change-color

Change Color Workflow

Change the color of clothing in product images using Visuals AI's color transformation workflow.

Supported inputs

• Local image files: PNG, JPG, JPEG (will convert to base64)
• Shopify product URLs: https://www.zenoficial.com.br/products/... (will fetch available images)
• Direct image URLs: Any publicly accessible image URL

Required environment variable

Ensure VISUALS_API_KEY is set in your environment. This is the master API key for Visuals AI.

Workflow steps

1. Determine input type

Check if the user provided:

• A local file path (check if file exists)
• A Shopify product URL (contains zenoficial.com.br/products/)
• A direct image URL (starts with http:// or https://)

2. Handle Shopify product URLs

If the input is a Shopify product URL, fetch available images first:

curl -X POST ${VISUALS_API_URL:-https://visuals-ai.vercel.app}/api/zen-product \
  -H "Content-Type: application/json" \
  -d '{"productUrl": "<shopify_url>", "action": "fetch"}'

This returns:

{
  "productName": "Product Name",
  "productUrl": "...",
  "images": [
    {"id": "...", "thumbnailUrl": "...", "fullUrl": "..."},
    ...
  ]
}

Present the images to the user using AskUserQuestion with the thumbnail URLs so they can choose which image to use. Show the product name and let them select from the available images.

After the user selects an image, upload it to get a fal.ai URL:

curl -X POST ${VISUALS_API_URL:-https://visuals-ai.vercel.app}/api/zen-product \
  -H "Content-Type: application/json" \
  -d '{"imageUrl": "<selected_fullUrl>", "action": "upload"}'

This returns:

{
  "url": "https://v3b.fal.media/files/...",
  "originalUrl": "..."
}

Use the url field as the image URL for the workflow.

3. Handle local files

If the input is a local file path, resize if needed to avoid 10MB payload limit:

file_path="<path>"

# Check file size and resize if necessary to avoid 10MB limit
# Base64 adds ~33% overhead, so target max 7MB original size
file_size=$(stat -f%z "$file_path" 2>/dev/null || stat -c%s "$file_path" 2>/dev/null)
max_size=$((7 * 1024 * 1024))  # 7MB

if [ "$file_size" -gt "$max_size" ]; then
  echo "Image is too large ($(($file_size / 1024 / 1024))MB), resizing to fit 10MB limit..."

  # Resize to max 1024px width using sips (macOS) or convert (ImageMagick)
  if command -v sips >/dev/null 2>&1; then
    sips --resampleWidth 1024 "$file_path" --out /tmp/resized-image.png >/dev/null 2>&1
    file_path="/tmp/resized-image.png"
  elif command -v convert >/dev/null 2>&1; then
    convert "$file_path" -resize 1024x "$file_path"
  else
    echo "Warning: Image may be too large. Install ImageMagick for auto-resize."
  fi
fi

# Detect mime type from extension
mime_type="image/jpeg"
if [[ "$file_path" == *.png ]]; then
  mime_type="image/png"
elif [[ "$file_path" == *.webp ]]; then
  mime_type="image/webp"
fi

# Convert to base64 data URL
base64_data=$(base64 -i "$file_path" | tr -d '\n')
image_url="data:${mime_type};base64,${base64_data}"