image-ecommerce

image-ecommerce

Use this skill for all e-commerce product photography requests on Starchild.

Covers: white-background hero shots, lifestyle product scenes, flat lay arrangements, detail/macro close-ups, packaging/unboxing shots, group/collection displays, scale reference images, seasonal themes (spring/summer/autumn/winter), 360-degree views, comparison layouts, infographic-style feature callouts, and platform-optimized images for Amazon, Shopify, Taobao, Instagram, Xiaohongshu, Etsy, eBay.

Core principle: call the provided script. Do not re-implement proxy/billing plumbing.

When to use image-ecommerce vs other image skills:

• image-ecommerce → user wants PRODUCT PHOTOS for e-commerce, catalogs, or marketing
• image-edit → user wants to EDIT or TRANSFORM an existing image (not product-specific)
• image-portrait → user wants a portrait with their face/identity preserved
• image-create → user wants to CREATE something from text (not product photography)
• image-tryon → user wants to try on clothing/accessories on a person

---

1. Quick start — single product photo (most common)

⚠️ Execution context — read this first. The code blocks below are Python, not shell commands. Starchild's bash tool runs /bin/bash -c, which cannot parse exec(open(...)) — pasting them directly into a bash command will fail with syntax error near unexpected token 'open'. Also, exec(open(...)) inside python3 -c fails with NameError: __file__ because the script uses __file__ for path resolution.

Use python3 - <<'EOF' with from exports import when calling via the bash tool:

python3 - <<'EOF'
import sys
sys.path.insert(0, "skills/image-ecommerce")
from exports import product_photo
result = product_photo(
    product_path="uploads/product.jpg",
    style="hero",
    background="white",
)
print(result)
EOF

The heredoc (<<'EOF') preserves all quotes and newlines — no escaping needed.

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    product_path="uploads/product.jpg",
    style="hero",
    background="white",
)
# result -> {"success": True, "images": [{"local_path": "output/images/..."}], ...}

The script reads the local file, base64-encodes it, and sends it to fal.ai as a data URI — no manual URL publishing needed.

2. Quick start — public URL

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    product_url="https://example.com/product.jpg",
    style="lifestyle",
    background="natural",
)

3. Quick start — text-to-image (no product photo)

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    prompt="premium wireless bluetooth headphones, matte black finish, over-ear design",
    style="hero",
    background="white",
)

When no product_path or product_url is provided, the script uses the text-to-image endpoint (no /edit suffix). A prompt describing the product is required in this mode.

4. Quick start — platform-optimized

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    product_path="uploads/product.jpg",
    platform="amazon",
)
# Automatically applies: style=hero, background=white, aspect_ratio=1:1

5. Quick start — complete product image set

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo_set(
    product_path="uploads/product.jpg",
    prompt="premium leather wallet",
    platform="amazon",
)
# Generates 7 images: hero, lifestyle, detail, scale, alternate angle, packaging, flat lay

Delivering the result to the user — IMPORTANT

Never hand the user the raw fal.media URL. fal serves files with restrictive CSP headers. The only reliable delivery path is the already-downloaded local file:

1. Use each image's local_path (e.g. output/images/xxx.png) — the script always downloads on success.
2. Tell the user the files are saved to output/images/ and viewable in the workspace file panel.
3. On Web channel, embed inline so the user can preview in chat:

   ![product](output/images/<filename>.png)
   

4. On Telegram / WeChat: send via send_to_telegram(file_path="output/images/...", message_type="image") or send_to_wechat(file_path="output/images/...", message_type="image").

---

6. Parameters — product_photo()

Parameter	Required	Default	Description
product_path	no	—	Local workspace file path to the product image
product_url	no	—	Public HTTPS URL of the product image
prompt	no	—	Custom prompt describing the product or desired photo
style	no	"hero"	Photography style preset (see §7)
background	no	"white"	Background type (see §8)
model	no	"nanopro"	Model: "nanopro" (fast ~25s) or "gpt" (best quality ~150s)
count	no	1	Number of images to generate (1–8)
aspect_ratio	no	"1:1"	Output ratio: 1:1, 3:4, 4:3, 9:16, 16:9
platform	no	—	Platform preset: amazon, shopify, taobao, instagram, xiaohongshu, etsy, ebay

Image input rules:

• Provide product_path OR product_url for edit mode (transform existing product photo).
• If both are given, product_path takes priority.
• Omit both for pure text-to-image generation (requires prompt).

Prompt priority: prompt + style/background (enhanced) > style + background templates.

Platform preset: When platform is set, it overrides default style, background, and aspect_ratio with platform-optimized values — unless you explicitly set them.

