Image Generation

Overview

Image Generation turns a written description into a picture, without leaving your conversation. Describe what you want, choose a shape if you care about one, and the image appears inline — ready to view, save, or keep building on with follow-up requests.

Why generate instead of search

Searching the web for an image means settling for whatever already exists — and dealing with licensing you may not be able to verify. Generating one gets you exactly the composition, subject, and style you describe, and it's yours to use in whatever you're working on.

How to use it

Open the composer's "+" menu (next to the message box).
Select Generate image. The composer switches into image mode — its placeholder changes to "Describe an image to generate…".
Type a description of the image you want.
Optionally choose a model from the dropdown (see below) and a ratio — Auto, Landscape, Portrait, or Square.
Send it, the same way you'd send a normal message.

While it generates

You'll see an animated placeholder sized to the ratio you picked while the image is being created. Once it's ready, it renders directly in the chat. Click it to view it larger. If the model rewrote or refined your description before generating, you'll see a small "Prompt used: …" caption underneath so you know exactly what was actually generated.

Choosing a model

The generate-image dropdown offers a few different image models. If you're not sure which to pick, the default works well for most requests — switch only if you want a specific style or are unhappy with a result and want to try an alternative.

A worked example

Say you're putting together a slide and want a simple illustration for it.

Click "+" in the composer, then Generate image.
Type: "A minimalist line-art illustration of a rocket launching, on a plain white background."
Choose Landscape since it's going on a slide, and send it.
The image renders inline once ready.
You decide the rocket should be blue: send a follow-up, "make the rocket blue," and a new version generates from that adjustment.
Click the final image to open it larger, then save it to your device the way you'd save any image from a webpage.

Example prompts

"A cozy reading nook with warm lighting, illustrated in a flat design style."
"A simple icon of a lightbulb, transparent background, minimalist."
"A photorealistic image of a mountain lake at sunrise."
"A diagram-style illustration of a funnel, three stages, labeled Awareness, Consideration, Decision."

Saving and sharing

The generated image is saved as part of the conversation, just like a text reply — it will still be there if you come back to this chat later.
If the chat is a Shared Room, everyone in the room sees the generated image too, the same way they'd see any other message.
To keep a copy on your own device, click the image to open it larger, then save it the same way you'd save any image from your browser.

Tips

Be as specific as you'd be with any image tool — subject, setting, style, and mood all help. "A cat" gives a generic result; "a orange tabby cat sitting on a sunlit windowsill, watercolor style" gives you more control.
Ask for changes as a follow-up. "Make the background darker" or "try it as a square instead" works as a normal next message.
Pick a ratio that matches how you'll use it — Square for an avatar or icon, Landscape for a banner or slide, Portrait for a poster-style image.
Switch out of image mode when you're done — send a normal question next and the composer returns to regular chat behavior on its own.

Troubleshooting

Nothing happened after I sent my description. Check that you actually selected Generate image from the "+" menu first — without it, your message is treated as a normal chat question instead of an image request.

The image doesn't match what I described. Try being more specific, or try a different model from the dropdown — models vary in how literally they follow a description versus how much creative interpretation they add.

The result looks different from what I typed. If a "Prompt used: …" caption appears under the image, the model adjusted your wording before generating — that's what actually produced the picture you see.

Frequently asked questions

Can I generate more than one image at once? Each request produces one image; ask again (with the same or a refined description) for another.

Can I edit an existing image, like removing an object from a photo? Image Generation creates new images from a text description — it isn't an editor for photos you already have.

Is there a limit to how many images I can generate? Image generation draws from your regular usage the same way other requests do — check Plans & Limits for how your plan's allowance works.

Will my generated images show up in Projects' knowledge base? No — Projects' knowledge base is for uploaded text documents. A generated image lives in the chat it was created in, like any other message.

Can I generate an image and then ask questions about it? Yes — once it's in the chat, you can ask Evelant to describe it, or refine it further with a follow-up description.

Does it work from the web dashboard as well as the extension? Yes — image generation is a chat composer feature available wherever you're chatting with Evelant, not something tied to a specific website you're browsing.

Related features

Code Sandbox & File Generation — for charts built from real data, rather than a generated illustration.
Interactive Charts — for pie and donut charts you can hover and click into right in the chat.
File & Image Attachments — for having Evelant look at an image you already have, the reverse of generating a new one.