image-ecommerce

image-ecommerce generates professional product photography for e-commerce listings, catalogs, and marketing on the Starchild platform. It covers white-background hero shots, lifestyle scenes, flat lay arrangements, detail and macro close-ups, packaging and unboxing shots, group/collection displays, scale references, seasonal themes, 360-degree views, comparison layouts, infographic-style feature callouts, and platform-optimized images for Amazon, Shopify, Taobao, Instagram, Xiaohongshu, Etsy, and eBay. Its core principle is to call the provided script rather than re-implement proxy or billing plumbing, and it clarifies when to use image-ecommerce versus related skills like image-edit, image-portrait, image-create, and image-tryon.

The main entry point is the product_photo() function (and product_photo_set() for a complete set). It supports three input modes: edit mode from a local file via product_path or a public HTTPS URL via product_url, and pure text-to-image mode when neither is provided (requiring a prompt). Parameters include style (default "hero"), background (default "white"), model ("nanopro" fast ~25s or "gpt" best quality ~150s), count (1-8), aspect_ratio (1:1, 3:4, 4:3, 9:16, 16:9), and platform, which overrides style, background, and aspect_ratio with platform-optimized values unless explicitly set. Photography style presets include hero, lifestyle, flat_lay, detail, packaging, group, scale, 360_view, comparison, infographic, and seasonal variants (seasonal_spring, seasonal_summer, seasonal_autumn, seasonal_winter). product_photo_set() generates 7 images (hero, lifestyle, detail, scale, alternate angle, packaging, flat lay). The script reads local files, base64-encodes them, and sends them to fal.ai as a data URI, and always downloads results on success to output/images/.

The skill is aimed at sellers and marketers producing listing and campaign imagery. It includes an important execution note: the code blocks are Python, not shell commands, and when calling via the bash tool you should use python3 - <<'EOF' with from exports import rather than exec(open(...)) inside python3 -c. For delivery, it warns never to hand the user the raw fal.media URL because fal serves files with restrictive CSP headers; instead use each image's downloaded local_path in output/images/, embed inline on Web, or send via send_to_telegram / send_to_wechat on those channels.