---

7. Photography styles

Core product shots

Style	Key	Best for
Hero shot	hero	Primary listing image, magazine ads, main product display
Lifestyle	lifestyle	Product in use, editorial, social media
Flat lay	flat_lay	Instagram, top-down arrangement, catalog
Detail close-up	detail	Material quality, texture, craftsmanship
Packaging	packaging	Unboxing experience, brand packaging
Group/collection	group	Multiple products, variants, bundles
Scale reference	scale	Size comparison, product in hand

Marketing & informational

Style	Key	Best for
360° view	360_view	Multi-angle showcase, turntable display
Comparison	comparison	Side-by-side, before/after, feature highlight
Infographic	infographic	Feature callouts, specs, dimensions

Seasonal campaigns

Style	Key	Best for
Spring	seasonal_spring	Cherry blossoms, fresh green, pastel
Summer	seasonal_summer	Beach, sunshine, tropical, vacation
Autumn	seasonal_autumn	Fall leaves, golden tones, harvest
Winter	seasonal_winter	Snow, holiday, festive, cozy

---

8. Background types

Background	Key	Best for
Pure white	white	Amazon, e-commerce standard, marketplace listings
Gradient	gradient	Hero shots, premium feel, modern
Studio	studio	Professional catalog, controlled lighting
Natural	natural	Outdoor products, organic brands
Lifestyle	lifestyle	Home/office context, in-use scenarios
Colored	colored	Brand-matching, vibrant marketing
Textured	textured	Luxury products, marble/wood surface
Transparent	transparent	Product cutout, PNG for design use

---

9. Platform presets

Platform	Aspect Ratio	Background	Style	Key Requirements
Amazon	1:1	white	hero	Pure white bg (RGB 255,255,255), product fills 85%+, no props/text/watermarks, min 1000px (1600px+ for zoom)
Shopify	1:1	white	hero	Square format, consistent catalog style, 2048x2048 recommended
Taobao	1:1	white	hero	800x800 minimum, white bg for main image
Instagram	1:1	lifestyle	lifestyle	1080x1080 feed, lifestyle context, visually appealing
Xiaohongshu	3:4	lifestyle	flat_lay	1080x1440 vertical, aesthetic flat lay, text overlay space
Etsy	4:3	natural	lifestyle	Handmade/artisan feel, natural backgrounds
eBay	1:1	white	hero	White background, clear product view, 1600px min for zoom

---

10. Model selection guide

Model	Key	Speed	Quality	Best for
NanoPro	nanopro	~25s	Good	Default for all requests. Fast iteration.
GPT Image 2	gpt	~150s	Best	When user explicitly asks for "highest quality" or "best quality". Complex scenes.

Decision rules:

1. Default: always use nanopro unless the user explicitly requests higher quality.
2. Use gpt when: user says "highest quality", "best quality", "premium", or the scene is very complex with many specific details.
3. Use nanopro when: user wants fast results, is iterating on styles, or generating multiple images.

# Default (fast)
result = product_photo(product_path="product.jpg", style="hero")

# High quality (user requested)
result = product_photo(product_path="product.jpg", style="hero", model="gpt")

---

11. Intent recognition guide

Use this table to map user requests to the correct style + background:

Product listing images

User says	Style	Background	Notes
"product photo", "listing image", "主图"	hero	white	Default e-commerce
"Amazon listing", "亚马逊主图"	hero	white	Use platform="amazon"
"Shopify product", "独立站产品图"	hero	white	Use platform="shopify"
"淘宝主图", "天猫主图"	hero	white	Use platform="taobao"
"white background", "白底图"	hero	white	Standard packshot
"product on white", "纯白背景"	hero	white	Amazon-style

Lifestyle & context

User says	Style	Background	Notes
"lifestyle photo", "场景图"	lifestyle	lifestyle	Product in context
"product in use", "使用场景"	lifestyle	lifestyle	Show product being used
"flat lay", "俯拍", "平铺"	flat_lay	textured	Top-down arrangement
"Instagram product", "小红书产品"	flat_lay	lifestyle	Social media optimized

Detail & technical

User says	Style	Background	Notes
"close-up", "detail shot", "细节图"	detail	studio	Macro/texture
"packaging", "包装图", "开箱"	packaging	studio	Box + product
"size comparison", "尺寸对比"	scale	studio	With reference object
"multiple products", "组合图"	group	white	Collection display
"360 view", "多角度"	360_view	white	Turntable style
"comparison", "对比图"	comparison	white	Side by side
"infographic", "功能标注"	infographic	white	Feature callouts

Seasonal & campaign

User says	Style	Background	Notes
"spring campaign", "春季"	seasonal_spring	auto	Cherry blossoms, pastel
"summer sale", "夏季"	seasonal_summer	auto	Beach, tropical
"autumn/fall", "秋季"	seasonal_autumn	auto	Golden leaves, warm
"winter/holiday", "冬季", "圣诞"	seasonal_winter	auto	Snow, festive

Complete product set

User says	Function	Notes
"complete set", "全套产品图", "listing images"	product_photo_set()	7 images covering all angles
"Amazon listing set", "亚马逊全套"	product_photo_set(platform="amazon")	Platform-optimized set

---

12. Usage examples by scenario

Amazon listing — white background hero shot

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    product_path="uploads/headphones.jpg",
    platform="amazon",
)

Lifestyle product photo

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    product_path="uploads/coffee_mug.jpg",
    style="lifestyle",
    background="lifestyle",
    prompt="premium coffee mug on rustic wooden table beside an open book, morning sunlight",
)

