Previously point cloud rendering was disabled on the HIPRT backend due
to unexpected performance regressions introduce by it.
With the recent update to HIP SDK 6.3 and HIPRT 2.5, these performance
regressions have been resolved and so this commit re-enables
point cloud rendering on HIPRT.
Pull Request: https://projects.blender.org/blender/blender/pulls/134902
This helps prevent long text from wrapping but still looks nice in my opinion.
In the future we might try to hide the base name part of the class but I think this would require a fix in sphinx.
This patch registers scrollbars with handles so that they shrink just
like their non-handle counterparts.
To avoid user errors related to clicking into the handle track instead
of the underlying view, if the final alpha of the track is invisible (0)
then it will range up to 0.25 instead.
Pull Request: https://projects.blender.org/blender/blender/pulls/135023
Our text wrapping code is written in a way that requires the Line Feed
character to be in our glyph cache. This is problematic as none of the
control characters will ever be there if not using fallback. With
fallback these glyphs could be a zero-width character from a symbol
font, or something with width if using a last resort font. So we'd have
to ignore the advance in the latter case. This came to a head recently
because we are turning off fallback in some circumstances. 12b1f8bd7a
removed the rendering of control characters so they are never in the
cache. That change requires this one, which does wrapping based on the
string's codepoint, not that of its glyph.
Pull Request: https://projects.blender.org/blender/blender/pulls/135196
Dropping the inlining hint for `light_tree_pdf` and reverting to the
default inlining thresholds for DPC++ compiler gives a ~4% speedup on
classroom and other scenes on Arc B580.
Pull Request: https://projects.blender.org/blender/blender/pulls/135042
This adds a new `Visual Geometry to Objects` operator which is added to the
`Apply` menu.
It creates new objects from the evaluated geometry of the active object. Other
than e.g. applying modifiers, this operator does keep all generated data and
creates new objects and collections for instances. It does have some
similarities to `Make Instances Real` with the main difference that it doesn't
actually realize instances, i.e. instanced geometry is still shared between all
objects using it.
It does intentionally not replace the original object, because the semantics of
that aren't all that obvious. It's especially tricky when the object was
referenced by other objects. Instance attributes are not preserved currently.
Pull Request: https://projects.blender.org/blender/blender/pulls/134119
Avoid non valid glTF file by avoiding exporting multiple slots with same target_id_type in the same animation for a single object:
- Make sure to create multiple animation when needed
- Use action.name for active slots (can have multiple slots when different target_id_type)
- Use action.name - slot.name_display for additional animations
The RE_render_result_rect_from_ibuf function (used only by VSE) was
allocating memory for the render result, clearing it to zero and
immediately overwriting it with the image pixel data. Remove the
"clear to zero" part.
Rendering a 1920x1080 resolution movie with VSE (using float
pixel data), average RE_render_result_rect_from_ibuf time goes from
6.8ms down to 3.8ms
Pull Request: https://projects.blender.org/blender/blender/pulls/135170
Same was true for SimpleDeform.
Invert vertex group influence should only have an effect if there is a
group specified.
Shrinkwrap always uses a weight (gets weights via
`BKE_defvert_array_find_weight_safe` -- which gets full weight for no
group specified, rightfully so), if this gets inverted, we get no
influence at all.
Now move the inverting of the weights to `BKE_defvert_array_find_weight_safe`,
adjust the other callers accordingly (Bevel also does not need to check for a valid
vertexgroup beforehand).
Pull Request: https://projects.blender.org/blender/blender/pulls/121625
This implements three improvements to the energy preservation and albedo
scaling logic, which help the Principled BSDF pass the white-furnace test
when using the coat layers at high roughness.
Specifically, at roughness 0.3, the albedo scaling brings it from 60% at
the edge to 95%, and with the energy preservation it's 99.8%.
Pull Request: https://projects.blender.org/blender/blender/pulls/134620
Since the transform system uses previous positions to update to
"original" positions upon cancel, the elastic transforms are not working
well with this (due to their `TransformDisplacementMode::Incremental`
nature).
To resolve, rely on restoring positions from undo (as done elsewhere in
sculpt).
Pull Request: https://projects.blender.org/blender/blender/pulls/134919
`BKE_modifier_copy_ex` did it (which is called e.g. when copying the
whole stack from one object to another), but when simply duplicating a
modifier (`modifier_copy`), it was just ensuring a unique name (based on
the default name) without copyig the existing name prior.
