The method of accumulating values to create a hash for each chunk has
been improved for ~16% better distribution of the resulting hashes.
Improve performance of array de-duplication, see: #105046.
Accumulating hashes with a byte/boolean array didn't include enough
information for a useful hash, creating hashes with many collisions.
This is the root cause of a performance regression since 3.3 where
mesh data (used for storing edit-mesh undo steps) was changed to store
selection in a boolean array, creating a bottleneck de-duplicating
chunks of that array for edit-mesh undo's custom-data de-duplication.
Resolve by increasing hash accumulation for arrays with smaller elements,
so each chunk of memory (a candidate for de-duplication) isn't as likely
to have hash collisions.
`char` (single byte) arrays now accumulate 22 values instead of 7, it's
taking more values into account was necessary as these are effectively
bits in the case of boolean arrays, 2-byte values accumulate 32 bytes,
4-byte elements accumulate 44 bytes, larger structs accumulate
`sizeof(type) * 7` bytes (as before).
Also ensure the accumulation read-ahead never exceeds the chunk size -
technically a fix although this would only happen when passing a small
`chunk_count` to BLI_array_store_create (in the range of 1-16) so this
didn't happen in practice.
Improve performance of array de-duplication, see: #105046.
Use uint32_t since it's sufficient for hashing, using an int64_t was
especially inefficient when allocating an int64_t for every boolean
(when compacting an array of booleans).
Improve performance of array de-duplication, see: #105046.
While this behavior can be useful in some cases, it can also create
issues (as in one of own recent commits, 3e5ce23c99), since it
implicetly keeps the removed linknode 'linked' to the listbase.
At least warn about it in the documentation of `BLI_remlink`.
This commit adds the default .usdz export capability.
The importer already supports usdz so no change is required other than updating the text in menu to match the updated exporter text.
On export, a .usd/a/c file has to be created first, and then converted to .usdz. A weird change of directory is required because of a quirk with the USD conversion to usdz. If an absolute filepath is passed into the `UsdUtilsCreateNewUsdzPackage` function, the usd files inside the usdz archive will have the same directory structure, i.e. if one tries to create a file at `C:\code\BlenderProjects\file.usdz`, when the usdz file is created, inside it will have the structure `\code\BlenderProjects\file.usdc`.
This is counteracted by setting the current working directory to the temporary session directory where both the usdc and usdz files are created, and just passing the file name to `UsdUtilsCreateNewUsdzPackage` without any filepath. Once the usdz file is created it is moved to the intended directory.
There is a separate `UsdUtilsCreateNewARKitUsdzPackage` capability for exporting usdz for iOS devices that will be implemented in a follow up patch as it will require some more small UI changes.
Co-authored-by: Charles Wardlaw (@CharlesWardlaw)
Co-authored-by: Sonny Campbell (@SonnyCampbell_Unity)
Co-authored-by: Bastien Montagne (@mont29)
Pull Request #105185, based on #104556.
Pull Request #105223
While some implementations of `getcwd` may return an allocated string
instead of the given char buffer in some cases, this is not the expected
behavior of the BLI wrapper. Not to mention the danger of returning a
pointer to a static char buffer...
Improve `ChangeWorkingDirectoryTest` to be more 'full check' regarding
behavior of both `BLI_current_working_dir` and `BLI_change_working_dir`.
Also move call to `BLI_threadapi_init` into proper `SetUp` method (to
have correct symmetry with the call to `BLI_threadapi_exit` in the
`TearDown` one).
Based on investigation by Charles Wardlaw (@CharlesWardlaw).
Pull Request #105220
Add `BLI_change_working_dir(path)` to change the current working directory.
This change is required for adding USDZ support to Blender. When exporting to that format, we are required to do a weird change of directory because of a quirk with the USD library's USDZ functionality. If an absolute filepath is passed into the `UsdUtilsCreateNewUsdzPackage` function, the USDZ archive will store that full path.
macOS uses `NSFileManager` through some new Mac-only wrapper functions.
Ref #99807
Pull Request #104525
The code below has to handle the case when `sample_length` is greater
or equal to the total length already anyway, so we can just make that
a valid input. Currently, the snake hook brush for curves also makes
use of passing in larger sample lengths which current results in crashes
in debug builds.
This reverts commit 19222627c6.
Something went wrong here, seems like this commit merged the main branch
into the release branch, which should never be done.
This reverts commit 68181c2560.
