Image Generation

Generate text-only and reference-guided image directions with Gemini or OpenAI.

Image providers

VideoBrainstorm sends Gemini and OpenAI image requests through authenticated backend routes. Provider keys stay server-side or in encrypted BYOK storage.

• Gemini image: default path for broad exploration and reference-guided concepts.
• OpenAI image: GPT Image 2 generations and edits.
• Reference images: add visual guidance when words are not precise enough.
• Direction labels: multi-image runs get short labels so results are easy to compare.

Text-only generation

Text-only generation is best for new directions. Keep the prompt visual and concrete.

Example: "Three distinct thumbnail directions for a premium iced coffee launch: bold studio lighting, lifestyle desk scene, and surreal morning energy."

Example outputs:

• Direction A, Bold studio: high contrast product hero with a clean launch feel.
• Direction B, Lifestyle desk: practical social frame with space for copy and logo.
• Direction C, Morning metaphor: more expressive option for campaign exploration.

Reference-guided generation

1. Add project references or direct prompt attachments.
2. Say how to use each reference: product, style, composition, or mood.
3. Generate a small set to compare adherence and range.

Say "use this as style guidance" when you want a look. Say "keep this product accurate" when fidelity matters.