Using ClangBuildAnalyzer on the whole Blender build, it was pointing
out that BLI_math.h is the heaviest "header hub" (i.e. non tiny file
that is included a lot).
However, there's very little (actually zero) source files in Blender
that need "all the math" (base, colors, vectors, matrices,
quaternions, intersection, interpolation, statistics, solvers and
time). A common use case is source files needing just vectors, or
just vectors & matrices, or just colors etc. Actually, 181 files
were including the whole math thing without needing it at all.
This change removes BLI_math.h completely, and instead in all the
places that need it, includes BLI_math_vector.h or BLI_math_color.h
and so on.
Change from that:
- BLI_math_color.h was included 1399 times -> now 408 (took 114.0sec
to parse -> now 36.3sec)
- BLI_simd.h 1403 -> 418 (109.7sec -> 34.9sec).
Full rebuild of Blender (Apple M1, Xcode, RelWithDebInfo) is not
affected much (342sec -> 334sec). Most of benefit would be when
someone's changing BLI_simd.h or BLI_math_color.h or similar files,
that now there's 3x fewer files result in a recompile.
Pull Request #110944
After some recent changes BLI_math_base got (indirectly) included
from DNA file, causing defines conflict in Cycles: Cycles wants the
default fast behavior of square root, and BLI color wants it to be
more preciese.
Proposed solution is to move the SSE block away from the math_base
closer to code which uses it. The initial intent was to make those
functions reusable, but for a long long time the color utilities
are the only users of those functions.
This change does not prevent the error from re-occurring in the
future if some code includes sse2neon and BLI color utilities, but
it makes such conflict situation much less likely to happen, for
now.
The downside of this change is that the code now need to include
BLI_simd.h explicitly to access BLI_HAVE_SSE2 instead of relying
on it being included indirectly with math headers. The mitigation
for this is to change semantic of the BLI_HAVE_SSE2: now it is
defined to 1 if SSE2 is supported and to 0 otherwise. This makes
it so the code needs to check if using `#if BLI_HAVE_SSE2` and
if the BLI_simd.h is not included it will generate warning when
using GCC or Clang.
This change in semantic is is something the current patches would
need to ensure is handled correctly.
Pull Request: https://projects.blender.org/blender/blender/pulls/109664
A lot of files were missing copyright field in the header and
the Blender Foundation contributed to them in a sense of bug
fixing and general maintenance.
This change makes it explicit that those files are at least
partially copyrighted by the Blender Foundation.
Note that this does not make it so the Blender Foundation is
the only holder of the copyright in those files, and developers
who do not have a signed contract with the foundation still
hold the copyright as well.
Another aspect of this change is using SPDX format for the
header. We already used it for the license specification,
and now we state it for the copyright as well, following the
FAQ:
https://reuse.software/faq/
Caused by 6769acbbba.
Some of the matrix variants are rather obscure from the
semantic: they don't really fully initialize the output:
as in, they only write to an upper-left 3x3 block.
A quick solution to fix the very commonly used constraint.
It feels that it is possible to minimize about of copy
operations.
Improve safety and correctness of matrix multiplication by using
temporary storage if one of the inputs is also the output.
No functional changes.
Differential Revision: https://developer.blender.org/D16876
Reviewed By: Campbell Barton, Sergey Sharybin
In some situations the zero axis can produce problems and need to add a small bias.
This function adds a small bias using the orthogonal result of the others valid axis.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D16158
6d
This patch implements the pixelate node for the realtime compositor.
Differential Revision: https://developer.blender.org/D15662
Reviewed By: Clement Foucault
Making the callers responsible for this isn't practical as matrices are
often passed indirectly to a functions such as mat3_to_axis_angle,
BKE_object_mat3_to_rot & BKE_pchan_mat3_to_rot.
Or the matrix is combined from other matrices which could be negative.
Given quaternions calculated from negative matrices are completely
invalid and checking only needs to negate matrices with a negative
determinant, move the check into mat3_to_quat and related functions.
Add mat3_normalized_to_quat_fast for cases no error checking on the
input matrix is needed such as blending rotations.
Use a more direct method of checking if a matrix is negative instead of
using cross & dot product.
Also replace some determinant_m3() < 0 checks with is_negative_m3.
Use a shorter/simpler license convention, stops the header taking so
much space.
Follow the SPDX license specification: https://spdx.org/licenses
- C/C++/objc/objc++
- Python
- Shell Scripts
- CMake, GNUmakefile
While most of the source tree has been included
- `./extern/` was left out.
- `./intern/cycles` & `./intern/atomic` are also excluded because they
use different header conventions.
doc/license/SPDX-license-identifiers.txt has been added to list SPDX all
used identifiers.
