Vec2Pix turns photos into editable vectors for pixel-precise AI edits

Vec2Pix introduces a hybrid workflow for image editing, converting source photos into vectorized representations that users can manipulate before reconstructing them back into photorealistic images. Once an image is decomposed into vector shapes, editors can stretch mountains, adjust object silhouettes, or reshape scene elements with exact control over geometry.

Beyond shape manipulation, Vec2Pix lets users recolor individual vector regions, reposition elements like roses, or add new vector objects such as sunglasses to the scene. The reconstructed outputs integrate these changes while preserving overall lighting and context, effectively combining traditional vector editing with AI-powered rendering.

The approach may be overkill for casual edits that modern text-prompt-based tools already handle well, but it offers a high‑precision option for designers who need tight control over composition. The project page notes that code is “coming soon”; however, the GitHub and data buttons are not yet active, leaving early adopters waiting for a public release.