* Support using devices from all OpenCL platforms, so that you can use e.g. both
Intel and NVidia OpenCL implementations if you have them installed.
* Fix compile error due to missing fmodf after recent math node change.
* Enable advanced shading for Intel OpenCL.
* CYCLES_OPENCL_DEBUG environment variable for generating debug symbols so you
can debug with gdb. This crashes the compiler with Intel OpenCL on Linux though.
To make this work the preprocessed kernel source code is written out, as gdb
needs this.
* Show OpenCL compiler warnings even if the build succeeded.
* Some small fixes to initialize cdDevice to NULL, add missing NULL check when
creating buffer and add missing space at end of build options for Apple OpenCL.
* Fix crash with multi device + opencl, now e.g. CPU + GPU render should work.
I did a few tweaks to the code and also:
* Fix viewport render failing sometimes with Apple CPU OpenCL, was not taking
workgroup size limits into account properly.
* Add compile error when advanced shading in the Blender binary and OpenCL kernel
are not in sync.
* Some closures (Toon, Diffuse Ramp) were not assigned to a CLOSURE_IS_* define, which made them invisible on render passes.
* Westin closures had wrong type, Sheen is Diffuse, Backscatter is Glossy.
* Rename fresnel_dielectric() to fresnel_dielectric_cos() to match SVM, easier when searching code.
* Also remove an old code comment in bsdf_reflection.h from Cycles branch days.
- Removed grid-snapping for area coordinates on scaling windows.
That caused the areas to shrink or expand, and eventually corrupt screen layouts.
- Added simple but efficient life resize for OSX. I need to know why this is so much
code for Windows... I suggest Windows to just copy same method; dispatch the queue,
and just let the event system draw.
This was caused by a "hack" Daniel Genrich introduced in his moving obstacles commit in r46050. I suppose it was originally added to prevent issues with too fast moving obstacles, but now it ended up limiting maximum velocity of higher resolution simulations.
Here is an comparision of 184 resolution simulation (simulation area limited by adaptive domain):
https://www.miikah.org/blender/smoke_with_pressure_limit_hack.pnghttps://www.miikah.org/blender/smoke_without_pressure_limit_hack.png
I now reverted that hack until a better solution is found. Daniel, can you check this out? Pressure was limited to maximum of dt * dx (= dt / res) which doesn't make sense to limit pressure based on grid resolution. Maybe better to limit with a constant factor instead?
This caused high resolution smoke to always regenerate new tile when domain was reinitialized, slowing down especially adaptive domain simulations. Now noise tile is saved in Blender temp directory instead.
* Added Westin Sheen and Westin Backscatter closures for testing, useful for Cloth like effects.
Only available via OSL, added an example OSL shader to the Templates (Text Editor).
* Also do pressure interpolation for brush size and spacing.
* Do smoothing of pressure when smooth stroke and sample average is enabled.
* Revert the OS X specific pressure change to pressure ^ 2.5, for low pressure
values like 0.05 it makes the pressure 100x lower, which is problematic. If
we need to adjust the pressure curve it should be done for all platforms.
Still weak:
* Pressure of first touch on tablet is difficult to control, usually it's low
which makes the stroke start out small or soft, but other times not. Finer
event capturing at ghost level would help, along with pressure changes without
mouse movement, but this may also need different paint stroke logic.
* Brush radius is rounded to integers, this gives noticeable stepping.
* Brush falloff is not antialiased, gives noticeable aliasing for small brush
sizes which was always a problem, but is more common with size pressure control.
So now, in the new "other" tex context, you can (depending on active data) have direct access to modifiers', force's or brushes' textures...
I also refactored a bit how texture contexts are handled (once again, we had some quite similar code in both space_buttons and RNA sources). This should also solve some harmless glitches like "no texture context selected in UI" sometimes when you remove data related to current texture (see e.g. after removing the material from default cube, in startup scene).
