Lighting and palette prompts: Sora guide vs Omni guide compared

Both prompt guides tell you to name the light, and both reward consistency: OpenAI's Sora guide asks for lighting plus a small set of color anchors, and Google's Omni guide asks for the light source and its quality. A prompt that names both transfers between the two models with little rewriting.

Compared from OpenAI's Sora 2 Prompting Guide and Google DeepMind's Omni prompt guide, both read on 2026-10-03. Neither page promises a result, so test.

Side by side

The Sora guide separates the look of the light from the palette so the frame stays coherent across shots. Google's guide places lighting among the elements of a good prompt together with shot type, camera, style and action.

A prompt that works for both

Write the light as a source and a quality, then the palette as three nouns: "soft window light from camera left, warm lamp fill; amber, cream and walnut tones." Avoid ten adjectives. If Omni ignores the palette, move the color words into the light sentence.

Request on Sume

Omni on Sume is gemini-omni-flash-1.1.

curl -X POST https://api.sume.com/v1/video-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: light-palette-001" \
  -d '{
    "model": "gemini-omni-flash-1.1",
    "prompt": "In a single continuous shot, a ceramic mug on a desk. Soft window light from camera left with a warm lamp fill; amber, cream and walnut tones. Static camera.",
    "resolution": "720p",
    "duration": 4,
    "aspect_ratio": "16:9",
    "mode": "async"
  }'

Keep a palette across clips

When you stitch several clips, reuse the same light and palette sentence in every prompt, and join with Timeline 1.0. The Video Router docs list the model's fields.