- In "Surface" mode the depth buffer needs to be cached after constructing the base DrawingPlacement, otherwise there is no depth data for projection and the origin is used as a fallback.
- The DrawingPlacement relies on the RegionView3D matrices to project from the rendered image back into 3D space. Because the boundary projection is called while the rv3d matrices are still set up for the image render it will look up the wrong pixels from the depth buffer. The solution is to make sure the viewport reset happens before projecting the boundary, i.e. by making sure the reset function is called before `process_image`.
- The aspect ratio needs to be taken into account for the boundary-to-strokes transform, otherwise the placement is wrong depending on how much the region aspect deviates from square.
The code for computing the zoom factor and offset has also been cleaned up and documented somewhat better, and now uses the `Bounds<float2>` struct instead of old `rctf`.
Pull Request: https://projects.blender.org/blender/blender/pulls/131321
Inside sequencer_main_cursor, the logic for showing a custom cursor
was already ignoring the time scrub ruler area, and not showing
the strip edge resizing cursors. However confusingly enough, actually
clicking there was still going into the
ED_sequencer_pick_strip_and_handle which found a strip edge and
returned it. Modify ED_sequencer_pick_strip_and_handle to return empty
selection when mouse is inside the time scrub ruler.
Pull Request: https://projects.blender.org/blender/blender/pulls/132891
Depth buffer caching was broken in 62897317bd. The `use_basic` parameter was removed but `use_gpencil`
should remain unmodified. This caused overlays to be included in the depth buffer for Grease Pencil.
Pull Request: https://projects.blender.org/blender/blender/pulls/132871
The `copy_layer_group_content` function was using the `LayerGroup::nodes()` API
for iterating over layers and inserting unique copies into the target geometry.
This function lists __all__ the layers inside a group, not just direct children.
A layer inside a group will be copied twice or more this way, once for the root
layer and for any of its parent groups. Looping only over directly child layers
and groups is the correct thing to do here.
Pull Request: https://projects.blender.org/blender/blender/pulls/132868
There are a few simple shader nodes that were missing test files.
This commit adds tests for those nodes:
- Camera data
- Fresnel
- Layer weight
- Hue/Saturation/Value
- Invert
- Normal
- Wavelength
Ref: blender/blender-test-data!39
Multi-column lists are supposed to collapse to single-column when they
are estimated to be wider than the available window width. However,
current code has two errors. First the calculation of the maximum
number of possible rows is based on entire screen height when the most
available is a bit less than half. This too-large value was used in
the calculation of widths, yet the maximum rows is actually clamped
to about 24 for uncategorized lists. Whenever the former is greater
than the latter this could cause collapsing to not occur. This issue is
only really noticeable for lists in Nodes as these can be shown much
smaller than regular size. This PR fixes the max-row calculation and
uses this correctly when breaking lists without categories.
Pull Request: https://projects.blender.org/blender/blender/pulls/132881
In a Camera's properties you can "Reset to Default" on "Focal Length"
because "lens" is a direct member of Camera and it has "50.0f" as its
default value in DNA_camera_defaults.h. However change the "Len Unit"
to "Field of View" and what you are changing is "angle". This is not
a member of Camera but is a calculated value based on sensor size. So
"reset to default" makes this "0", which is invalid and otherwise not
possible to enter. This PR uses RNA_def_property_float_default to set
a default value of 0.6911504f radians which is the 39.6° FOV for the
default 50mm lens.
Pull Request: https://projects.blender.org/blender/blender/pulls/132364
For node region, `event_cursor` is true, it sends update to the cursor
every draw call, this restores the cursor `WM_CURSOR_X_MOVE`.
Issue is originally caused by e6d941cdf4 and it introduced couple other
issues. So best fix would be to revert the commit. And to fix#129178,
restore modal cursor after drawing context menu, i.e. execution of
`ui_popup_context_menu_for_button`.
Pull Request: https://projects.blender.org/blender/blender/pulls/132848
Unfortunately, the tests as they are right now will likely not work with
4.2 and previous versions, as the brush asset changes will need to be
accounted for.
Pull Request: https://projects.blender.org/blender/blender/pulls/132827
This creates more boilerplate code, but it's easier to extend,
it's follows the same structure as the mesh extraction, and
it's more aligned to planned future performance improvements.