I merged 3.6 into 3.5 by mistake. Basically I had a PR against main,
then changed it in the last minute to be against 3.5 via the
web-interface unaware that I shouldn't do it without updating the
patch.
Original Pull Request: #104889
Note that the node group has its sockets names
translated, while the built-in nodes don't.
So we need to use data_ for the built-in nodes names,
and the sockets of the created node groups.
Pull Request #104889
This better aligns with OSX/Linux warnings.
Although `__pragma(warning(suppress:4100))` is not the same as
`__attribute__((__unused__))` in gcc (which only affects the attribute
instead of the line), it still seems to be better to use it than to
hide the warning entirely.
Using larger integer types allows for more efficient code, because we
can use the hardware better. Instead of working on individual bytes,
the code can now work on 8 bytes at a time. We don't really benefit
from this immediately but I'm planning to implement some more optimized
bit vector operations for #104629.
Pull Request #104658
Add a test to address the issue raised in #103913, where zero area
triangles could be created from polygons that have co-linear edges
but were not degenerate.
During hair grooming in curves sculpt mode, it is very useful when hair strands
are prevented from intersecting with the surface mesh. Unfortunately, it also
decreases performance significantly so we don't want it to be turned on all the time.
The surface collision is used by the Comb, Pinch and Puff brushes currently.
It can be turned on or off on a per-geometry basis.
The intersection prevention quality of this patch is not perfect yet. This can
be improved over time using a better solver. Overall, perfect collision detection
at the cost of bad performance is not necessary for interactive sculpting,
because the user can fix small mistakes very quickly. Nevertheless, the quality
can probably still be improved significantly without too big slow-downs depending
on the use case. This can be done separately from this patch.
Pull Request #104469
Implements virtual shadow mapping for EEVEE-Next primary shadow solution.
This technique aims to deliver really high precision shadowing for many
lights while keeping a relatively low cost.
The technique works by splitting each shadows in tiles that are only
allocated & updated on demand by visible surfaces and volumes.
Local lights use cubemap projection with mipmap level of detail to adapt
the resolution to the receiver distance.
Sun lights use clipmap distribution or cascade distribution (depending on
which is better) for selecting the level of detail with the distance to
the camera.
Current maximum shadow precision for local light is about 1 pixel per 0.01
degrees.
For sun light, the maximum resolution is based on the camera far clip
distance which sets the most coarse clipmap.
## Limitation:
Alpha Blended surfaces might not get correct shadowing in some corner
casses. This is to be fixed in another commit.
While resolution is greatly increase, it is still finite. It is virtually
equivalent to one 8K shadow per shadow cube face and per clipmap level.
There is no filtering present for now.
## Parameters:
Shadow Pool Size: In bytes, amount of GPU memory to dedicate to the
shadow pool (is allocated per viewport).
Shadow Scaling: Scale the shadow resolution. Base resolution should
target subpixel accuracy (within the limitation of the technique).
Related to #93220
Related to #104472
The ear clipping method used by polyfill_2d only excluded concave ears
which meant ears exactly co-linear edges created zero area triangles
even when convex ears are available.
While polyfill_2d prioritizes performance over *pretty* results,
there is no need to pick degenerate triangles with other candidates
are available. As noted in code-comments, callers that require higher
quality tessellation should use BLI_polyfill_beautify.
Straightforward port. I took the oportunity to remove some C vector
functions (ex: copy_v2_v2).
This makes some changes to DRWView to accomodate the alignement
requirements of the float4x4 type.
`9c14039a8f4b5f` broke blenlib tests in release builds, due to how
`EXPECT_BLI_ASSERT` works (in release builds it just calls the given
function, so if that crashes then the test fails).
For now remove that check in the test.
If GetFileAttributesW returns an error, only debug assert if the reason
is file not found.
See D17204 for more details.
Differential Revision: https://developer.blender.org/D17204
Reviewed by Ray Molenkamp
Straightforward port. I took the oportunity to remove some C vector
functions (ex: `copy_v2_v2`).
This makes some changes to DRWView to accomodate the alignement
requirements of the float4x4 type.
After Win32 API call GetFileAttributesW, check for
INVALID_FILE_ATTRIBUTES, which is returned on error,
usually if file not found.
See D17176 for more details.
Differential Revision: https://developer.blender.org/D17176
Reviewed by Ray Molenkamp
Thoses are added for better component masking syntax.
This avoids the harder to read `float2(my_vec4)`.
This is not meant to be fully usable swizzling support but could be
extended in the future.