This usage of two different systems for textures, and the handling of switches between them, has been a bit tricky to get working right, but it is OK now I think. I also had to add a bool flag to buttons space, SB_TEX_USER_LIMITED (use_limited_texture_context in RNA), which indicates "new shading" texture code whether it has to ignore materials, lamps etc. (BI) or not (Cycles).
Btw, pinned textures from modifiers/force/etc. were also broken (showing nothing), now it should work too.
Thanks to Brecht for reviewing.
Another issue with the recent Ghost changes here. For some reason key up events
are not coming through when the command key is pressed. I can't figure out why,
for now just always handle them, still fixes the original bug.
interact better with system shortcuts.
This is a special shortcut for switching between views and does not get
delivered directly to our view when we pass it through the application key
event handling path. We only have a single OpenGL view, so there's no need to
pass it on to the application, instead just interpret it directly.
insert text in the text editor and do the associated operation like minimizing the
window or switching windows.
The code was always doing both without trying to ensure only one is done. Now we
integrate a bit better with the event handling and pass the event to NSApp, which
then decides to handle the event itself or pass it on to the window, from where
we then send it back to be handled.
Problem was that due to group proxy node the anisotropic node did not detect
early enough that it needs generated texture coordinate data to generate the
tangent. Now the proxy nodes are removed earlier.
for Apple OpenCL on OS X 10.8 and simple AO render.
Also environment variable CYCLES_OPENCL_TEST can now be set to CPU, GPU,
ACCELERATOR, DEFAULT or ALL values to test particuler devices.
give results that were either too weak or too strong, this makes it give more
predictable results. The downside is that it breaks backwards compatibility but
the previous behavior was almost broken.
* Code refactor of tile ordering to simplify the code and avoid some branching.
* Changed the Center method, so it really follows center -> corners, instead of the BI method, which was confusing sometimes.
the Bump node has a Normal input, so you can chain it after a Normal Map node.
Note that normal mapping always has to be done first because it is tied to the
particular mesh surface and tangents.
fullscreen option. It was possible to enable both at the same time which got
you stuck in a state where it was impossible to exit fullscreen. Now I've made
them mutually exlusive, only one can be enabled at the same time.
Note the reason we need to support both is because the new Lion fullscreen does
not work with multiple monitors, it will just give black screens on the other
monitors. This is a limitation of OS X, you can find many complaints about this
online.
after a bugfix for precision issues with low roughness. Now it renders them as
perfectly sharp which avoids the problematic calculations rather than increasing
the roughness.
Fixed by adding a "texture_user_property" member to spacebuts' context, and using it to get the prop identifier in ui script.
Thanks to Brecht for its advice!
The problem was (again) the x86 extended precision float register being used for
one float value while the other was rounded to lower precision. This caused the
strictly weak order requirement for std::sort to be broken.
a maximized Blender window in Ubuntu Unity. The window size would slightly change
as part of the unminimize effect.
Now cycles waits 0.2 seconds before restarting rendering after a viewport resize,
also a bit less flickery when changing the 3d view size in Blender itself.
and preview running at the same time.
It seems there's something in OSL/LLVM that's not thread safe, but I couldn't
figure out what exactly. Now all renders share the same OSL ShadingSystem which
should avoid the problem.
This is a experimental collision detection function, so the API might
change in the future.
Note: The simulation needs to be stepped before this function can be
used, otherwise the rigid body world might not be valid.
Patch [#34989] Bullet Convex sweep test API
by Vilem Novak (pildanovak), thanks!
After planar codecs support minimal FFmpeg was
bumped to 0.10 which was not so much nice because
it was only released only later last year.
Didn't find a way to make compatibility code local
in ffmpeg_compat, so there're some ifdefs in
audaspace and writeffmpeg.
Not entirely happy, but having a bit of ifdefs in
code better than lots of real PITA for platform
maintainers.
Now there is a single BVH traversal code with #ifdefs for various features.
At runtime it will then select the appropriate variation to use depending if
instancing, hair or motion blur is in use.
