What this changes
PDF pages are not editable DOM documents. Text is often part of a rendered page surface, so moving or editing text requires cleanup, replacement drawing, async tile handling and preview stability without erasing unrelated content.
No results found.
The render graph is the browser-side planning layer that answers what needs to be drawn now for the current document preview and local edits.

From PDF bytes to final composite, all inside the browser preview path.
PDF pages are not editable DOM documents. Text is often part of a rendered page surface, so moving or editing text requires cleanup, replacement drawing, async tile handling and preview stability without erasing unrelated content.
Browser Core turns each local operation into structured visual responsibilities before anything is drawn: source cleanup, replacement content, selection/debug overlays and destination cleanup only when it is safe. Those responsibilities become command buffers that are resolved into a final composite preview.
Separates what the user did from what the browser must draw
Keeps source cleanup separate from replacement text so moves do not erase unrelated PDF content
Lets async textless tiles resolve without forcing destructive fallback masks into the preview
Supports replay after refresh by rebuilding the same visual responsibilities from local operations
Preview planning layer
Inspectable command buffers
Dirty-region awareness
Safe final composite filtering
Backend export remains separate