Images and image editing
Create images from a sentence with Image Generation, and change an existing picture by describing the edit with Image Manipulation. Both return the result right in the chat.
Need a launch banner, a campaign shot or a quick visual for a slide? Describe it and StickyPrompts generates it. Already have a picture that is almost right? Attach it and say what to change. Both happen in a chat, so you can keep refining with follow-up messages until the image is exactly what you need.
Generate an image
- Open Image Generation
Go to Apps and choose Image Generation. You get a focused composer titled Generate images with AI.
- Describe the image
Say what you want to see, as specifically as you can. For example, the prompt used in the recording above:
Product photo for a launch banner: a sleek laptop on a warm oak desk showing a retail analytics dashboard... - Send and review
Select Send. The image comes back in the chat, with buttons to view it full size or download it.
- Refine with a follow-up
Not quite right? Reply in the same chat: “make the light warmer”, “wider crop, more of the store”. Each reply produces a new image, and the earlier ones stay in the conversation for comparison.
Edit an existing image
Image Manipulation changes a picture you already have. Drop in the image and describe how you want to transform or extend it. It works with PNG and JPG files.
- Open Image Manipulation and add your image
Go to Apps, choose Image Manipulation and attach the picture - a product shot, a campaign photo, a previous result. You can also Attach an image in any image chat.
- Say what to change - and what to keep
Be explicit about both. The edit in the screenshot below used:
Change the sweater to deep forest green and make it golden hour. Keep her face, pose and the store exactly the same. - Compare and download
The edited image appears under your original, so you can compare them before you download.
Prompting tips for images
A useful image prompt answers five questions:
| Question | Example |
|---|---|
| What kind of image? | Product photo, editorial campaign shot, flat illustration, icon |
| What is the subject? | A sleek laptop showing a retail analytics dashboard |
| Where? | On a warm oak desk; in a sunlit independent clothing store |
| What light and mood? | Natural light, warm film tones, candid |
| What shape? | 4:5 for social, wide for a banner |
More tips:
- Lead with the purpose. “Product photo for a launch banner” tells the model about composition, polish and empty space for text before it reads anything else.
- Describe, do not list keywords. A clear sentence gets better results than a pile of adjectives.
- Put the aspect ratio in the prompt, as in “candid, 4:5”.
- Ask for one change at a time when editing. Two small edits are more predictable than one big one.
- Check text in images. If you need exact wording on the image, keep it short and review it closely.
Images elsewhere in StickyPrompts
- Agents and saved prompts can generate images inline when the Image Generation tool is switched on. See Agents and Prompt library.
- For moving pictures, see Video generation. A generated image makes a good starting frame.