Pull Request: https://projects.blender.org/blender/blender/pulls/132866
Was missed in the move to Overlay Next.
Now bring it back in `class Axes: Overlay`
Thx @pragma37 making me aware chaging draw state (not writing depth)
can be done using a subpass!
Pull Request: https://projects.blender.org/blender/blender/pulls/132794
Main goals of this refactor:
* Make it more obvious which update function should be used.
* Make it more obvious which parameters are required by using references instead
of pointers.
* Support passing in multiple modified trees instead of just a single one.
No functional changes are expected.
Pull Request: https://projects.blender.org/blender/blender/pulls/132862
The `bNode.type_legacy` is still used in many places to check if a node is a
specific type. However, going forward it's better to rely more on the idname
than on this legacy integer type. Some more information is available in #132858.
The added utility method can be used like so:
`node->is_type("GeometryNodeMenuSwitch")`. Previously one would have written
`node->type_legacy == GEO_NODE_MENU_SWITCH`. The `is_type` method internally
checks that the passed in string is a valid node identifier to make it more
likely that we catch typos early.
Pull Request: https://projects.blender.org/blender/blender/pulls/132863
As proposed in #98518, this moves all `View2D` files into
a `source/blender/editors/inteface/view2d/` directory. This helps
keeping the `interface/` directory clean. In general I think we should
promote a more modular thinking, where coherent parts of the code form a
module or sub-module. The directory structure and its files can reflect
that nicely.
Pull Request: https://projects.blender.org/blender/blender/pulls/132853
Simply load factory startup file, and recursively iterate over
all (valid) `POINTER` and `COLLECTION` properties of all IDs.
NOTE: A few cases are currently returning invalid rna paths (at least
from quick look), these are skipped for now.
Pull Request: https://projects.blender.org/blender/blender/pulls/132809
The issue was that we were passing a pointer-to-a-pointer to
`BLI_addtail()`, which expects a pointer to something castable to a
`Link`. This in turn led to an invalid memory access when trying to
access the fields of the supposed `Link`.
This fixes the issue by passing a pointer to a zero-initialized `Link`
instead.
This also takes the opportunity to more simply zero-initialize the
`bAction` structs used in the tests as well.
Pull Request: https://projects.blender.org/blender/blender/pulls/132860
The new description for `bNode.type_legacy`:
```
/**
* Legacy integer type for nodes. It does not uniquely identify a node type, only the `idname`
* does that. For example, all custom nodes use #NODE_CUSTOM but do have different idnames.
* This is mainly kept for compatibility reasons.
*
* Currently, this type is also used in many parts of Blender, but that should slowly be phased
* out by either relying on idnames, accessor methods like `node.is_reroute()`.
*
* A main benefit of this integer type over using idnames currently is that integer comparison is
* much cheaper than string comparison, especially if many idnames have the same prefix (e.g.
* "GeometryNode"). Eventually, we could introduce cheap-to-compare runtime identifier for node
* types. That could mean e.g. using `ustring` for idnames (where string comparison is just
* pointer comparison), or using a run-time generated integer that is automatically assigned when
* node types are registered.
*/
```
Pull Request: https://projects.blender.org/blender/blender/pulls/132858
In various previous commits all the remaining usages of this file were removed
(b1d7e8fcb1, b1d7e8fcb1, 2afd946ba, b43e2168e3). Now this file is finally
completely unused, which means we can remove it.
The main reason why we don't want this file is that it heavily relied on the
preprocessor and makes it hard to decentralize implementation details of
individual nodes.
Pull Request: https://projects.blender.org/blender/blender/pulls/132855
This scene consists of two spheres (shade flat and shade smooth)
having their normals baked onto a plane using the "Bake from selected"
option.
Ref: blender/blender#132716
Ref: blender/blender-test-data!37
When the material index of a stroke is out-of-range of the material slots, the
strokes are never editable. This makes it hard to even re-assign the material or
otherwise remove such strokes. A negative index can happen when curves are added
while a Grease Pencil object has no material slots (see #123887).
