Rendering Backends

VeloxKit ships with three rendering backends. The default auto mode picks the best one for the machine at startup — no configuration needed for most apps.

The memory story

The backend choice is the single biggest factor in your app's RAM footprint:

TinySkia at 97 MB is what makes VeloxKit's headline number possible. It is not a fallback or a limited mode — it is the default for most hardware and produces visually identical output to Vello at typical UI complexity.

auto (default)

auto queries the GPU adapter at startup and picks the appropriate backend:

No GPU / software adapter (WARP, llvmpipe)  →  TinySkia
Integrated GPU (Intel UHD, Apple M-series)   →  TinySkia
Intel Arc discrete GPU                        →  FemtoVG
NVIDIA / AMD discrete GPU                     →  Vello

You don't need to set anything — this is the default when renderMode is omitted.

// veloxkit.config.ts — auto is the default, this is explicit
window: {
  renderMode: 'auto'
}

The selected backend and adapter are logged at startup:

[velox] renderMode=auto → skia (Intel(R) UHD Graphics 770, IntegratedGpu)
[velox] renderMode=auto → vello (NVIDIA GeForce RTX 4080, DiscreteGpu)

skia — TinySkia (CPU rasterizer)

Pure CPU rendering via tiny-skia (opens in a new tab). Zero GPU memory — no wgpu device allocation, no buffer pools, no compute pipelines.

window: { renderMode: 'skia' }

• ~97 MB RSS on hello-world
• No GPU required
• Fastest cold start (no shader compilation)
• Full visual parity with Vello at typical UI density
• Best choice for: productivity apps, low-end hardware, VMs, headless/CI

femtovg — FemtoVG (OpenGL)

GPU-accelerated rendering via FemtoVG (opens in a new tab) — a lightweight OpenGL triangle tessellation renderer. Lighter than Vello because it avoids wgpu's compute pipeline overhead.

window: { renderMode: 'femtovg' }

• ~103 MB hello-world, ~153 MB on a richer app (Calculator)
• GPU-accelerated anti-aliasing and smooth animations
• Works on OpenGL ES — broader hardware support than DX12/Vulkan
• Best choice for: Intel Arc systems, mid-range hardware, smooth animation-heavy apps

gpu / vello — Vello (wgpu compute)

GPU compute rendering via wgpu (opens in a new tab) + Vello (opens in a new tab). The highest quality backend — analytically anti-aliased gradients, shadows, and glyphs rendered directly on the GPU.

window: { renderMode: 'gpu' }   // explicit Vello GPU path

• ~285–328 MB RSS (wgpu device + resource pool + V8)
• Best visual quality for gradients, shadows, high-density content
• Required for 3D Canvas
• Requires DX12 (Windows), Metal (macOS), or Vulkan (Linux)
• Best choice for: visual showcase apps, games, apps with heavy 3D content

renderMode: 'cpu' uses Vello's Cranelift CPU path (~35–40 MB cheaper than GPU Vello, but heavier than TinySkia):

window: { renderMode: 'cpu' }   // Vello on CPU — still ~250 MB

Choosing explicitly

Force skia at runtime without changing config:

VELOX_CPU_RENDER=1 veloxkit dev

Fallback chain

If a backend fails to initialise (driver crash, missing GPU), VeloxKit falls back:

vello   → fail → femtovg → fail → skia
femtovg → fail → skia
skia    → fail → ERROR

The fallback is logged. Production apps should test on their target hardware to avoid silent fallbacks.

See also

• veloxkit.config.ts reference — window.renderMode field
• Renderer & engine selection guide — full heuristics matrix and persona recommendations