This makes scenes without hair render a bit faster, especially after the
minimum width feature was added. It's not the most beautiful code, but we can't
use c++ templates and there were already 4 copies, adding 4 more to handle the
hair case separately would be too much.
panel now has an option to specify how to use them. There's three options:
* Use: render layer samples override scene samples
* Bounded: bound render layer samples by scene samples
* Ignore: ignore render layer sample settings
Code is added to restrict the pixel size of strands in cycles. It works best with ribbon primitives and a preset for these is included. It uses distance dependent expansion of the strands and then stochastic strand removal to give a fading. To prevent a slowdown for triangle mesh objects in the BVH an extra visibility flag has been added. It is also only applied for camera rays.
The strand width settings are also changed, so that the particle size is not included in the width calculation. Instead there is a separate particle system parameter for width scaling.
AVCODEC_MAX_AUDIO_FRAME_SIZE was deprecated and
finally removed from current trunk.
Initial patch by Lawrence D'Oliveiro (ldo) with
own modification, Thanks!
The latest ffmpeg versions include a workaround to deal with a certain
pecularity in Canon DSLR footage: instead of decoding pictures with the
proper resolution of 1920x1080 they decode it with 1920x1088 and add a
black bar at the bottom.
Needless to say, that this screws up things in a lot of areas within blender
(proxy indices, mask animations etc.)
Since all blender versions besides Linux x86 32bit seem still to include
older ffmpeg versions which still contain this bug, this patch adds
a workaround for older versions until we have all versions on all platforms
up to date.
See also: http://git.libav.org/?p=libav.git;a=commit;h=30f515091c323da59c0f1b533703dedca2f4b95d
Custom implementation for resizing (GHOST_SizerWin32)
Some things still don't work:
* esc cancel
* max windows size
* aero (sizing) snap on win7
hbrBackground = 0 to disable clear screen.
Thanks to dfelinto for help in finding of root cause.
low roughness and same index of refraction.
Problem was bad float precision due to low roughness, which caused the pdf for
the different closures to not match properly.
* Cycles Render layers UI was broken after freestyle merge (changes were not merged). Did manual edits now with some tweaks.
* Some layout fixes for Mask Layer.
per render layer samples in addition to the progress bar.
Also fixed job progress bar not working at all on high DPI / retina, was so small
the actual progress was not visible.
Issue was caused by some mesa drivers does not support GL_RGBA16F.
Now added check around glTexImage2D to verify whether requested
internal format is actually supported. If not blender will fall
back to non-GLSL image display.
- GLSL shader wasn't aware of alpha predivide option,
always assuming alpha is straight. Gave wrong results
when displaying transparent float buffers.
- GLSL display wasn't aware of float buffers with number
of channels different from 4, crashing when trying to
display image with different number of channels.
This required a bit larger changes, namely now it's
possible to pass format (GL_RGB, GL_RGBAm GL_LUMINANCE)
to glaDrawPixelsTex, This also implied adding format to
glaDrawPixelsAuto and modifying all places where this
functions are called.
Now GLSL will handle both 3 and 4 channels buffers,
single channel images are handled by CPU.
- Replaced hack for render result displaying with a bit
different hack.
Namely CPU conversion will happen only during render,
once render is done GLSL would be used for displaying
render result on a screen.
This is so because of the way renderer updates parts
of the image -- it happens without respect to active
render layer in image user. This is harmless because
only display buffer is modifying, but this is tricky
because we don't have original buffer opened during
rendering.
One more related fix here was about when rendering
multiple layers, wrong image would be displaying when
rendering is done. Added a signal to invalidate
display buffer once rendering is done (only happens
when using multiple layers). This solves issue with
wrong buffer stuck on the display when using regular
CPU display space transform and if GLSL is available
it'll make image displayed with a GLSL shader.
- As an additional change, byte buffers now also uses
GLSL display transform.
So now only dutehr and RGB curves are stoppers for
using GLSL for all kind of display transforms.
well as I would like, but it works, just add a subsurface scattering node and
you can use it like any other BSDF.
