WAVE · Enhance
AI super-resolution for your video — no re-encode pipeline to build.
POST a clip to /v1/enhance and get back a sharper, AI-upscaled 672×672 artifact — rendered in-graph by FFmpeg 9's ONNX Runtime DNN backend, no separate inference service to stand up. v1 ships ESPCN super-resolution (CPU); GPU serving is the named fast-follow.
One call, a sharper clip back
Point WAVE at a video and get a fixed 672×672 AI-upscaled artifact back (3× the model's 224×224 working canvas) — pay by the minute of output rendered, settled through the same WAVE gateway key/meter every other spoke uses. v1 is CPU-EP and input-size-bounded (see /llms.txt for the exact constraint); GPU-EP + native-resolution tiling are the named fast-follow.
endpointPOST /v1/enhance ?model=espcn → enhanced video
authAuthorization: Bearer <key> (via gateway)
healthGET /v1/enhance/healthz