菜单
When adding STORAGE_BINDING to compact texture formats in vendored wgpu-types, both |s_ro_wo| flags and |TextureUsages::STORAGE_BINDING| in allowed_usages are needed — the flags element must include s_ro_wo or the HAL receives zero storage access bits
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when working on 3D-specific rendering — deferred chain, G-buffer, clustered shading, Stratum GI, ReGIR, SVGF/A-SVGF, TAA, shadow techniques, depth/HZB, anything 3D-specific. ALWAYS load titan-rendering-core first; this skill assumes core conventions. Loads Titan-specific overrides for realtime-rendering. NOT for 2D rendering or editor panels.
Use when diagnosing a Titan renderer artifact — flicker, ghosting, fireflies, banding, light leaks, seams, boil, NaN/Inf, aliasing, smear/trails, dark voids, shadow artifacts, GI instability, denoiser/TAA issues. The frame-isolation-first debugging method: capture 60Hz, diff frames / A-B vs classical, consult literature, isolate via systematic-debugging, propose (and prove) a fix, record empirically. NOT for landing the production fix (hand to titan-rendering-3d-agent), 2D rendering, or editor panels.
Use when working in titan-rendering-frame-graph, titan-rendering-core, or titan-rendering-post-process-core — frame graph DAG, RenderTarget, DevicePool, PostProcessHook, CameraFrameData, HDR intermediate, tonemap, color management. Foundational layer shared between current 3D and future 2D rendering. Loads Titan-specific overrides for realtime-rendering. NOT for 3D-specific concerns (deferred chain, GI, shadows) — load titan-rendering-3d after this.
When a Titan bench tool using Renderer3d lacks on_resize → renderer.resize(), depth_prepass_view and HZB occlusion resources are never created — DepthPrepass, HzbBuild, and SceneCull passes silently skip, and taps produce no output
When adding a camera UBO shared between Rust and WGSL, verify field-by-field that the CPU layout matches the WGSL struct — especially packed vec4 fields where reordering silently breaks the shader
When building a depth prepass for reverse-Z HZB occlusion culling, use Greater compare (not Always) — the closest fragment must win so the HZB stores the nearest occluding surface, not the farthest
| name | wgpu-format-patching |
| description | When adding STORAGE_BINDING to compact texture formats in vendored wgpu-types, both |s_ro_wo| flags and |TextureUsages::STORAGE_BINDING| in allowed_usages are needed — the flags element must include s_ro_wo or the HAL receives zero storage access bits |
| source | auto-skill |
| extracted_at | 2026-06-15T03:49:26.570Z |
When modifying vendored wgpu-types guaranteed_format_features() to add
STORAGE_BINDING to compact formats, both elements of the returned tuple
must be updated:
flags (first element) — must include s_ro_wo (= STORAGE_READ_ONLY | STORAGE_WRITE_ONLY). Without this, conv.rs:100-115 gates each HAL storage
access bit on the corresponding flag, producing zero storage access bits
despite allowed_usages passing validation.
allowed_usages (second element) — must include
TextureUsages::STORAGE_BINDING.
Why the split exists: allowed_usages controls validation at texture-creation
time. flags controls what the HAL actually enables. The pattern in every
storage-capable format (Rgba8Unorm, Rgba16Float, etc.) uses both. They were
decoupled in the BGRA8UNORM_STORAGE feature because write-only storage on BGRA8
is sometimes available when read-write is not.
Correct pattern per format:
// R8Unorm
Self::R8Unorm => (msaa_resolve | s_ro_wo, attachment | TextureUsages::STORAGE_BINDING),
// R16Float
Self::R16Float => (msaa_resolve | s_ro_wo, attachment | TextureUsages::STORAGE_BINDING),
// Rg16Float
Self::Rg16Float => (msaa_resolve | s_ro_wo, attachment | TextureUsages::STORAGE_BINDING),
// Rg11b10Ufloat (two branches — feature-gated)
let (rg11b10f_f, rg11b10f_u) =
if device_features.contains(Features::RG11B10UFLOAT_RENDERABLE) {
(msaa_resolve | s_ro_wo, attachment | TextureUsages::STORAGE_BINDING)
} else {
(msaa | s_ro_wo, basic | TextureUsages::STORAGE_BINDING)
};
Vendoring approach: cargo vendor + [patch.crates-io] is sufficient —
the native backends (Vulkan/DX12/Metal) query the driver for format
capabilities directly; only guaranteed_format_features() in wgpu-types
needs changes. The wgpu-hal backend tables do NOT need modification.
Verification: cargo check -p titan-rendering-3d — if the vendored crates
compile, the format changes are syntactically correct. The runtime probe
adapter.get_texture_format_features(R16Float).allowed_usages.contains(STORAGE_BINDING)
confirms the formats are exposed at device creation time.