# Canvas render probe — Godot 4.x Vita port, Phase 0 (neutral proof) Answers one question, without committing to path A or B and without touching Godot or VitaGL source: > **Can VitaGL compile and render a shader shaped like Godot's 2D canvas > material?** ## What it does Hand-writes the GLSL **ES 1.00** equivalent of what `drivers/gles3/shaders/canvas.glsl` does at its core — `mvp * vertex`, texture sample, modulate multiply — then: 1. compiles + links it through VitaGL (logging compile/link status), 2. uploads a known 2×2 texture `(255,128,64,255)`, 3. draws a fullscreen textured quad modulated by `(1, 0.5, 1, 1)`, 4. **reads back the center pixel** and checks it equals the expected `(255, 64, 64)` — proving the draw actually happened (no display capture needed). The shader is deliberately the *down-emit* of Godot 4's real one: `#version 300 es`→(none), `in/out`→`attribute/varying`, `layout(std140)` UBO→plain uniforms, `texture()`/frag-out→`texture2D()`/ `gl_FragColor`, `uvec4` integer attribs→dropped. ## Build & run ```bash export VITASDK=/usr/local/vitasdk cd platform/vita/tools/canvas_render_probe cmake -B build -S . && cmake --build build ``` Install `build/canvas_render_probe.vpk`, run it (shows the quad ~3s, then exits), and read: ``` ux0:/data/canvas_render_probe.txt ``` Requires VitaGL built with shader-compiler support and `libshacccg.suprx` installed/decrypted on the console. ## Reading the result - **`RESULT: PASS`** → VitaGL renders a Godot-shaped 2D shader. **Path B** (down-emit ES 1.00 from `drivers/gles3`) is viable — scope that fork next. - **`RESULT: FAIL (shader compile)`** → even the ES 1.00 down-emit is rejected; the shader-compiler story is worse than expected. - **`RESULT: FAIL` with "pixel is background"** → it compiled/linked but drew nothing; a GXM/VitaGL draw-state issue to dig into before investing. Pair this with `../es3_probe/` (which probes the raw ES3 API surface). Together they bound the renderer effort precisely.