make-tape-collage

Invoke the skill with $make-tape-collage and attach your photo, asking for a tape-collage transformation. The subject is rebuilt from roughly 10-18 coherent tape pieces on a 3:4 warm-white journal page, and the source photo does not appear in the output unless you request preservation.

Explicitly ask to preserve, keep, or show the original photo when invoking the skill. It generates only a transparent tape motif, then uses scripts/compose_direct_split.py to mount the unchanged photo pixels beside a paper panel with torn edges, shadows, and a typewriter caption.

The compositor script requires Python and Pillow for image composition, paper texture synthesis, and typography. Image generation itself uses the built-in image generation tool, so Pillow is only needed when running scripts/compose_direct_split.py.

Facial identity is not preserved in tape-only transformations, and complex scenes are intentionally simplified to one to three identification anchors. The style favors soft structural distillation with a limited source-derived palette rather than literal reproduction.

Generative full-composite edits can leak paper seams, stains, or color casts and alter photo pixels. The deterministic compositor guarantees unchanged source pixels, one continuous synthesized paper sheet, exact crop limits, and precise caption placement.

Yes, exact user-supplied wording overrides the default one-to-three-word English summary title. Text is rendered deterministically in faded typewriter lettering at 26-32 pt equivalent, and an explicit no-text request removes it entirely.