This operator is a combination of the previous delete operator `frame` mode
and the `active_frames_delete_all` operator.
This also add the delete menu and key binds.
The new node interface items can be dragged and dropped to change
their hierarchy. The messages indicating where an element is dropped
were extracted using N_(), but not actually translated.
This commit enables this translation using TIP_(). It also extracts a
missing message ("Insert into panel").
Pull Request: https://projects.blender.org/blender/blender/pulls/112108
- Use context variables for drawing functions.
- Move conditions to functions drawing specific elements as
preconditions
- Disentangle calculation of common coordinates.
- Use more descriptive variable names (in context at least)
Individual drawing functions are mostly unchanged, so improvements there
are minimal.
No intended functional changes.
This adds a new Skip input to the Simulation Output node (design task: #112082).
It is a convenience feature that makes it easy to conditionally forward the
output of the Simulation Input node to the Simulation Output node, without the
need for potentially multiple Switch nodes. When Skip is enabled, the other inputs
of the Simulation Output node are not evaluated, i.e. the nodes in the simulation
zone are ignored.
The implementation adds this new functionality directly to the `LazyFunction`
of the Simulation Output node. It has new inputs that are linked directly
to the Simulation Input node, so that the simulation state can be forwarded.
Pull Request: https://projects.blender.org/blender/blender/pulls/112140
Merge all the small UBOs used by the engine to save binding slots.
Each module is still responsible for filling its own data (by storing a
reference to it at construction) and this data is still private for other
modules.
The engine instance pushes the data to the GPU at the end of
`end_sync`, so only the modules that modify their data outside of the
sync functions need to manually call `Instance::push_uniform_data()`.
Pull Request: https://projects.blender.org/blender/blender/pulls/112046
was mixing real geometry normal, smoothed geometry normal and
bump-mapped normal.
Use `(sd->type & PRIMITIVE_CURVE) ? sc->N : sd->Ng` consistently instead.
Users expect to be able to scale the uppermost edge of the asset shelf,
which actually belongs to the asset shelf header region only. So it
would only work to hide/unhide this. They expect this because they seem
like one region, but they are actually implemented as two (to have
separate layout and scrolling mostly).
This adds a region flag so that scaling a region can actually affect the
previous one instead. Something similar is already used for split
regions.
Part of #107881.
We already have two other functions doing very much the same thing.
`bsdf_microfacet_sample()` seems to be the only place where this
function was used; there we always sample visible normals, so the extra
`inside` check is not needed.
1. move early-out logic even earlier
2. reduced the scope of some variables
3. return `label` at the end of `bsdf_microfacet_sample()`. Return
`LABEL_NONE` in the invalid case. The previous distinction was
unnecessary because samples with zero contribution are assigned with
`LABEL_NONE` in `integrate_surface_bsdf_bssrdf_bounce()` anyway.
the motivation was to give closures with low weight a higher pdf to pick
at the first bounce, in case the next interaction has high contribution.
However, there are several issues:
1. this is too much fine-tuned for a specific case, and only works well
when there is a strong contribution after reflection and very little
contribution after the transmission;
2. the logic in `bsdf_microfacet.h` was added when merging reflection
and refraction into a glass closure, since then it doesn't even work
well in the above case when mixed with other closures;
3. The behavior is inconsistent in `bsdf_microfacet_eval()` and
`bsdf_microfacet_sample()`;
4. such cases should be handled by more modern and more general methods
such as path guiding and denoiser;
5. it makes the code flow harder to follow
Delete this trick for now to pick the closures solely based on their
`sample_weight`. Can be added back (with proper fix in
`bsdf_microfacet`) if indeed necessary.
bd9f94e917 made it so the file browser doesn't bring files online for
the purpuse of creating their thumbnail, because that can take a while.
Instead it uses a previously cached thumbnail if available. This should
be the behavior for all cases thumbnails are requested, it's not only
the file browser that does this.
In fact it makes sense to move this into the normal function to "manage"
thumbnails (that is, load and if necessary (re)create cached
thumbnails) since there are no currently known use-cases for
different behavior.
Also, seems like the previous solution didn't work when loading ID
previews from offline .blend files. For that we need to use the path to
the .blend file to check the offline status, not the full path to the ID.
Found while working on #109234 (Use UI preview system for async loading of
file/asset previews).
Pull Request: https://projects.blender.org/blender/blender/pulls/112101
We need a separate array that we can change in during the parallel
group construction. That array tells where in each group the index
is added. Building this array is expensive, since construcing a new
`Array` fills its elements serially. There are two possible solutions:
1. Use a copy of the offsets to increment result indices directly
2. Rely on OS-optimized `calloc` instead of `malloc` and a copy/fill
Both depend on using `fetch_and_add` instead of `add_and_fetch`.
The vertex to corner and edge to corner map creation is optimized
by this commit, though the benefits will be useful elsewhere in the
future.
| | Before | 1. offsets copy | 2. calloc |
| -------- | ------- | --------------- | --------------- |
| Grid 1m | 3.1 ms | 1.9 ms (1.63x) | 1.8 ms (1.72x) |
| Grid 16m | 51.8 ms | 33.3 ms (1.55x) | 32.7 ms (1.58x) |
This commit implements the calloc solution, since it's slightly faster
and simpler. In the future, `Array` could do this optimization itself
when it detects that its fill value is just zero bytes.
Pull Request: https://projects.blender.org/blender/blender/pulls/112065
When duplicating the node group from the geometry node editor, show the
data-block selector in the modifier interface. Otherwise it's not clear
that the modifier is using a local data-block, not the original asset.
While it isn't "wrong" to have the name listed below too, In Blender
the name of an item is usually edited by double clicking inside
of list/tree views, and it's better to be consistent.
Recent versions of DPC++ dropped using the environment variable
SYCL_PI_LEVEL_ZERO_USE_COPY_ENGINE_FOR_IN_ORDER_QUEUE=0 we were setting.
We're now also setting SYCL_PI_LEVEL_ZERO_USE_COPY_ENGINE=0 by default
to keep a consistent behavior.
The API for the kernels library is defined, there is no need to
export more than that. This change only affects linux since hidden
visiblity is the default on Windows.
Use the asset icon for the "Mark Asset" operator in menus.
Using icons is not only good for accessibility, but also to create
a connection with the Asset Browser and the icon shown in the
data-block template once marked as asset.
Pull Request: https://projects.blender.org/blender/blender/pulls/112111
The recursive building has an early output, which could leave the node
without the requested child initialized, causing the access to the left
child to crash later on in the builder function.
Simply add an extra is-canceled check to avoid access of possibly
non-initialized tree structure.
Pull Request: https://projects.blender.org/blender/blender/pulls/112132
Reasons to keep calling it Geometry Nodes:
* People can search for Geometry or Nodes and find it.
* It makes a connection with the editor called Geometry Nodes.
Interesting enough these were similar reasons to the original discussion
we had when Geometry Nodes first got added to Blender. Basically the
reasoning is still valid.
Counter-arguments:
* Over-time more and more modifiers will be geometry-nodes based, it
could be seem as strange to single out one of them.
* Once we have different node-base systems that may also work as
modifiers (e.g., collection modifiers?) we will need a more unified
name.
The response to the latter is that once we get to this we will also
rename the editor. Until then we keep what is working.
Regression [0] which merged index assignment into an earlier loop.
This caused the indices to be overwritten by p_chart_abf_solve
causing unwrap to fail.
[0]: 8ace65e3c6
