Canvas overlay editor

Three layers

The editor ships with three pre-defined text layers:

• Title, large, top-third by default, display font.
• Subtitle, smaller, optional, hidden by default.
• Byline, bottom of the canvas, sans-serif.

Per-layer controls

• Text, the string to render.
• Font family, Display, Serif, Sans, or Mono. Each maps to a curated stack so your cover looks consistent across devices.
• Font size, pixel value. Defaults are sensible; tune per layer.
• Color, hex picker.
• Alignment, left, center, right.
• Drop shadow, on/off. Useful when text sits over a busy area of the artwork.
• Visibility, toggle off to omit a layer entirely (e.g. no subtitle).
• Position, drag any text layer on the canvas to reposition. Position stores as 0..1 ratios so it survives canvas resizing.

Save composites in-place

When you save, Pendraic flattens the canvas (artwork plus your text layers) into a single image and that becomes the new cover. The bookshelf shows the typed version immediately. Layers are baked in at save time, so what you see on the shelf is exactly what you saw in the editor.

You can re-open the editor against a saved cover, but the previous layer state is not preserved, you start from the defaults. (The editor knows the writer's typography choices via the project metadata, so the title and byline come back populated.)

Why a separate editor

The diffusion model that paints the artwork can't render legible book titles. Diffusion models butcher letterforms 9 times out of 10. Pendraic addresses this structurally, the AI prompt never carries the title or byline; the canvas editor handles type entirely. Two layers, one clean cover.