Chrome Sprint Study
A reflective running shoe floats above a clay track in low golden light for a high-energy sports campaign.

STARTING PROMPT
Reusable prompt
The original English prompt is kept unchanged so you can reproduce the Chrome Sprint Study example.
A dynamic editorial product photograph: a futuristic chrome running shoe suspended above a red clay athletics track, long late-afternoon shadow, tactile mesh and reflective metal details, energetic fashion campaign lighting.
What we saw when we re-ran it
On 2026-10-27 we re-ran the Chrome Sprint Study prompt twice on Nano Banana 2 (gemini-3.1-flash-image) at 1:1 and 1K through POST /v1/media/tasks. Each task finished in about 18 to 24 seconds end to end, with the status polled every 5 seconds.
- Both runs printed made-up lettering on the shoe ("AVANT-GARDE" on one, "AETHEL" on the other). The prompt does not ask for an unbranded shoe.
- The light drifted: one run kept a warm low sun, the other came back under a bright midday blue sky, though both kept a long shadow on the track.
- Lane lines, lane numbers and stadium stands appeared in both runs, and the chrome finish and mesh detail held.
Make it yours
- 01
Ban letters on the shoe
Add "no lettering, no logos and no brand marks anywhere on the shoe" after the mesh and metal clause. Both runs printed an invented name on the upper, so this sentence controls whether the product arrives clean enough for a campaign layout.
- 02
Pin the light
Replace "late-afternoon" with an explicit time and direction, for example "low warm sun from camera left just before sunset", and repeat it in the shadow sentence. One run came back at bright midday, so the time of day decides whether the track stays golden.
- 03
Clear the stadium
Add "an empty track with no lane numbers, no spectators and no signage" when the background must stay plain. Both runs filled the frame with lane lines, lane numbers and stadium stands, which turns a clean product shot into a race-day scene.
Check before you rely on it
- Both runs invented a name on the shoe, and in our Nano Banana 2 tests a pottery scene that asked for no text still showed SHIMPO on the wheel. State that the shoe is unbranded and inspect the mesh for small text.
- The light is not stable between requests. With one prompt, one image came back in warm late sun and another at bright midday, so run the shot more than once when a set has to sit together in one layout.
- gemini-3.1-flash-image was deprecated on 6 October 2026 with no shutdown date named. Copy the id exactly while it answers, and start new campaign work on gemini-nano-banana-2.1.
Run it through the API
Built from the template's model id, prompt and settings. Set OMNIAKEY_API_KEY in your environment first.
curl --fail-with-body https://api.omniakey.com/v1/media/tasks \ -H "Authorization: Bearer $OMNIAKEY_API_KEY" \ -H "Content-Type: application/json" \ -H "Idempotency-Key: $(uuidgen)" \ -d '{ "model": "gemini-3.1-flash-image", "type": "image", "operation": "generate", "input": { "prompt": "A dynamic editorial product photograph: a futuristic chrome running shoe suspended above a red clay athletics track, long late-afternoon shadow, tactile mesh and reflective metal details, energetic fashion campaign lighting." }, "parameters": { "aspect_ratio": "1:1", "image_size": "1K" } }' # Then poll the task until its status is succeeded curl --fail-with-body https://api.omniakey.com/v1/media/tasks/<task-id> \ -H "Authorization: Bearer $OMNIAKEY_API_KEY"
Questions
Why does my chrome shoe carry a brand name?
The prompt never rules lettering out, and this model adds text uninvited. Both runs on 2026-10-27 printed an invented name on the upper, "AVANT-GARDE" in one and "AETHEL" in the other. Add a no-lettering sentence and inspect the mesh before the image goes out.
How do I call this template from the API?
Use POST /v1/media/tasks with model gemini-3.1-flash-image, the prompt as the task input, and aspect_ratio 1:1 plus image_size 1K as request parameters. Poll the task, then read assets[].url; the signed link expires, and querying the same task again returns a fresh one. Billing is per generation.
Will 2K or 4K sharpen the mesh detail?
A larger image_size gives the reflective metal and mesh more pixels: Nano Banana 2 accepts 1K, 2K and 4K. The template is set to 1:1 at 1K for the gallery grid, and 2K and 4K are billed at higher tiers, listed in the price table on the model page.
More image templates
- Midnight Ramen CounterSteam rises from a ramen bowl at a rainy midnight counter under warm paper lanterns.
- Citrus Morning TableBlood oranges, blue timber, linen, and hard morning light form a bright hospitality still life.
- Sunken Listening RoomWalnut panels, green modular seating, and warm lamps create a tactile 1970s listening room.
ONE ACCOUNT · ONE API
Create with OmniaKey
Start with the prompt and settings above, then adapt the idea to your project.