Social Carousel Template

Use this template when the artifact is a curated multi-card anthology — 6–10 images that share an editorial identity and want to be experienced in rhythm, not as a list.

Distinct from digital-eguide/social-carousel.html (12-frame field-notes guide with 50/50 photo + metadata). This one is gallery-paced, no per-card metadata table, sculptural geometric masks.

Use When

• The user wants to publish a series / anthology / 选辑 / collection — photography, art direction, branding moments, mood reel.
• Each card carries one image + a short serif title + a volume number (第 0X 辑).
• The viewing context is browsing / pause / browse — not step-by-step instruction.
• The output is a fixed-image set (HTML-to-image) or an embedded auto-scrolling strip.

Avoid When

• The content is a tutorial or numbered steps — use digital-eguide/social-carousel.html (it has progress + per-step metadata) or social-card-swiss/.
• The series is < 4 entries — make individual cards instead.
• The content is text-heavy (quotes, body copy) — gallery rhythm hides text.
• The platform expects discrete swipeable posts (XHS multi-image post) — this is one scrolling strip, not 9 separate posts.

Required Reading

Read this SKILL.md first. Do not read scroll-gallery.html by default.

Read scroll-gallery.html only when you need: the geometric-mask shapes (arch / oval / half-dome / diagonal), the auto-scroll easing values, the seamless-loop double-render trick, or the hover-slowdown behavior. Extract patterns; do not copy the placeholder Songti titles, photo URLs, or class names wholesale.

Design Strategy

Commit to the anthology spine before designing cards:

• one cover frame role per card (full-bleed, arched, oval, diagonal, masked half-dome)
• a single rotating ornamental SVG mark style (don't mix flower + grid + arrow)
• consistent corner metadata (volume + tag + year, picked once)
• one title font (Songti SC), one meta font (PingFang SC), no third voice

If the cards don't share a visual system, the auto-scroll exposes the inconsistency.

Layout Bias

Prefer:

• 2:3 or 7:9 portrait per card; mix sparingly, never randomly
• restrained warm palette (creams / terracottas / sage) over white or near-black
• one geometric mask per card, rotated through the set
• ornamental marks small and consistent, in a single accent ink
• continuous slow auto-scroll, pause on hover, infinite loop

Avoid:

• emoji or 3D plastic stickers in corner metadata
• mixing portrait + landscape card aspects randomly
• adding step numbers (this is not a tutorial)
• decorative gradients behind images — the mask is the decoration
• 黑金 / neon palettes — the template's identity is editorial / warm

Platform / Size

A horizontal strip of vertical cards (~ 360×540 each). Designed for embedding on a long page or capturing as a wide image. If the user wants discrete XHS posts, export each card individually.