This value was meant to be used for keeping images that are slowest to
render in cache. Method of measurement was flawed, because it doesn't
take UI overhead into consideration.
Cache panel is to be removed because users should not have to tweak
settings like this. It is not useful for development either, therefore
it is removed completely.
If image is cached twice, it's size has been counted twice as well, but
only image reference count is increased, not memory usage.
Use `MEM_get_memory_in_use()` instead of size own tracking.
This new node increases the radiance of an image by a scalar value.
Previously, the only way to adjust the the exposure of an image was with
math node or using the scene's color management.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D9677
This bug was caused by making it so that non-embossed modifier icon
buttons could become an operator button and retain their red highlight
for disabled modifiers. The icon button needs emboss turned off, but
in earlier versions of Blender, `UI_EMBOSS_NONE` would be overridden
by animation or red alert states.
Instead of abusing "NONE" to mean "none unless there is animation or
red alert", this commit adds a new emboss flag for that situation,
`UI_EMBOSS_NONE_OR_STATUS`, which uses no emboss unless there is an
animation state, or another status. There are only a few situations
where this is necessary, so the change isn't too big.
Differential Revision: https://developer.blender.org/D9902
The previous design is rather old and has a couple of problems:
* Scalability: The current solution of adding little icon buttons next to the
data-block name field doesn't scale well. It only works if there's a small
number of operations. We need to be able to place more items there for better
data-block management. Especially with the introduction of library overrides.
* Discoverability: It's not obvious what some of the icons do. They appear and
disappear, but it's not obvious why some are available at times and others
not.
* Unclear Status: Currently their library status (linked, indirectly linked,
broken link, library override) isn't really clear.
* Unusual behavior: Some of the icon buttons allow Shift or Ctrl clicking to
invoke alternative behaviors. This is not a usual pattern in Blender.
This patch does the following changes:
* Adds a menu to the right of the name button to access all kinds of operations
(create, delete, unlink, user management, library overrides, etc).
* Make good use of the "disabled hint" for tooltips, to explain why buttons are
disabled. The UI team wants to establish this as a good practise.
* Use superimposed icons for duplicate and unlink, rather than extra buttons
(uses less space, looks less distracting and is a nice + consistent design
language).
* Remove fake user and user count button, they are available from the menu now.
* Support tooltips for superimposed icons (committed mouse hover feedback to
master already).
* Slightly increase size of the name button - it was already a bit small
before, and the move from real buttons to superimposed icons reduces usable
space for the name itself.
* More clearly differentiate between duplicate and creating a new data-block.
The latter is only available in the menu.
* Display library status icon on the left (linked, missing library, overridden,
asset)
* Disables "Make Single User" button - in review we weren't sure if there are
good use-cases for it, so better to see if we can remove it.
Note that I do expect some aspects of this design to change still. I think some
changes are problematic, but others disagreed. I will open a feedback thread on
devtalk to see what others think.
Differential Revision: https://developer.blender.org/D8554
Reviewed by: Bastien Montagne
Design discussed and agreed on with the UI team, also see T79959.
Since "Float Color" is more commonly used than "Byte Color",
which is not exposed in the interface yet anyway, it makes sense to
drop the "Float" label on the color data type name.
This uses the "id" attribute to randomly pick instances from a collection
for each point.
There is one issue. When the collection is updated (e.g. when an object is
added to it), the nodes modifier is not automatically updated. It seems
like we don't have the infrastructure to support this dependency yet.
The same issue exists in the Boolean modifier and with collision collections.
This should be solved separately soonish.
When "Whole Collection" is disabled, one direct child of the input collection
is instanced at each point. A direct child can be an object or a collection.
Currently, all objects are picked approximately equally often. In the future,
we will provide more control over which point gets which instance.
Differential Revision: https://developer.blender.org/D9884
Ref T82372.
Caused by rB7470c10601d0 where iterating nodetree nodes was changed from
backwards to forwards by mistake.
Maniphest Tasks: T83916
Differential Revision: https://developer.blender.org/D9890
Change the top coordinate of the animation channel list UI elements to
`rect->ymin` so they correctly span from `rect->ymin` to
`channel_height`.
Some buttons in the dope sheet channels didn't scale properly with the
`Keyframe Scale Factor` preferences setting. This was caused by using
the `ymid` value (`ymid = BLI_rctf_cent_y(rect) - 0.5f * ICON_WIDTH`) to
position the buttons that supposed to fill all vertical space in the
channel (with `channel_height` height). The `ymid` value is only
appropriate for the UI elements that with `ICON_WIDTH` height.
Maniphest Tasks: T83716
Reviewed by: sybren
Differential Revision: https://developer.blender.org/D9841
This adds a GPointer class, which is mostly the same as GMutablePointer.
