Conversation
…k for transformations Signed-off-by: pengyiqiang <pengyiqiang@xiaomi.com>
VG_LITE_FILTER_POINT computes the initial sample coordinate as: c_step = 0.5 * step + translate (from inverse matrix) When the fractional part of translate is exactly 0.5, c_step lands on a pixel boundary where hardware floor-rounding is ambiguous -- this causes visible artifacts (confirmed: 0.5 fails, 0.25/0.75 are fine). Fix lv_vg_lite_matrix_has_transform() to detect sub-pixel translations whose fractional part falls within [0.4, 0.6] of a pixel boundary, and return true to force VG_LITE_FILTER_BI_LINEAR in those cases. Signed-off-by: pengyiqiang <pengyiqiang@xiaomi.com>
FASTSHIFT
requested review from
AndreCostaaa,
axos88 and
kisvegabor
as code owners
September 21, 2026 13:42
Contributor
PR report
Coverage — 100.0% of 1 changed line(s) (too few to enforce 50%) · overall 82.22% (vs master 82.22%, -0.00%)1 of 1 newly coverable line(s) are covered by the test suite. Overall: 56010 of 68126 coverable line(s) are covered by the test suite. The latest master build, Performance — no change in render_time
The full per-scene tables are in the job summary. Memory Usage — no changeLinked image (
|
| Target | Config | Flash | master | Δ Flash | RAM | Δ RAM |
|---|---|---|---|---|---|---|
| arm-cortex-m4 | standard | 381.926 KB | 381.926 KB | +0 B | 365.559 KB | +0 B |
| arm-cortex-m4 | rgb565 | 340.785 KB | 340.785 KB | +0 B | 365.559 KB | +0 B |
| arm-cortex-m4 | minimal | 156.406 KB | 156.406 KB | +0 B | 75.746 KB | +0 B |
| arm-cortex-m0plus | standard | 392.184 KB | 392.184 KB | +0 B | 365.559 KB | +0 B |
| arm-cortex-m0plus | rgb565 | 348.312 KB | 348.312 KB | +0 B | 365.559 KB | +0 B |
| arm-cortex-m0plus | minimal | 164.464 KB | 164.464 KB | +0 B | 75.746 KB | +0 B |
| xtensa-esp32 | standard | 405.805 KB | 405.805 KB | +0 B | 365.582 KB | +0 B |
| xtensa-esp32 | rgb565 | 354.602 KB | 354.602 KB | +0 B | 365.582 KB | +0 B |
| xtensa-esp32 | minimal | 175.027 KB | 175.027 KB | +0 B | 75.746 KB | +0 B |
| riscv32-esp32c3 | standard | 417.809 KB | 417.809 KB | +0 B | 365.582 KB | +0 B |
| riscv32-esp32c3 | rgb565 | 369.906 KB | 369.906 KB | +0 B | 365.582 KB | +0 B |
| riscv32-esp32c3 | minimal | 178.504 KB | 178.504 KB | +0 B | 75.746 KB | +0 B |
LVGL library, nothing dead-stripped
liblvgl.a in full. The linked image above only contains what the test application reaches. Useful to know how much memory LVGL will require when using a binding like MicroPython
| Target | Config | Library | master | Δ |
|---|---|---|---|---|
| arm-cortex-m4 | standard | 339.410 KB | 339.410 KB | +0 B |
| arm-cortex-m4 | rgb565 | 301.298 KB | 301.298 KB | +0 B |
| arm-cortex-m4 | minimal | 228.775 KB | 228.775 KB | +0 B |
| arm-cortex-m0plus | standard | 352.305 KB | 352.305 KB | +0 B |
| arm-cortex-m0plus | rgb565 | 311.644 KB | 311.644 KB | +0 B |
| arm-cortex-m0plus | minimal | 238.922 KB | 238.922 KB | +0 B |
| xtensa-esp32 | standard | 437.639 KB | 437.639 KB | +0 B |
| xtensa-esp32 | rgb565 | 387.404 KB | 387.404 KB | +0 B |
| xtensa-esp32 | minimal | 290.195 KB | 290.195 KB | +0 B |
| riscv32-esp32c3 | standard | 428.764 KB | 428.764 KB | +0 B |
| riscv32-esp32c3 | rgb565 | 382.111 KB | 382.111 KB | +0 B |
| riscv32-esp32c3 | minimal | 284.627 KB | 284.627 KB | +0 B |
Flash is text + data, RAM is data + bss.
Comparing 652979577 against 7cf49a06c.
Generated automatically
Contributor
There was a problem hiding this comment.
All reported issues were addressed across 2 files
Reply with feedback, questions, or to request a fix.
Re-trigger cubic
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Check if the fractional part of a float is near 0.5 (pixel boundary).
When using
VG_LITE_FILTER_POINT, the GPU samples at pixel centers. The initial sample coordinate c_step = 0.5 * step + translate. If the fractional part of translate is exactly 0.5, the sample lands on a pixel boundary where hardware rounding is ambiguous. Use a threshold around 0.5 to catch this case and force bilinear filtering for correct results.The threshold 0.1 means: frac in [0.4, 0.6] is considered "near half".
Values like 0.25 and 0.75 (frac distance to 0.5 = 0.25) are safe.
Notes
lv_conf_template.hrun lv_conf_internal_gen.py and update Kconfig.scripts/code-format.py(astyle v3.4.12needs to installed by runningcd scripts; ./install_astyle.sh) and follow the Code Conventions.