Wan 2.2 I2V A14B Image-to-Video on Modular

  import base64
  from openai import OpenAI
  
  client = OpenAI(
      base_url="https://model.api.modular.com",
      api_key="<your_api_token>",
  )
  
  prompt = """A 2D pop art comic book animation. Thick black ink outlines,
  flat color fills, Ben-Day halftone dots covering every surface,
  limited palette of pink, cyan, red, and black. A rainy back alley
  in Shinjuku at night, rendered as a comic panel: kanji neon signs
  drawn as flat graphic shapes with halftone glow, puddles drawn as
  flat cyan shapes with white highlight marks, rain shown as diagonal
  ink lines. A helmeted astronaut figure in a black raincoat walks
  away from the viewer down the alley, never turning. Steam from a
  ramen vent drawn as cartoon curls. The viewpoint pushes slowly
  forward through the alley. Hand-drawn cel animation feel, no
  photorealism, no 3D shading, no lens effects."""
  
  response = client.responses.create(
      model="Wan-AI/Wan2.2-I2V-A14B-Diffusers",
      input=prompt,
      extra_body={
          "provider_options": {
              "video": {
                  "height": 512,
                  "width": 512,
                  "steps": 28,
                  "num_frames": 81,
                  "frames_per_second": 16,
                  "response_format": "b64_json",
              }
          }
      },
  )
  
  video_data = response.output[0].content[0].video_data
  
  with open("output.mp4", "wb") as f:
      f.write(base64.b64decode(video_data))