue with svn 52328 (recent BF buildbot compile)
|
|
I see we explicitly set the edge visibility in the code for a few
modifiers because of this flag missing. The only place that this flag is
not set is during subsurf so maybe it would make a lot more sense to set
it by default on newly created edges (currently off by default) through
bmesh and turn it off for any modifiers that may need it to be off.
|
|
all edges are set to draw. Thanks to Campbel for pointing that out!
|
|
|
|
|
