Applying overlay materials into multi-surface meshes currently
requires adding a next pass material to all the surfaces, which
might be cumbersome when the material is to be applied to a range
of different geometries. This also makes it not trivial to use
AnimationPlayer to control the material in case of visual effects.
The material_override property is not an option as it works
replacing the active material for the surfaces, not adding a new pass.
This commit adds the material_overlay property to GeometryInstance
(and therefore MeshInstance), having the same reach as
material_override (that is, all surfaces) but adding a new material
pass on top of the active materials, instead of replacing them.
Implemented in rasterizer of both GLES2 and GLES3.
having the raycast distance hardcoded to `10000` caused input events
to not be registered in very large 3D scenes.
This resolves the issue by using the cameras far distance instead.
Creating the more predictable behavior of if an object is visible,
it will be picked by the viewport.
resolves: #49735
(cherry picked from commit 02b6bbc5df)
Now uses Ctrl + Scrollwheel for vertical scrolling
Ctrl + Shift Scrollwheel for horizontal scrolling
Also converts some macros to constants
(cherry picked from commit 13ba788b63)
Previously a crude metric was used to decide on the roaming expansion margin, but it created unexpected results in some scenarios. Instead this setting is exposed to the user via the RoomManager, allowing them to tailor it to the world size, room sizes, roaming objects sizes and the speeds of movement.
Reporting rest collision information is needed for move_and_collide and
move_and_slide so floor detection can be done properly, but in the case
of just testing the motion for collision, it makes sense to return false
if the body is able to move all along the path without being stopped.
Updated the logic in test_move and clarified the documentation for
test_move and move_and_collide.
In all physics servers, body_get_direct_state() now silently returns
nullptr when the body has been already freed or is removed from space,
so the client code can detect this state and invalidate the body rid.
In 2D, there is no change in behavior (just no more errors).
In 3D, the Bullet server returned a valid direct body state when the
body was removed from the physics space, but in this case it didn't
make sense to use the information from the body state.
Async. compilation via ubershader is currently available in the scene and particles shaders only.
Bonus:
- Use `#if defined()` syntax for not true conditionals, so they don't unnecessarily take a bit in the version flagset.
- Remove unused `ENABLE_CLIP_ALPHA` from scene shader.
- Remove unused `PARTICLES_COPY` from the particles shader.
- Remove unused uniform related code.
- Shader language/compiler: use ordered hash maps for deterministic code generation (needed for caching).
NodePath properties are designed to be relative to the given node, so
validity checks are failing in the editor for Polygon2D nodes, which are
relative to the Skeleton2D node rather than the Polygon2D node.
Fixed by saving bone paths as String properties instead of NodePath.
Shouldn't cause a difference for performance since NodePath properties
are technically saved as String anyway.
(cherry picked from commit 8d9619ad46)
Changes the Path2D drawing to use POLYLINE instead of thick lines.
Add a path to translate thick lines (that are not using anti-aliasing) to draw as polygons instead. This should be faster because polygons can be batched.
Sets `AlignOperands` to `DontAlign`.
`clang-format` developers seem to mostly care about space-based indentation and
every other version of clang-format breaks the bad mismatch of tabs and spaces
that it seems to use for operand alignment. So it's better without, so that it
respects our two-tabs `ContinuationIndentWidth`.
* Changed another instance of ERR_FAIL_COND in the same file to
ERR_FAIL_NULL_MSG instead.
* Checked for potential access of the viewport pointer elsewhere in same
file.
Fixes#54098
(cherry picked from commit 497b00a937)
Add new error type to BakeError to signalise that there is no root node to
start baking from.
Document the new error type, as well as BAKE_ERROR_NO_LIGHTMAPPER.
When switching the mesh at runtime, the physics server wasn't properly
updated with the new mesh. Now we keep track of the soft body mesh to
make sure everything is properly initialized on pre-draw.
Also cleaned a few things around private methods.
Fix some typoed names from the doc
Add _index to "index" parameters of *_shape_* signals, this is both in doc and in the template. This makes the code, signature and doc easier to understand
Add method to get Node from the _index params of those signals. This was not as easy to find as one would expect. Putting this information where it is needed will help.
Very low wait times behave in unpredictable ways depending on the
rendered frame rate. This is because the timeout signal is only emitted
once per rendered frame (or physics frame, depending on the timer's
process mode).
When packing a scene node which is not the root, errors where caused
by internal checks in is_editable_instance method.
This check can be safely made outside instead.
(cherry picked from commit 2f6cf3b8a8)
In case the reference is stored in script, create a new instance to
avoid overriding the previous values.
Otherwise, re-use the reference as before to avoid extra allocations.
Fixes shear effect with `BILLBOARD_FIXED_Y` when the camera is rotated around the z-axis by rotating the mesh correctly into view space.
Also removes shearing effects that occur when rotating the mesh by excluding the model rotation and scale from the billboard matrix.
(cherry picked from commit d3850a0954)
Fixes#43733: "creating SpatialMaterial in a separate thread creates invalid
shaders (temporarily)."
The bug occurred because various setters called in materials' constructors add
materials to queues that are processed on the main thread. This means that
when the materials are created in another thread, they can be processed on the
main thread before the constructor has finished.
The fix adds a flag to affected materials that prevents them from being added
to the queue until their constructors have finished initialising all the
members.
Extra:
- Optimized the debug-only check about why the object is null to determine if it's because it has been deleted (the RC is enough; no need to check the ObjectDB).
- Because of the previous point. the debugger being attached is not required anymore for giving the "Object was deleted" error; from now, it only matters that it's a debug build.
- `is_instance_valid()` is now trustworthy. It will return `true` if, and only if, the last object assigned to a `Variant` is still alive (and not if a new object happened to be created at the same memory address of the old one).
- Replacements of `instance_validate()` are used where possible `Variant::is_invalid_object()` is introduced to help with that. (GDScript's `is_instance_valid()` is good.)
The icon was present in `editor/icons/`, but it was never implemented
in the editor gizmos code.
This also removes some unused gizmo drawing code (overridden methods
that are no longer called anywhere).
At runtime, packed scenes with nodes marked as editable instance where
saved with node type tags, which prevented the scene to be then loaded
as an instance, causing duplicated nodes in the tree.
This change ensures nodes marked as editable instances and their owned
children are properly set as instances.
That doesn't make a difference in the editor, since such nodes where
already set as instances based on their instance state, but it helps
at runtime where instance states are disabled.
Co-authored-by: latorril <latorril@gmail.com>
(cherry picked from commit fab88a810c)
