Hatch fills drew the full per-family pattern loop every frame regardless
of zoom. Two LOD tiers, neither of which existed:
* **Shader-side solid substitute.** When the densest line family's
spacing × scale projects to less than 2 px, individual lines blur
into a solid color and the per-fragment family loop is wasted ALU
(and produces moiré). Pass `world_per_pixel` through Uniforms; the
hatch fragment shader now returns `h.color` directly in that case.
* **CPU-side sub-pixel skip.** When the entire hatch AABB projects
to less than 2 px, drop the draw call. Adds `world_aabb` to
HatchGpu plus a per-frame `compute_hatch_lod` pass (mirrors the
existing scissor pass) and `hatch_skip_flags`. The render pass
skips flagged hatches.
Factored a shared `aabb_below_pixel` helper for future LOD callers.
The Uniforms layout is unchanged in size (96 B) — the trailing
`_pad: vec2<f32>` becomes `world_per_pixel: f32, _pad: f32`, so wgsl
modules that don't read the new field keep working with their old
`_pad: vec2<f32>` declarations.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Wire pass already uses `WireModel.vp_scissor` to clip viewport-projected
geometry to its paper-space viewport rect; hatch / image / wipeout
passes drew across the whole canvas. When a future paper-space path
projects fills/images into a viewport, they would bleed past the
viewport frame.
Add `vp_scissor: Option<[f32; 4]>` to HatchModel and ImageModel
(mirroring WireModel), forward it through HatchGpu / ImageGpu, and add
per-pass `compute_*_scissors` / `set_scissor_rect` machinery. Factored
the wire-side scissor-projection math into a shared `project_scissor`
helper. Field defaults to None at every constructor site — no producer
sets it yet, so this is plumbing for the paper-viewport fill projection
that Phase 1.4 unblocks without changing current visual output.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Block defns and top-level entities now know their world-space XY
AABB, and `tessellate_entity` / `expand_insert` early-out for anything
that doesn't intersect the camera's view rect. Tested commit-only on
time_open (correct subset emitted with default camera); GUI run
pending user verification.
Cache plumbing:
- `LocalWire` / `BlockDefn` carry `aabb_local` (post-build recursive
union over nested defns).
- `Scene::view_world_aabb()` derives a 25%-margined world rect from
the active camera + last-seen render aspect.
- `wire_cache` / `paper_sheet_cache` keys widen to
`(geometry_epoch, camera_generation)` so pan/zoom invalidates the
cached wire list — and pipeline `cached_epoch` widens to match
so the GPU re-uploads on view change.
- `XrefStatus::Unloaded` ack already shipped — orthogonal to this.
Also adds `examples/xref_states.rs` to dump per-block_record xref
metadata for diagnosing the is_loaded / unloaded distinction.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Remove MAX_SEGS_PER_WIRE = 100_000 and MAX_TOTAL_SEGS = 3_000_000.
These were stop-gap GPU-memory guards that mutated visible output:
single complex entities were cut mid-geometry, and entire wires past
the 3M total were dropped by arrival order with no spatial awareness.
Either is worse than honestly drawing whatever the file contains —
truncated arcs and missing road segments look like bugs to the user,
not "we ran out of GPU memory."
The matching "Temporary Workarounds" section in
RENDERING_OPTIMIZATION.md goes with them. Per-entity LOD and
viewport culling (Phase 1 + 3.2 in the same doc) remain the right
long-term answer; until they land, prefer correctness over a soft
crash on pathological files.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds two-layer protection against large entities crashing the renderer:
- MAX_SEGS_PER_WIRE (100K): per-entity hard cap, excess segments truncated
- MAX_TOTAL_SEGS (3M): scene-level budget in upload_wires, excess wires skipped
Both limits log a warning to stderr when triggered.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
PolyfaceMesh and PolygonMesh were wireframe-only. They now render with
the same flat-color solid fill as Face3D entities.
- Add fill_tris: Vec<[f32;3]> to TruckEntity and WireModel
- PolyfaceMesh: fan-triangulate each face into fill_tris
- PolygonMesh: triangulate each grid quad into fill_tris
- tessellate.rs Lines branch: apply world_offset to fill_tris
- Face3DGpu::from_wires: collect fill_tris from all wires in addition
to Face3D key_vertices quads
- upload_face3d: pass all_wires so fill_tris are batched into one draw call
Viewport-projected wires now clip at pixel level to their viewport boundary.