The main difference is that GPointer references const data, while GMutablePointer
references non-const data.
Ref: T82651
Normally people use "Combine XYZ" to input a vector, but it is more
interesting to have an explicit vector input.
So this is basically "Combine XYZ" without any input sockets, the values
are stored in the node itself.
Differential Revision: https://developer.blender.org/D9885
In glsl the clamp function has undefined behavior when min > max. For
the clamp node this resulted in differences between cycles and eevee.
This patch adds the expected implementation for minmax.
The old clamp function is still used in cases where we know for certain
that the input values are correct (math node clamp option). GPU uses
optimized code and silicon in these cases.
Some GPU platforms don't support having more than one underscore in
sequence in an attribute name. This change will remove the underscore
as a possible character when encoding to save names.
This patch introduces a partial update of GPUTexture. When rendering
a large image the GPUTexture could have been scaled. The old implementation
would rescale the image on CPU and create a new GPUTexture. This
resulted in flooding the PCI bus.
The new solution would only scale and upload the parts of the GPUTexture
that has been changed. It does this by keeping track of areas of 256x256
pixels. When something changes the tiles that cover that changes will be
rescaled and uploaded the next time the GPUTexture is requested.
Test situation: Default Cube, 4 samples, 19200x10800 tile size 512.
Blender 2.83.9: 4m27s.
Blender 2.91: 20+m (regression)
This patch: 1m01s.
There is still room for more optimizations:
* Reduce the time that an image is locked.
** Use task scheduling to update the tiles of an image.
** Generic optimization of the ImBuf scale method.
Maniphest Tasks: T82591
Differential Revision: https://developer.blender.org/D9591
The outliner already expands the panel for the modifier you click on,
this just extends that idea to also set it active, which is consistent
with behavior of active and selected items elsewhere in Blender.
After rB15083d9e1 the outliner tree is not rebuilt after expanding or
collapsing rows. Because the tree is no longer rebuilt the positions
and flags of the elements are not cleared when collapsing a row. This
caused hover highlights and selections on the collapsed child icons to
be incorrect in many cases.
For example, only the direct children of a collapsed element are drawn
inline. If any grandchild elements had been previously icon row flagged
they would continue to be evaluated as icon row elements despite being
hidden. In this case the x coordinates of the child and grandchild would
overlap causing selection to appear erratic.
Now the flags for inline row icons are explicitly cleared, which was
previously only done because the tree was rebuilt on collapsing rows.
Expand abbreviations for words like "Bright" (instead of "Brightness"),
"Premul", "Lib", "Dir", etc.
Differential Revision: https://developer.blender.org/D9862
Since creating the attribute node, a helper function has been added to
automatically get the input attribute or a constant value, depending on
the "input type" values for the node. This commit replaces the specific
implementation of that behavior with the new helper function.
The versioning is necessary since the node now has a "storage" struct.
Previously these HSV values were in the color picking space, which meant the
relation to the scene linear RGB values was confusing.
The new situation:
* RGB number buttons: scene linear color space
* HSV number buttons: scene linear color space
* Picker widgets: color picking color space
* Hex: sRGB color space
Fixes T69562, T83853, Ref T68926
This patch adds two related nodes, a node for separating points
and mesh vertices based on a boolean attribute input, and a node
for creating boolean attributes with comparisons.
See the differential for an example file and video.
Point Separate (T83059)
The output in both geometries is just point data, contained in the mesh
and point cloud components, depending which components had data in the
input geometry. Any points with the mask attribute set to true will be
moved from the first geometry output to the second. This means that
for meshes, all edge and face data will be removed. Any point domain
attributes are moved to the correct output geometry as well.
Attribute Compare (T83057)
The attribute compare does the "Equal" and "Not Equal" operations by
comparing vectors and colors based on their distance from each other.
For other operations, the comparison is between the lengths of the
vector inputs. In general, the highest complexity data type is used
for the operation, and a new function to determine that is added.
Differential Revision: https://developer.blender.org/D9876
In case of being in Asset browsing mode, the search field is located in
the header (RGN_TYPE_HEADER not RGN_TYPE_UI as for file browsing mode).
To be future proof, now iterate all regions and act if the
"filter_search" can be activated.
Maniphest Tasks: T83888
Differential Revision: https://developer.blender.org/D9882
Caused by Caused by rB83980506957c.
Since above commit, the modifier was created with wrong initial values
[amount was 0.0 and offset was 1.0 -- instead of the other way around].
Since there is no way to fix existing files in a reasonable way I guess,
all we can do here is to make sure that from now on, the defaults are
correct.
