This commit adds a new navigation preset called `Tablet/Trackpad` which enables "Emulate 3 Button Mouse" to more quickly set up good default keys for tablet users.
It also adds support for mouse buttons 4 and 5 in the navigation settings which will be helpful if users want to customize 3D navigation further for specific pens/mice.
The issue was that Node3DEditorViewport was using the render server's BVH to filter out nodes,
which is not correct for gizmos that have no renderable components,
or have collision triangles that exceed the bounds of their renderable components.
This commit adds 3 new editor settings for orbit, pan, and zoom mouse buttons, and 6 new shortcuts which act as modifiers for the navigation controls. These new shortcuts replace the old orbit, pan, and zoom modifier settings.
The `navigation_scheme` setting now acts as a preset which changes the new options added above, and the new settings are what drives 3D navigation instead. A new struct is used for ordering the navigation logic so that actions with fewer shortcuts are checked first. When the editor starts, the preset detection will run to automatically update user settings from old Godot versions. When the setting is changed, the hint values for the mouse buttons are dynamically updated to show the user the corresponding shortcut values.
The new doc fields have been generated and the description for the new settings are filled out. The `navigation_scheme` entry now has more consistent styling and added control descriptions that were missing before.
- Create AudioStreamPlayer if dropped in between nodes in the Scene dock
- Create AudioStreamPlayer2D if dropped into 2D editor
- Create AudioStreamPlayer3D if dropped into 3D editor
Fixes certain issues where sub-pixel motions would get discarded while
the mouse is captured, such as when free look is enabled in the editor
(at least when turned on while holding right click).
Very slightly compat breaking, as actual public APIs are changed,
although with "compatible" types (Point2i->Point2).
Each time an AABB is rotated, it gets bigger. That means opposite rotations don't cancel out.
The previous implementation repeatedly rotates children AABBs as it climbs up the tree. This often resulted in selection boxes looking bigger than their contents.
This implementation calculates and applies a single final transformation to each AABB before it is merged with the others. After merging, there are no additional rotations, so AABBs remain accurate.
Co-Authored-By: Robert Yevdokimov <105675984+ryevdokimov@users.noreply.github.com>
This change introduces a new EditorThemeManager class
to abstract theme generatio and its subroutines.
Logic related to EditorTheme, EditorColorMap, and editor
icons has been extracted into their respective files with
includes cleaned up.
All related files have been moved to a separate folder to
better scope them in the project. This includes relevant
generated files as well.
Introduces support for FSR2 as a new upscaler option available from the project settings. Also introduces an specific render list for surfaces that require motion and the ability to derive motion vectors from depth buffer and camera motion.
This allows the user to input numbers during an "instant" (blender
style) transform operation to specify exactly how far to transform the
object. For example:
g2.5xx: Translate 2.5 units along the local x-axis
ry-45: Rotate -45 degrees around the y-axis
s.25Z: Scale by a factor of .25 on the xy plane
Some shared code between the traslate/rotate/scale branches of update_transform
was refactored into apply_transform so numeric transforms could reuse it.
This removes any "{X,Y,Z}-Axis Transform" messages. These prevented the
"Transforming: (x,y,z)" messages from showing, and the latter are more
useful, as they tell you the actual units.
This also rearranges finish_transform to clear _edit before updating
the axis rendering, so an axis doesn't remain highlighted.
Co-authored-by: Rémi Verschelde <rverschelde@gmail.com>
When performing a blender-style translate or scale (but not rotate),
wrap the mouse inside the bounds of the viewport. This allows moving the
mouse indefinitely for large translate or scale operations, and is
consistent with how blender works.
This enables global input processing during blender-style transforms, so
we can capture and wrap mouse events outside the bounds of the viewport.
Other wrapping operations use _sinput, which I assume is possible
because a mouse button is held, so the control receives input even when
the mouse exits the control.
This does not implement wrapping for regular (gizmo-driven) transforms,
though this could be done if desired.
Fixesgodotengine/godot-proposals#4255.
Do not try to format fields with garbage values in _snap_update()
Initialize grid_enable[] before use.
Initialize previewing_camera before use.
These are all cases found live, with valgrind.
Fixes#76925