Since `material_index` is a generic attribute and should be allowed any value,
strokes with out-of-range indices should be considered editable by default.
This patch slightly changes the `get_editable_material_indices` function for
tools, such that only __valid__ materials with the "Locked" flag set are
considered non-editable.
Pull Request: https://projects.blender.org/blender/blender/pulls/132806
The `smooth_curve_attribute` function smoothes individual
ranges of points that are selected within a stroke.
When smoothing a single range, the `gaussian_blur_1D` function
shouldn't treat this range as cyclic.
Instead, we check if the entire curve is selected and only use
the `cyclic` boolean in this case. For ranges, we always
treat them as non-cyclic.
Pull Request: https://projects.blender.org/blender/blender/pulls/132850
The `ZDepth` output of the camera data node was different between SVM
and OSL.
SVM would output the `ZDepth`, with negative distances for points
behind the camera. While OSL would output the absolute of the distance,
which resulted in points behind the camera becoming positive.
Align OSL to SVM and allow outputting the negative distance as it
allows users to differentiate between what's in front or behind the
camera.
Pull Request: https://projects.blender.org/blender/blender/pulls/132837
This removes the last usage of `NOD_static_types.h` after #132815. It does so by
essentially inlining it into `rna_nodetree.cc` in a more compact way.
To avoid the need for especial cases for some nodes, all node-specific functions
now also take the `BlenderRNA` as parameter instead of only the `StructRNA`.
This allows such node-specific functions to register additional subtypes if
necessary. I think it should be possible to use this to fully define more
complex dynamic nodes such as repeat zone nodes, but that's not included in this
patch.
Now a line to define a node rna struct looks like so: `define("GeometryNode",
"GeometryNodePointsOfCurve");`
The `GeometryNode` part of it is intentionally somewhat redundant to allow for
better searchability.
Pull Request: https://projects.blender.org/blender/blender/pulls/132823
This removes the second to last usage of `NOD_static_types.hh` which we intend
to remove. A nice benefit is that the idname is now finally more explicit when a
node is registered. Previously it was difficult to search for the definition of
a node in the code when one had only the idname, which is the main identifier
for nodes.
The main change is in `node_type_base`.
Pull Request: https://projects.blender.org/blender/blender/pulls/132815
The material index in Grease Pencil attributes can be negative, since they are a
generic attribute. The shader, however, requires clamping of the material index
to avoid an invalid lookup. This was already happening for the material buffer
setup (`grease_pencil_object_cache_populate`), but not for the actual VBO.
Fixes#131670
Pull Request: https://projects.blender.org/blender/blender/pulls/132804
When no image is present the depth and color buffers aren't correctly
cleared, resulting in showing the previous buffers.
Replaced the clear on bind with regular clearing operations.
Alternative implementation for !132751
Pull Request: https://projects.blender.org/blender/blender/pulls/132844
- Remove redundant .HasValue() calls
We are calling `GetPrimvarsWithValues` which will do the HasValue
checks for us already.
- Consistently skip non-array primvars
We only want to import in the array attributes. Consistently check for
this in each of our loops.
Pull Request: https://projects.blender.org/blender/blender/pulls/132816
- Prefer term "size" over "width" as this was used for the height.
- Simplify calculation for drawing the "+".
- Add note that the "+" is slightly off-center (ref #112653).
Implicit sharing dramatically reduces the cost of coping geometry
which is extremely helpful for read-only and slight modification.
But case of Sort Elements node implies a total rewrite of whole domain,
so implicit sharing only adds redundant cost for write access. This fix
uses kernel API to create domains without attributes, manually forcing
implicit sharing for offset indices and uses kernel attribute copy
functions to force sharing for all attributes when this is possible.
In all other cases new attributes are created from scratch and copied
as re-ordered only once.
In attached example file shown the difference depend on the number of
attributes on the mesh, benchmark result:
| Attributes | Before | After |
| ---------- | ------ | ----- |
| 1 | 2 ms | 1.5 ms |
| 5 | 5 ms | 4 ms |
| 10 | 9 ms | 6 ms |
| 50 | 45 ms | 30 ms |
| 100 | 89 ms | 59 ms |
Pull Request: https://projects.blender.org/blender/blender/pulls/117910
