Class RenderContextInternal

The View we are currently rendering.

Whether backfaces are currently enabled during the current frame.

Statistic that counts how many times gl.bindTexture() has been called so far within the current frame.

Which texture is currently bound to each texture unit, indexed by unit.

Lets the draw path skip a redundant gl.bindTexture when the unit already holds the texture about to be bound — saving the repeated binds of frame-wide textures (camera matrix, IBL maps) across a bin's batches.

Only valid within one uninterrupted batch-draw sequence. Cleared by resetTextureBindings at the start of every such sequence (each draw bin, pick, snap, and the SAO/shadow/cap prep passes), because sub-pipelines and atlas mipmap refreshes bind textures outside this tracking.

Whether debugging is enabled.

The vertex winding order for what we currently consider to be a backface during current frame: true == "cw", false == "ccw".

The WebGL rendering context.

IBL Layer-2 prefiltered samplers — populated by the IBL prefilter pipeline before the BRDF passes run. The smooth-shaded technique variants bind these for the split-sum specular IBL approximation:

• iblIrradianceCubemap: cosine-convolved cubemap for diffuse IBL
• iblPrefilteredCubemap: GGX-prefiltered cubemap with mip chain
• iblBRDFLUT: shared 2D (NdotV, roughness) → (scale, bias) LUT

iblMaxSpecularMipLevel is the highest valid mip on the prefiltered cubemap, used to scale roughness into the lod range.

All four are null outside the BRDF binding window — the renderer sets them in RenderManager._renderScene and clears them in reset.

View-to-world rotation (3×3, transposed view matrix's upper-left) uploaded as a mat3 so the BRDF shader can transform view-space normals/reflection vectors into world space for cubemap sampling. Populated alongside the cubemaps. Float32 column-major.

ID of the last WebGLProgram that was bound during the current frame. Used to avoid redundant program binds.

Render pass of the last _bind call. Used together with lastProgramId to detect when pass-dependent uniforms (e.g. silhouetteColor) need re-uploading even though the same program is still bound.

The current line width.

The memory configuration for the WebGLRenderer.

Whether to render a quality representation for triangle surfaces.

When false, we'll render them with a fast vertex-shaded Gouraud-shaded representation, which is great for zillions of objects.

When true, we'll render them at a better visual quality, using smooth, per-fragment shading and a more realistic lighting model.

TODO

Whether the WebGLRenderer is currently picking invisible objects.

The 4x4 orthographic projection transform matrix the WebGLRenderer is currently using when rendering a ray-pick.

The 4x4 viewing transform matrix the WebGLRenderer is currently using when rendering a ray-pick.

This sets the viewpoint to look along the ray, when picking with a ray.

Distance to the far clipping plane when rendering depth fragments for GPU-accelerated 3D picking.

Distance to the near clipping plane when rendering depth fragments for GPU-accelerated 3D picking.

Whether we are currently picking with a ray.

The RenderInspector for logging draw calls.

The occlusion rendering texture.

See sceneRenderWidth.

Width of the FBO the scene is currently being rendered into. Equal to drawingBufferWidth when Tonemap.renderScale is 1; larger when supersampling is on. Post-process passes (tonemap, FXAA) use drawingBufferWidth — they sample this bigger scene texture down to canvas size on the way out.

Number of shadow cascades active this frame, in [1, MAX_SHADOW_CASCADES].

View-space |z| boundaries between cascades. Entry i is the far edge of cascade i (equivalently the near edge of cascade i+1). Only the first shadowCascadeCount - 1 entries are meaningful.

Light direction in camera-view space (xyz). Populated alongside shadowLightViewProjMatrices and used by shadow-aware draw techniques for slope-scaled bias computation.

Concatenated array of up to MAX_SHADOW_CASCADES mat4s (16 floats each) — shadowLightViewProjMatrices.subarray(i*16, (i+1)*16) is the camera-view-space → cascade-i light-clip matrix.

Light view-projection matrix for the cascade currently being rendered, used by the shadow-depth technique which only ever processes one cascade per draw pass. Set by ShadowPipeline before each cascade's depth dispatch; ignored by shadow-aware color techniques (they read the array form, shadowLightViewProjMatrices).

One shadow-map depth texture per cascade (indices 0..shadowCascadeCount-1 are populated). Unused slots alias to cascade 0's texture so every sampler in shadow-aware shaders has a valid binding. Null array entries indicate shadows aren't in effect this frame.

Snap framebuffer size in pixels — the snap shaders use the ratio (drawingBufferSize / snapBufferSize) to convert a canvas-pixel offset into the matching offset on the small snap FBO, so the cursor's surroundings land flush against its edges.

Cursor position in clip space (range [-1, 1] on each axis), set by the SnapManager before binding a snap technique. The snap pass uses this to centre its viewport-remap helper on the cursor — same shape as pickClipPos but kept separate so pick and snap can run side by side without clobbering each other.

The next available texture unit to bind a texture to.

The Viewer.

The HTML canvas element used for WebGL rendering.