Pull Request: https://projects.blender.org/blender/blender/pulls/135172
When called on an object that you cannot get a mesh from (e.g. Empties),
you would run into an unhelpful "SystemError: <built-in method
FromObject of type object at ...> returned NULL without setting an
exception"
Now be more specific in the error message.
Pull Request: https://projects.blender.org/blender/blender/pulls/135162
This patch refactors GPU shaders to remove includes to the utility
gpu_shader_common_math.glsl file. This is done because it has duplicate
functions that exist in other files, and it was really created for use
in GPU material nodes.
The safe_divide and hypot functions were removed since they exist in
gpu_shader_math_base_lib.glsl.
The compatible_[mod|pow] and wrap functions were moved into
gpu_shader_math_base_lib.glsl.
The floor_to_int function was inlined since it was trivial and only used
in one place.
The quick_floor was removed because it was unused.
The euler_to_mat3 function was replaced with the from_rotation function
from gpu_shader_math_matrix_lib.glsl.
Now the file only contains some GPU material node utility functions.
Pull Request: https://projects.blender.org/blender/blender/pulls/135160
Looks like "divers" comes from ancient times, Dutch word meaning "misc".
But by now, everything in that file is about conversion between different
pixel data types.
Pull Request: https://projects.blender.org/blender/blender/pulls/135165
There's no point in having non-threaded image color space conversion functions.
So merge the threaded and non-threaded functions and clarify names while at it:
- IMB_colormanagement_transform & IMB_colormanagement_transform_threaded
-> IMB_colormanagement_transform_float
- IMB_colormanagement_transform_byte & IMB_colormanagement_transform_byte_threaded
-> IMB_colormanagement_transform_byte
- IMB_colormanagement_transform_from_byte & IMB_colormanagement_transform_from_byte_threaded
-> IMB_colormanagement_transform_byte_to_float
These places were doing single-threaded colorspace conversion previously, and
thus now are potentially faster:
- IMB_rect_from_float (used in many places)
- EXR image "save as render" saving (image_exr_from_scene_linear_to_output)
- Object baking (write_internal_bake_pixels, write_external_bake_pixels)
- General image saving, clipboard copy, movie preparation
(IMB_colormanagement_imbuf_for_write)
- Linear conversion when reading HDR images/movies
(colormanage_imbuf_make_linear)
- EXR multi-layer conversion (render_result_new_from_exr)
For one case I benchmarked, which is to render out a 2D stabilized 10 bit input
movie clip out of VSE, the total render time went from 49sec down to 44sec
(Ryzen 5950X), one of the single-threaded parts was the colorspace conversion
in the movieclip code.
Pull Request: https://projects.blender.org/blender/blender/pulls/135155
In this case, the evaluator cache was never referenced and
the subdiv free queue empty. So the freeing of the cache
never happened.
This function is called once per frame and is unlikely
to generate overhead by doing one lock.
Speedup IMB_rotate_orthogonal (used for example in auto-rotating
videos that were shot sideways on a phone) by: 1) not copying previous
pixel values into new result, only for them to be immediately
overwritten by rotated pixels, and 2) using multi-threading.
Performing rotation of 1920x1080 resolution HDR (float) video frame
goes from 20ms down to 5ms (Ryzen 5950X, Windows)
Pull Request: https://projects.blender.org/blender/blender/pulls/135158
This patches removes common_math_utils includes from compositor shaders
and replaces them with math lib includes. This involves moving some
functions from that file to to the math lib files.
Pull Request: https://projects.blender.org/blender/blender/pulls/135157
Only a single `select` prop was exposed, and this was set to "bezt.f1"
(so it affected the left handle only).
Now we add properties for `left`, `right` and the `single` handle, the
selection prop for the control point is renamed to
`select_control_point` [which is now in line with the API for a bezier
curve].
NOTE: setting/getting a `single` handle uses the existing functions for
this -- handle is selected if any of the right/left handle is selected,
setting the handle results in both handles getting selected.
Pull Request: https://projects.blender.org/blender/blender/pulls/135114
This happened for context attributes if already "complete" attributes
were typed/pasted in.
Reason is that this copies functionality from our intellisense
codecompletion where this is expected behavior, for context attributes
we cannot really use "tab" to expand, so it is less useful here.
When confirming (with the additional "."), the keymap entry would be
broken, so better solve in a way that we dont end up with those "." by
removing it.
NOTE: there might be ways to enhance the string-search behavior here (by
properly supporting "tab-expanding"), but this is a bit out-of-scope for
this bugfix.
Pull Request: https://projects.blender.org/blender/blender/pulls/135098
