Pick an AI video model by the input you have: photo, clip, audio, none
Match the Sume video row to your input: nothing gets Wan 3.0 or Omni, a photo gets a first-frame row, a clip gets Omni edit or Recast. Read 2026-10-13.
Start from what you hold. With nothing but text, use wan-3.0, seedance-2.5, gemini-omni-flash-1.1 or sume/auto. With one photo, use a first-frame row. With a clip you want changed, use an edit row. With several photos of one subject, use a reference row.
Input to row
The table lists the rows the docs name for each input.
| You have | Use |
|---|---|
| Text only | wan-3.0, seedance-2.5, Omni, H3, sume/auto |
| One photo as the first frame | kling-3, wan-3.0, Seedance, H3, Omni, grok-imagine-video-1.5 |
| Start and end photo | kling-3, wan-3.0, Seedance, H3, Omni |
| Many photos of one subject | wan-3.0, H3, Omni, Seedance |
| A clip to edit | Omni edit, h3-max-recast, higgsfield-genjutsu, hailuo-h3-regen |
| Audio to follow | Wan 3.0, H3 rows, Seedance rows (reference audio) |
Notes
grok-imagine-video-1.5 is image-to-video only on Sume, so it needs a photo. Omni takes no reference audio. If your input is a still plus a voice track and you want the person to speak, see the lip-sync post.
Sources
Related posts
More in Use cases
- Place your character in concept art scenes with two references
Design locations first, then put the hero in each one with two input_references. Recraft's Oct 9 method as a Sume API script with costs.
- Smoke alarm battery reminder video for a fire department or HOA
One 10-second Gemini Omni Flash clip with a real chirp, captioned with Ready.gov's smoke alarm rules, costs about $1.45 on Sume. Prompt, cues and checks.
- SynthID Detector is public: what it checks for AI ad creative
Google opened SynthID Detector to everyone on Oct 7, 2026 for Google, OpenAI, NVIDIA and Kakao marks. What a result means for AI ads and Sume files.
- TikTok Shopping Assistant: what should product videos say?
TikTok's Shopping Assistant answers sizing, shipping and availability in chat. Make product clips that match your listing, with photo-to-video and burned cues.
Written by Sume