See P2788 for the script that automated these edits.
Reviewed By: brecht, mont29, sergey
Ref D14069
This is almost always meaningless as most code has changed
since the comment was added. Besides this, version control can be used
to check if/when a file was modified.
Some cases of this were kept when they contain details
about the original copyright holder.
Viewport cull bones during selection to avoid depth-picking
reading the depth buffer for bones that aren't in the viewport.
Files with thousands of bones could hang blender for seconds while
selecting. The issue could still happen with overlapping bones or when
zoomed out so all bones are under the cursor, however in practice this
rarely happens.
Now files with many bones select quickly.
Related changes include:
- Split `BKE_pchan_minmax` out of `BKE_pose_minmax`.
- Add `mat3_to_size_max_axis` to return the length of the largest
axis (used for scaling the radius).
Reviewed By: sybren
Maniphest Tasks: T91253
Ref D13990
Adding better support for drawing huge images in the image/uv editor. Also solved tearing artifacts.
The approach is that for each image/uv editor a screen space gpu texture is created that only contains
the visible pixels. When zooming or panning the gpu texture is rebuild.
Although the solution isn't memory intensive other parts of blender memory usage scales together with
the image size.
* Due to complexity we didn't implement partial updates when drawing images tiled (wrap repeat).
This could be added, but is complicated as a change in the source could mean many different
changes on the GPU texture. The work around for now is to tag all gpu textures to be dirty when
changes are detected.
Original plan was to have 4 screen space images to support panning without gpu texture creation.
For now we don't see the need to implement it as the solution is already fast. Especially when
GPU memory is shared with CPU ram.
Reviewed By: fclem
Maniphest Tasks: T92525, T92903
Differential Revision: https://developer.blender.org/D13424
- Added space below non doc-string comments to make it clear
these aren't comments for the symbols directly below them.
- Use doxy sections for some headers.
- Minor improvements to doc-strings.
Ref T92709
Separate the huge test into huge logical parts and add more cases
to check. Also add a utility to check that the matrix is orthogonal,
with arbitrary epsilon values and calculations in double.
A couple of tests deliberately fail, to be fixed in following commits.
Ref D9551
Commit e1665c3d31 added math to do 2D affine transformations with 3x3
matrices, but these matrices are also used for 3D transformations.
Remove added functions and use 4x4 matrices for 2D transformation.
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D12510
This constraint can be naturally viewed as a prototype for a future
4x4 matrix math node (or subset thereof), since its basic semantics
already is matrix assignment. Thus it makes sense to add math options
to this constraint to increase flexibility in the meantime.
This patch adds support for several operations that would be useful:
- An option to remove shear in the incoming target matrix.
Shear is known to cause issues for various mathematical operations,
so an option to remove it at key points is useful.
Constraints based on Euler like Copy Rotation and Limit Rotation
already have always enabled shear removal built in, because their
math doesn't work correctly with shear.
In the future node system shear removal would be a separate node
(and currently Limit Rotation can be used as a Remove Shear constraint).
However removing shear from the result of the target space conversion
before mixing (similar to Copy Rotation) has to be built into
Copy Transforms itself as an option.
- More ways to combine the target and owner matrices.
Similar to multiple Inherit Scale modes for parenting, there are
multiple ways one may want to combine matrices based on context.
This implements 3 variants for each of the Before/After modes
(one of them already existing).
- Full implements regular matrix multiplication as the most basic
option. The downside is the risk of creating shear.
- Aligned emulates the 'anti-shear' Aligned Inherit Scale mode,
and basically uses Full for location, and Split for rotation/scale.
(This choice already existed.)
- Split Channels combines location, rotation and scale separately.
Looking at D7547 there is demand for Split Channels in some cases,
so I think it makes sense to include it in Copy Transforms too, so that
the Mix menu items can be identical for it and the Action constraint.
Differential Revision: https://developer.blender.org/D9469
Expose a this function to initialize any zeroed axes
to an orthogonal vector based on other non-zeroed axes.
This functionality already existed for `invert_m#_m#_safe_ortho`,
expose as a public function as it's useful to be able to fill in zeroed
axes of transformation matrices since they may be used in matrix
multiplication which would create degenerate matrices.
This commit refactors the point distribute node to skip realizing
the instances created by the point instance node or the collection
and object info nodes. Realizing instances is not necessary here
because it copies all the mesh data and and interpolates all
attributes from the instances when this operation does not
need to modify the input geometry at all.
In the tree leaves test file this patch improves the performance of
the node by about 14%. That's not very much, the gain is likely larger
for more complicated input instances with more attributes (especially
attributes on different domains, where interpolation would be necessary
to join all of the instances). Another possible performance improvement
would be to parallelize the code in this node where possible.
