image-processing

Image Processing

A real estate platform receives listing photos in arbitrary resolutions and formats. Each image needs validation (minimum resolution, supported format), resizing to standard dimensions, metadata extraction (EXIF, dimensions), and watermarking before it can appear on the website. A corrupt JPEG should not block the entire batch.

Pipeline

[ip_load_image]
     |
     v
     +──────────────────────────────────────────────+
     | [ip_resize] | [ip_watermark] | [ip_optimize] |
     +──────────────────────────────────────────────+
     [join]
     |
     v
[ip_finalize]

Workflow inputs: imageUrl, sizes, watermarkText

Workers

FinalizeWorker (task: ip_finalize)

Finalizes image processing by assembling all outputs.

• Reads resized. Writes outputUrl, totalOutputs

LoadImageWorker (task: ip_load_image)

Loads an image from a URL.

• Sets format = "png"
• Reads imageUrl. Writes imageData, width, height, originalSize, format

OptimizeWorker (task: ip_optimize)

Optimizes/compresses an image.

• Sets result = "optimized"
• Reads format. Writes result, optimizedSize, compressionRatio, format

ResizeWorker (task: ip_resize)

Resizes an image to multiple variants.

• Rounds with math.round()
• Reads sizes. Writes variants, variantCount

WatermarkWorker (task: ip_watermark)

Applies a watermark to an image.

• Sets result = "watermarked"
• Reads text. Writes result, applied, position, opacity

---

40 tests | Workflow: image_processing | Timeout: 60s

See RUNNING.md for setup and usage.