Product detail close-up

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    product_path="uploads/leather_bag.jpg",
    style="detail",
    background="studio",
    prompt="extreme close-up of leather stitching and grain texture",
)

Seasonal campaign — winter holiday

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    product_path="uploads/candle.jpg",
    style="seasonal_winter",
    prompt="luxury scented candle in cozy holiday setting with pine branches and warm glow",
)

Text-to-image — generate product from description

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    prompt="sleek minimalist smartwatch with black silicone band and OLED display showing time",
    style="hero",
    background="gradient",
    model="gpt",
)

Flat lay for Instagram / Xiaohongshu

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    product_path="uploads/skincare_set.jpg",
    style="flat_lay",
    background="textured",
    platform="xiaohongshu",
)

Multiple images — batch generation

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo(
    product_path="uploads/sneakers.jpg",
    style="hero",
    background="white",
    count=4,
)
# Generates 4 variations of the hero shot

Complete product image set

exec(open('skills/image-ecommerce/product_photo.py').read())
result = product_photo_set(
    product_path="uploads/wallet.jpg",
    prompt="premium leather bifold wallet",
    platform="amazon",
)
# result -> {"success": True, "sets": [...], "total_images": 7, ...}
# Generates: hero, lifestyle, detail, scale, alternate angle, packaging, flat lay

---

13. Prompt engineering best practices

The product photography prompt structure

Every effective product photo prompt should include these elements:

[product description], [photography style], [lighting], [background/surface], [composition], [quality modifiers]

Key principles (derived from product-photography, eachlabs-product-visuals, image-create skills)

1. Product preservation is critical — when editing an existing product image:

   • Always emphasize "keep the product exactly as it is"
   • Preserve shape, color, branding, and details
   • Only change the background/context/lighting

2. Lighting specificity — always specify lighting type:

   • Studio: "soft diffused studio lighting", "even lighting with no shadows"
   • Dramatic: "dramatic rim lighting", "edge light for premium feel"
   • Natural: "natural window light", "golden hour warm light"
   • Flat: "flat even lighting" (for e-commerce white background)

3. Background precision — vague backgrounds produce poor results:

   • ❌ "nice background"
   • ✅ "pure white background #FFFFFF, no shadows"
   • ✅ "rustic wooden table with morning sunlight"
   • ✅ "soft gradient from white to light grey"

4. Composition rules (from product-photography skill):

   • Hero shot: product fills 80% of frame, slight 15-30° angle
   • Packshot (Amazon): product dead center, fills 85%+
   • Flat lay: bird's eye view, organized arrangement
   • Group: odd numbers (3 or 5), triangle composition

5. Shadow types matter:

   • No shadow: Amazon/e-commerce requirements
   • Contact shadow: grounded but clean
   • Drop shadow: adds depth, professional
   • Reflection: tech, luxury, premium feel

6. Material and texture — for detail shots, specify:

   • "visible leather grain and stitching"
   • "brushed metal finish with subtle reflections"
   • "soft fabric texture, thread detail visible"

7. Platform compliance — when targeting a specific platform:

   • Amazon: pure white (RGB 255,255,255), no props/text/watermarks
   • Instagram: lifestyle context, visually appealing
   • Xiaohongshu: vertical format, aesthetic, text overlay space

Example: building a custom prompt

User request: "I need a hero shot of my leather wallet for Amazon"

result = product_photo(
    product_path="uploads/wallet.jpg",
    platform="amazon",
    prompt="premium leather bifold wallet, rich brown color, slight angle showing card slots",
)

The script automatically builds:

Transform this product image into a professional e-commerce photo.
Keep the product exactly as it is — preserve its shape, color, details, and branding.
premium leather bifold wallet, rich brown color, slight angle showing card slots.
Photography style: professional product hero shot, clean composition, studio lighting...
Background: pure white background #FFFFFF, clean, e-commerce standard, no shadows.

---

14. E-commerce image set guide

A complete product listing needs 7-9 images. Use product_photo_set() for automatic generation, or create individual shots:

Position	Image Type	Style	Background	Purpose
1	Hero / packshot	hero	white	Primary listing image
2	Lifestyle	lifestyle	lifestyle	Product in use/context
3	Detail close-up	detail	studio	Material quality, craftsmanship
4	Scale reference	scale	studio	Size in hand or next to known object
5	Alternate angle	hero	white	Back or side view
6	Packaging	packaging	studio	Unboxing experience
7	Flat lay	flat_lay	textured	Arranged composition
8	Infographic	infographic	white	Dimensions, specs, features
9	Seasonal	seasonal_*	auto	Campaign-specific

---

15. Error handling

The script returns structured results. Always check success:

result = product_photo(product_path="uploads/product.jpg")
if result["success"]:
    for img in result["images"]:
        print(f"Saved: {img['local_path']}")
else:
    print(f"Error: {result.get('error')}")

Common errors:

• "File not found" — check the product_path
• "Unsupported image format" — use JPG, PNG, or WebP
• "Image too large" — max 10 MB
• "Either a product image or a prompt is required" — provide product_path/product_url or prompt
• "Unknown style/background" — check available presets in §7/§8