Esports team intro sting: 5 seconds with a logo reference

For a 5-second esports intro, send the team logo as a reference image to Gemini Omni Flash 1.1 and describe the motion: $0.625 at 720p on Sume, or $0.9375 at 1080p. Omni on Sume accepts up to 10 reference images and up to 3 reference clips of 3 seconds, and the prompt refers to them as <IMAGE_REF_0> and <VIDEO_REF_0>. Prices read 2026-10-08.

The risk with a logo is the logo. Generated lettering drifts, so plan for one shot where the mark is clean and the rest is motion.

Make the logo stay readable

Google's Omni tips say text rendering works when the prompt gives the exact text, its placement and its animation. Reference the logo and say what must not change.

Fast energy sting. <IMAGE_REF_0> is the team logo: keep its shape, letters and colours exactly as in the image.
0-2s: dark arena, light beams sweep, camera rushes forward.
2-4s: the logo slams in at centre with a flash and sparks.
4-5s: hold the logo, light beams settle. Loud synth hit on the slam. No dialogue.

Cost and length

The minimum clip is 3 seconds, so 5 seconds is inside the range. Draft at 360p to check the motion.

If the lettering still drifts

Two repairs are cheaper than re-rolling. First, use the logo itself as the last frame (end_image_url) so the final frame is the real artwork. Second, keep the generated clip for motion only and place the real logo above it in a Timeline. Sume's video trim step can cut the best 3 seconds out of a longer clip, and returns a new MP4.

Stream overlays usually want 16:9, which Omni offers. For a vertical highlight reel, use 9:16.

Sound

Omni always makes an audio track and the API rejects generate_audio: false. A loud synth hit is a good thing to ask for here. If your stream software already plays its own sound, mute the clip's audio there.

Variants for each player

A roster needs one sting per player or one per map. Keep the prompt fixed and change only the reference image and the line about the name, then run the set as a batch. Ten variants at 360p cost $1.875, and ten finals at 720p cost $6.25 on Sume.

Submit with an Idempotency-Key header so a retry after a dropped connection returns the original job instead of a second charge; the same key with a different payload returns 409. Then poll GET /v1/jobs/:id/status and fetch GET /v1/jobs/:id/result when it finishes.

Do not ask the model to spell gamer tags. Put tags in your overlay software, where a typo is a text edit and not a $0.625 retry. Keep the generated clip to light, motion and impact.