- WireModel.vp_scissor: Option<[f32; 4]> — paper-space bbox set on projection
- WireGpu.vp_scissor carries it through to the draw loop
- Pipeline.compute_wire_scissors() converts paper→NDC→pixel each frame
- render() calls set_scissor_rect() per wire, reset after viewport group
- INTEGRATION_GAPS.md: Viewport ⚠️ → ✅, all 51 gaps resolved
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- fix: polyline arc center sign was inverted (- → +) causing arcs to render on the
wrong side in LwPolyline and Polyline2D
- fix: store glyph strokes in local space with f64 origin so world_offset subtraction
uses f64 precision; prevents blocky text at large UTM coordinates
- fix: Face3D fill darkened to 45% so edge wires are visually distinct from fill
- fix: wire batch vertex buffer chunked to 256 MB GPU limit (was panic on large files)
- fix: gpu_face3d_edges changed from Option<WireGpu> to Vec<WireGpu> to support chunks
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
3DFACE entities previously created one WireGpu buffer per entity,
resulting in N separate draw calls per frame. For files with large
triangle meshes this was the dominant GPU CPU overhead.
New pipeline:
- face3d.wgsl: simple flat-color vertex + fragment shader (MVP only,
no quad expansion, no pattern distance computation)
- Face3DGpu: batches ALL Face3D entities into a single TriangleList
buffer (6 vertices per face, per-vertex color) — 1 draw call
- WireGpu::from_batch: merges N Face3D edge wire models into one
quad-expanded buffer for edge rendering — 1 draw call
- split_face3d_wires: separates face3d from other wires at Primitive
build time via document handle lookup (O(N), once per epoch)
Hit-testing via WireModel (CPU) is unaffected — face3d wires remain
in entity_wires_arc() and hit_test_wires() as before.
Before: N 3DFACE → N draw calls, N × 96 B/vertex WireVertex
After: N 3DFACE → 2 draw calls (fill + edges), 28 B/vertex fill
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Changed selection overlay pipeline from depth_compare=Greater to
depth_compare=Always so selected wires are redrawn at full brightness
after all other passes, making them visible regardless of what is in
front of them.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Selected wires are now visible through occluding objects. A second
wire_xray pipeline (depth_compare=Greater, depth_write_enabled=false)
renders 25%-alpha ghost copies of selected wires only where they are
behind the depth buffer, leaving fully-visible portions at full glow
brightness unchanged.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add two complementary render caches that make mouse navigation essentially
free regardless of file size:
Option A – Wire tessellation cache: entity_wires_arc() caches the
tessellated Vec<WireModel> keyed by geometry_epoch and returns an Arc,
so navigation frames (no preview wires) incur zero Vec clones and zero
tessellation work. Primitive::wires changed to Arc<Vec<WireModel>>.
Option B – GPU buffer cache: Pipeline stores cached_epoch; prepare()
skips all upload_wires/hatches/images/meshes calls when the epoch is
unchanged, writing only the 192-byte camera uniform instead.
geometry_epoch is a process-wide AtomicU64 incremented by every Scene
mutation (add/erase entity, selection, layer visibility, layout switch,
populate_*, grip, transform, copy, clear, preview wires). Using a global
counter rather than a per-scene counter prevents two tabs that happen to
have the same local bump count from sharing a cached_epoch value, which
was causing the first file's geometry to remain on screen after opening a
second file.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
MSAA and depth textures are now sized to the shader widget's clip
bounds instead of the full surface. The resolve writes to a matching
clip-sized intermediate texture; a blit pass (blit.wgsl, viewport
trick) then copies the result to the exact clip region of the surface
target, leaving all other widgets untouched.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
All main render pipelines (wire, hatch, image, mesh) now use 4×
multisampling. Drawing passes render to a dedicated MSAA color buffer;
a final empty resolve pass transfers the anti-aliased result to the
iced surface target. The ViewCube renders post-resolve at 1× to avoid
pipeline incompatibility. Depth buffer also upgraded to 4× MSAA.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Previously wipeout solid fills were part of the hatch pass (pass 1),
drawn before wires (pass 5), so all wires appeared on top of the
wipeout. This fix moves wipeout fills into a dedicated pass 6 that
runs after wires, so the background-colored polygon correctly covers
the lines/text beneath it — matching AutoCAD behaviour.
- Scene.wipeout_models() separated from synced_hatch_models()
- Primitive.wipeout_hatches populated via wipeout_models()
- Pipeline.gpu_wipeouts + upload_wipeouts() + pass 6 after wires
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