The point distribution code unfortunately gets quite a bit more
complicated because it has to handle the complexity of having many
inputs instead of just one.
Note that this commit changes the randomness of the distribution
in some cases, as if the seed input had changed.
Differential Revision: https://developer.blender.org/D10596
This patch changes behavior of strip transform and crop feature.
Purpose of this change is to allow display arbitrary portion of input
image, simplify user interface and workflow.
Offset and Crop values in old files are converted in versioning.
Offset animation is also converted. Crop animation and animation of
crop or offset enable properties is not taken into account
Changes in behavior and interface:
- If image is added to timeline it is scaled to fit inside preview area
while maintaining aspect ratio. Image is centered. This is considered
as a baseline for further transformation.
- Scale and rotation was added, so it is possible to transform image at
it's original resolution.
- Crop will not affect image transformation (does not move image).
- Values of Crop and Transform Position are in pixels, these values are
corrected if preview is fraction of project resolution.
- Transform and Mirror panel has been removed and new Transform panel
and Crop panel is moved to Adjust panel. Mirror is now part of new
Transform panel.
Technical changes:
- Preprocessing stage must work on duplicated image, because original is
cached. Previously Crop and Offset could run at once and required only
one duplication of image. This is not the case with new algorithms, so
duplication on demand is implemented. Transformation can read original
image and will output new image that is safe to modify. It should be
possible to add crop step to transform algorithm, so that Crop won't
require previous duplication though.
- Use Crop and Use Translation checkboxes were removed. Individual
values are compared to default values to check if image needs to be
processed. In case of transform this will be done also if resolution of
source.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8393
This patch changes behavior of strip transform and crop feature.
Purpose of this change is to allow display arbitrary portion of input
image, simplify user interface and workflow.
Offset and Crop values in old files are converted in versioning.
Offset animation is also converted. Crop animation and animation of
crop or offset enable properties is not taken into account
Changes in behavior and interface:
- If image is added to timeline it is scaled to fit inside preview area
while maintaining aspect ratio. Image is centered. This is considered
as a baseline for further transformation.
- Scale and rotation was added, so it is possible to transform image at
it's original resolution.
- Crop will not affect image transformation (does not move image).
- Values of Crop and Transform Position are in pixels, these values are
corrected if preview is fraction of project resolution.
- Transform and Mirror panel has been removed and new Transform panel
and Crop panel is moved to Adjust panel. Mirror is now part of new
Transform panel.
Technical changes:
- Preprocessing stage must work on duplicated image, because original is
cached. Previously Crop and Offset could run at once and required only
one duplication of image. This is not the case with new algorithms, so
duplication on demand is implemented. Transformation can read original
image and will output new image that is safe to modify. It should be
possible to add crop step to transform algorithm, so that Crop won't
require previous duplication though.
- Use Crop and Use Translation checkboxes were removed. Individual
values are compared to default values to check if image needs to be
processed. In case of transform this will be done also if resolution of
source.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8393
We have our own assert implementation, `BLI_assert()` that is prefered over the
C standard library one. Its output is more consistent across compilers and
makes termination on assert failure optional (through `WITH_ASSERT_ABORT`).
In many places we'd include the C library header without ever accessing it.
Unlike invert_m4_m4_safe, this calculates zeroed axes.
Useful when we need to use the inverse of an objects matrix,
keeping the valid axis, only filling in the zeroed ones.
The matrix interpolation function `interp_m3_m3m3()` decomposes the
matrices into rotation and scale matrices, converts the rotation
matrices to quaternions, SLERPs the quaternions, and converts the result
back to a matrix. Since quaternions cannot represent axis flips, this
results in interpolation problems like described in T77154.
Our interpolation function is based on "Matrix Animation and Polar
Decomposition", by Ken Shoemake & Tom Duff. The paper states that it
produces invalid results when there is an axis flip in the rotation
matrix (or negative determinant, or negative scale, those all indicate
the same thing). Their solution is to multiply the rotation matrix with
`-I`, where `I` is the identity matrix. This is the same as element-wise
multiplication with `-1.0f`. My proposed solution is to not only do that
with the rotation matrix `R`, but also with the scale matrix `S`. This
ensures that the decomposition of `A = R * S` remains valid, while also
making it possible to conver the rotation component to a quaternion.
There is still an issue when interpolating between matrices with
different determinant. As the determinant represents the change in
volume when that matrix is applied to an object, interpolating between a
negative and a positive matrix will have to go through a zero
determinant. In this case the volume collapses to zero. I don't see this
as a big issue, though, as without this patch Blender would also produce
invalid results anyway.
Reviewed By: brecht, sergey
Differential Revision: https://developer.blender.org/D8048