Maniphest Tasks: T83886
Differential Revision: https://developer.blender.org/D9881
Currently, the random attribute node doesn't work well for most
workflows because for any change in the input data it outputs
completely different results.
This patch adds an implicit seed attribute input to the node, referred
to by "id". The attribute is hashed for each element using the CPPType
system's hash method, meaning the attribute can have any data type.
Supporting any data type is also important so any attribute can be
copied into the "id" attribute and used as a seed.
The "id" attribute is an example of a "reserved name" attribute,
meaning attributes with this name can be used implicitly by nodes like
the random attribute node. Although it makes it a bit more difficult
to dig deeper, using the name implicitly rather than exposing it as an
input should make the system more accessible and predictable.
Differential Revision: https://developer.blender.org/D9832
The size of the nodes is not enough to give enough context to users what
the sockets are about.
Minimum Distance -> Distance Min
Maximum Density -> Distance Min
Note this does not handle doversion. That means users will have to
manually.
Some weird proxies apparently can have a local collection instancing...
Not sure this is even really valid for proxies, but in any case we
cannot override that, just detect and properly cancel the operation
then.
Should be backported to 2.91.1 should we do it.
Data of the File Browser context callback needs to be validated and
return `CTX_RESULT_NO_DATA` if the context member is valid but not set
in the current context.
Callback function was using int while update_render_passes_cb_t was
using eNodeSocketDatatype leading to a build warning about different
argument types with MSVC.
Crash happens when using relative path to image in operator properties
and checking image dimensions by loading image with `IMB_loadiffname()`
Ensure path is absolute.
Crash on null dereference in `SEQ_timeline_boundbox()`. This function was
generalized in rB9e4a4c2e996c to work on arbitrary `seqbase`.
Fixed by refactoring `SEQ_timeline_boundbox()` functions to return default
sane values if `seqbase` is `NULL`
Reviewed By: HooglyBoogly
Differential Revision: https://developer.blender.org/D9878
Anchored brushes with spherical falloff start off with zero radius, thus
we have no pbvh nodes on the first brush step. This would prevent
initializing the automasking cache [which only happens on the first brush
step].
Maniphest Tasks: T83856
Differential Revision: https://developer.blender.org/D9873
Since a few versions (even before 2.79) we have an option that allows to restrict the vertex tools to operate only on deform groups. This was originally implemented for working with vertex groups for skeletal animation. In that case it is fortunate to have weight tools operate only on deforming vertext groups (vgroups assigned to bones)
In previous versions of Blender (up to 2.79) we have been able to use this option in Mesh Edit mode regardless of the armature mode. The current implementation (since 2.80 as far as i know) enables this option only when the associated armature is in pose mode. this has a bad consequence:
It is not at all intuitive that you have to put the armature into Pose mode before you can make use of the option in mesh edit mode.
Besides this it is not even necessary in the case when the user wants to restrict the tool only to all pose bones. In that case the armature can safely be kept in Object mode. However, when the tool shall apply only to selected pose bones, then it actually makes sense to have the armature in pose mode (as it is implemented right now)
I do not know why this feature has been restricted as described above. It must have got lost somewhere on the way to Blender 2.90
This patch fixes the issue as it allows to select the "restrict to pose bones" option when the armature is in any mode. I see no downsides of this change, actually this is a fix for a feature that once worked and apparently got forgotten in newer releases.
Reviewed By: sybren, campbellbarton
Differential Revision: https://developer.blender.org/D9658
This commit resolves problem introduced in e1665c3d31 - it was
difficult to import media at their original resolution.
This is done by using original resolution as reference for scale.
All crop and strip transform values and their animation is converted
form old files.
To make both workflows easy to use, sequencer tool settings have been
created with preset for preffered scaling method. This setting is in
sequencer timeline header and add image or movie strip operator
properties.
Two new operators have been added:
`sequencer.strip_transform_fit` operator with 3 options: Scale To Fit,
Scale to Fill and Stretch To Fill.
Operator can fail if strip image or video is not loaded currently, this
case should be either sanitized or data loaded on demand.
`sequencer.strip_transform_clear` operator with 4 options:
Clear position, scale, rotation and all (previous 3 options combined).
Reviewed By: sergey, fsiddi
Differential Revision: https://developer.blender.org/D9582
This adds a boolean attribute and custom data type, to be used in the
point separate node. It also adds it as supported data types in the
random attribute and attribute fill nodes.
There are more clever ways of storing a boolean attribute that make
more sense in certain situations-- sets, bitfields, and others, this
commit keeps it simple, saving those changes for when there is a proper
use case for them. In any case, we will still probably always want the
idea of a boolean attribute.
Differential Revision: https://developer.blender.org/D9818
