Stitch Your Pets — Internal Pattern Generator

The AI's only job is to look at the photos, pick out this cat's own 5-or-fewer real coat colors, and describe where each goes — it never touches stitch counts or construction. Photos are sent securely to our AI service for analysis and are not stored.

Structured result from the AI. You can hand-edit this JSON before applying it to the stitch maps — nothing here touches stitch counts, only which colors get suggested where.

Once you've run an analysis, tell it what to fix in plain language — it keeps the photos in mind and revises the marking report above. Examples: "the belly is more cream than white," "there's a dark patch on the left ear you missed," "make the tail solid ginger, no tip." Each reply updates the Marking Report textarea above — click "Paint Stitch Maps From Report" again to push the change into the grids.

General swatch library. Edit to match what you actually stock. The AI reuses one of these ids whenever it's a genuinely close match to what it sees in the photos.

Whenever none of the yarn palette swatches above are a close enough match, the AI reads this cat's actual photos and adds up to 5 new colors of its own (exact hex values it read off the coat) here, below the palette. These are added alongside the palette above, not instead of it — painting, the yarn shopping list, the PDF, and the generated preview photo can all draw on either list. Filled in automatically by Analyze Photos; edit name/hex here (e.g. to note which yarn skein you're using for it) if you want to nudge one.

Every stitch already exists in the base pattern — the AI (and you) only ever assign a color to a square. Click a swatch below, then click, drag-paint, or bucket-fill squares. Internally every square has a permanent ID like HEAD-R08-S14 (you never need to see these — it's just how the software keeps the pattern and the picture in sync).

Approximate preview only, built from the same colored squares you just painted. Each round is a true ring: its circumference comes from that round's real stitch count, so the pattern's own numbers define the shape — no separate 3D model to keep in sync. Self-contained, no internet needed to view. Drag to rotate, scroll to zoom.

Pulled from the written base pattern — these never change and aren't affected by coloring, but they're easy to lose track of, so they're included on the printed pattern below.

Sends only three kinds of images to our AI service: the approved colored stitch maps, four fixed 3D-render views (front/left/right/back), and a bundled set of real gray finished crochet-cat reference photos (shown below — front/side/back/top/close-ups, always sent automatically, nothing to upload). Your customer's real cat photos are never sent here — only what you've already approved through Steps 2–3. Coat colors and markings come only from the stitch maps and 3D render, plus an explicit hex-code color list included as text; the gray photos are used only to match crochet texture, yarn look, stuffed shape, and construction (ears, tail, safety eyes, embroidered nose/mouth, whiskers) — their gray color itself is not copied unless your stitch maps are actually gray.

This creates a customer-ready PDF by placing each custom stitch map and its color instructions beside the matching section of the original base pattern. It also swaps the cover page's cat photo and name for the approved generated front photo and customer name above (keeping the rest of the cover design as-is), then adds the finished-cat preview (all 4 angles) right after the cover, and adds the yarn colors page right after the Materials page. The base pattern itself is bundled with this site (base-pattern.pdf and cover-badge.png, sitting in the same folder) — nothing to upload.