AI Images
Sidebar: Content → Images
Module: AI Creatives — Plus and Prime.
Describe a picture in your own words — with or without a reference from your gallery — and get it back ready to use on a page, a post or an e-mail.
Like the AI video generator, this is a paid-plan feature that runs on your own OpenRouter account. No Flights are charged.
Where your data goes. Your OpenRouter key lives in your browser only, and every request goes straight from your browser to OpenRouter. Your briefing, your prompt and your reference images do not pass through FlyB’s servers. FlyB has no control over how OpenRouter and the model providers behind it handle them — see OpenRouter’s privacy policy and our AI disclosure.
Connect and choose models
The key is the same one used by the video generator — connect it on either screen and both are ready.
| Role | What it does |
|---|---|
| Prompt model | Reads your reference pictures, when there are any, and writes the prompt in English |
| Image generation | Draws the image |
The prompt model is shared with the video generator: picking it here picks it there too.
Each image model publishes its own capabilities, and only what it supports is offered in the next steps: how many reference images it accepts (or whether one is required), how many images it can produce per run, transparent backgrounds and SVG output.
Format
Aspect ratio (including Automatic, where the model decides — usually matching your reference), resolution, images per run (several options generated on the same call) and transparent background where the model supports it, which returns the subject cut out — for logos, stickers and product shots.
Some models publish no format options at all and generate in their own default shape.
Reference images
From a reference builds the image from pictures in your gallery — your product, your photo, your brand asset. From the prompt only builds it from text alone, in which case every visual decision has to be described in words.
References are read, not framed: pick the whole picture the model has to look at.
The prompt
Write your briefing in any language — the prompt model turns it into English, which is what image models read best. Or paste your own prompt and skip the step. The prompt is editable, and the exact text on screen is what the image model receives.
Generating and saving
Generation usually takes a few seconds, or a minute or more at 4K.
Do not leave or reload the screen while it generates. The run happens in this tab. Leaving loses the result while OpenRouter still charges for it.
Results appear as cards. Each one can be expanded, downloaded, or saved to your gallery — and until you save it, the image exists only in this tab. The screen warns you when you have unsaved results.
Saved images go into a dedicated folder in your photo gallery and are then usable anywhere in the platform.
Starting over
Start a new image clears the format, references, briefing, prompt and generated images. Your key, your chosen models and anything already saved to the gallery are kept.
Rights in what you generate
Generated images come back to you subject to the terms of the model provider that produced them, and AI output may not be eligible for copyright protection in every jurisdiction. Before using an image commercially, check the licence of the model you chose. See Terms of Use → Ownership.