It is using fully raytraced sampling compatible with progressive rendering
and other more advanced rendering algorithms we might used in the future, and
it uses no extra memory so it's suitable for complex scenes.
Disadvantage is that it can be quite noisy and slow. Two limitations that will
be solved are that it does not work with bump mapping yet, and that the falloff
function used is a simple cubic function, it's not using the real BSSRDF
falloff function yet.
The node has a color input, along with a scattering radius for each RGB color
channel along with an overall scale factor for the radii.
There is also no GPU support yet, will test if I can get that working later.
Node Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/Shaders#BSSRDF
Implementation notes:
http://wiki.blender.org/index.php/Dev:2.6/Source/Render/Cycles/Subsurface_Scattering
Apparently C== allowed to have implementations of
OCIOImpl::setupGLSLDraw, OCIOImpl::finishGLSLDraw and
OCIOImpl::freeGLState in two different files.
STUPID!
* Unnecessary shader inputs inside the Mix Shader are now ignored, in case the factor is 0.0 / 1.0 and not connected.
This way we save some render time for complex node graphs.
Example: http://www.pasteall.org/pic/show.php?id=48226
Check the Mix Shader at the end: In this case, Cycles will now avoid the complete huge shader tree, and only calculate the Diffuse Shader.
Rendertime decreased from 1:50 min to 1:20 min on CPU. GPU rendering benefits as well from this.
This only affects SVM, OSL was already doing these optimizations.
Implemented using GLSL API from OpenColorIO library and
some general functions were added to it's c-api:
- OCIO_setupGLSLDraw prepares OpenGL context for GPU-based
transformation for a giver processor.
This function compiles and links shader, sets up it's
argument. After this transformation would be applied
on an image displaying as a 2D texture.
So, glaDrawPixelsTex called after OCIO_setupGLSLDraw will
do a proper color space transform.
- OCIO_finishGLSLDraw restores OpenGL context after all
color-managed display is over.
- OCIO_freeOGLState frees allocated state structure used
for cacheing some GLSL-related stuff.
There're some utility functions in IMB_colormanagent which
are basically proxies to lower level OCIO functions but
which could be used from any place in blender.
Chacheing of movie clip frame on GPU is also removed now,
and either glaDrawPixelsTex or glaDrawPixelsAuto are used
for display now. This is so no code duplication happens
now and no large textures are lurking around in GPU memory.
Known issues:
- Texture buffer and GLSL are no longer checking for
video card capabilities, possibly could lead to some
artifacts on crappy drivers/cards.
- Only float buffers are displaying using GLSL, byte
buffers will still use fallback display method.
This is to be addressed later.
- If RGB curves are used as a part of display transform,
GLSL display will also be disabled. This is also thing
to be solved later.
Additional changes:
- glaDrawPixelsTexScaled will now use RGBA16F as an
internal format of storing textures when it's used
to draw float buffer. This is needed so LUT are
applied without precision loss.
- pass string size to BLI_timestr() to avoid possible buffer overrun.
- quiet warning for mingw.
- include guards for windows utf conversion funcs.
- fix for mistage in edge-angle-selection check.
- some style cleanup.
Added new build option WITH_JACK_DYNLOAD for CMake and
WITH_BF_JACK_DYNLOAD for SCons, which means there'll be
no build-time linking against libjack and getting symbols
from libjack will happen runtime using dlopen and dlsym
tricks.
Alternative would be to use weak linking, but it'll require
having wrapper for preloading libjack.
This new options are disabled by default and they only
intended to be used on linux. Other platforms shall not
be using this and there shall be no functional changes
on non-linux platforms at all.
Mac OS X full screen: the old option to go full screen now didn't hide the
dock/topbar anymore. Also made it use dock auto-hide now, not permanent hide.
* Move OpenGL settings out of the film panel into its own.
Imho these should go completely elsewhere, but better separated than mixed with Cycles settings.
Blender now supports the 10.7+ "Full screen" mode, which pushes a window to a permanent
other "screen", with animated zoom and sliding.
