DDColor Colorization Worker (#8)
RunPod Serverless worker that colorizes black-and-white images using the
ModelScope DDColor pipeline (damo/cv_ddcolor_image-colorization).
Contract
Input:
{ "image": "<base64 (optionally data:...;base64, prefixed)>", "input_size": 512 }
Output:
{ "image": "<base64 PNG, same dimensions as input>" }
On error: { "error": "..." }.
Deploy
- Dockerfile path:
/workers/colorize/Dockerfile - Build context: repo root (
docker build -f workers/colorize/Dockerfile .) - GPU: medium (RTX 3080 / 3090, ~10-12 GB VRAM)
- App env var to set:
RUNPOD_COLORIZE_URL(point your app at this endpoint)
Notes
- Model weights are cached to the network volume via
MODELSCOPE_CACHE(/runpod-volume/modelscope) when/runpod-volumeis mounted, so the first request downloads and subsequent cold starts reuse the cached weights. input_sizecontrols the DDColor internal working resolution (default 512). The output is always resized back to the exact input dimensions.