Available via the icon in window header (right), Apple+F, or "Window" menu in top.
Works much nicer than Blender's own "full screen" option.
Todo: the zoom effect is still draws a bit ugly, because Blender doesn't have "live resize"
yet.
This commit basically implements frames prefetching for
movie clip datablock.
Number of frames to be prefetched is controlled in User
Preferences, System tab, Prefetch Frames option.
Currently prefetching is destructive-less for movie cache,
meaning mo frames will be removed from the cache when while
prefetching. This is because it's half of simplier to
implement, but it also makes sense from tracking point of
view -- we could want to playback in both directions and
removing frames from behind time cursor is not always a
good idea.
Anyway, smarter prefetching strategy could be developed
later.
Some implementation notes:
- Added MEM_CacheLimiter_get_memory_in_use function to get
memory usage of specified memory limiter.
- Fixed prototype of MEM_CacheLimiter_get_maximum which
was simply wrong (used wrong data type for output).
- Added some utility functions to movie clip and movie
cache for direct cache interaction and obtaining cache
statistics.
- Prefetching is implemented using general jobs system.
which is invoking from clip draw function.
- Prefetcing will stop as soon other job or playback starts.
This is done from performance point of view. Jobs will
likely require lots of CPU power and better to provide
whole CPU to it.
Playback is a bit more complicated case. For jpeg sequence
playback prefetching while paying back is nice. But trying
to prefetch heavy exr images and doing color space
conversion slows down both playback and prefetching.
TODO:
- Think of better policy of dealing with already cached frames
(like when cached frames from other clips prevents frames
from current clip to be prefetched)
- Currently a bit funky redraw notification happens from
prefetch job. Perhaps own ND_ is better to have here.
- Hiding clip while prefetch is active in theory shall stop
prefetching job.
- Having multiple clips opened on file load will prefetch
frames for only one of them.
The issue here was that the proxy nodes created for connecting extern group node sockets to the internal nodes were generated by the input/output nodes themselves.
0 input/output nodes: there would be no proxy that external group node sockets can map to
2+ input/output nodes: additional nodes would overwrite entries from previous nodes, so that only one of the input/output nodes would be used.
Solution is to always generate exactly 1 proxy node for every group socket in advance, regardless of whether it is used internally. Internal node sockets can then all map to this proxy node.
In the case out output nodes there should only ever be one active node, otherwise the connection to the proxy would be ambiguous. For this purpose the NODE_DO_OUTPUT flag has been exposed to RNA, so that cycles can check it and only use the active output.
PyNodes opens up the node system in Blender to scripters and adds a number of UI-level improvements.
=== Dynamic node type registration ===
Node types can now be added at runtime, using the RNA registration mechanism from python. This enables addons such as render engines to create a complete user interface with nodes.
Examples of how such nodes can be defined can be found in my personal wiki docs atm [1] and as a script template in release/scripts/templates_py/custom_nodes.py [2].
=== Node group improvements ===
Each node editor now has a tree history of edited node groups, which allows opening and editing nested node groups. The node editor also supports pinning now, so that different spaces can be used to edit different node groups simultaneously. For more ramblings and rationale see (really old) blog post on code.blender.org [3].
The interface of node groups has been overhauled. Sockets of a node group are no longer displayed in columns on either side, but instead special input/output nodes are used to mirror group sockets inside a node tree. This solves the problem of long node lines in groups and allows more adaptable node layout. Internal sockets can be exposed from a group by either connecting to the extension sockets in input/output nodes (shown as empty circle) or by adding sockets from the node property bar in the "Interface" panel. Further details such as the socket name can also be changed there.
[1] http://wiki.blender.org/index.php/User:Phonybone/Python_Nodes
[2] http://projects.blender.org/scm/viewvc.php/trunk/blender/release/scripts/templates_py/custom_nodes.py?view=markup&root=bf-blender
[3] http://code.blender.org/index.php/2012/01/improving-node-group-interface-editing/