Appearance
Images
Open a tool from Tools, or ask KimtAI in a chat to do the job for you. Most tools here save their result to Files, next to your other work, and it counts towards your storage. Look is the exception: it saves nothing.
Image to text
Pull the text out of a photo, a screenshot or a scanned PDF.
Give it: one image or PDF. Pages that already hold text are copied; scanned pages are read.
You get: the text on screen, and a file in Files. If an image has no text, the tool says so, lists what it tried and offers to read it with the vision model.
Options:
- Language of the text: one language reads more accurately than all of them.
- Pages (PDF only): for example
1-3,5or10-. - Read pages as images (PDF only): use it when the copied text comes out wrong.
Ask KimtAI: "Copy the text from this screenshot."
Remove background
Remove the background from a photo and keep a clean cut-out.
Give it: one photo with a clear subject: JPG, PNG, WebP or HEIC.
You get: a PNG with a transparent background, saved to Files. The result shows the cut-out beside the original so you can compare. Very large photos are scaled down to 4000 pixels on the long side first.
Options:
- Crop to the subject: trims the empty space around it.
Ask KimtAI: "Cut me out of this photo for my profile picture."
Generate image
Describe a picture in a sentence or two and get new images.
Give it: a description (up to 2000 characters). Four examples are on the page if you want a start.
You get: PNG pictures saved to Files. Every run makes new pictures, even for the same words. If fewer pictures were made than you asked for, the page says so. Pictures made earlier in the same visit stay on the page.
Options:
- Shape: Square, Wide or Tall.
- How many: 1 to 4.
TIP
If image generation is not turned on for your Kimtos, the page says so. Nothing is lost while you wait.
Ask KimtAI: "Make a picture of a calm lake at sunrise."
Look
Say what is on a picture, a screenshot or one page of a PDF, and find named things on it.
Give it: a picture, a screenshot or a PDF.
You get: an answer about the page. Each thing you named is boxed on the page, with its place given as a distance from the left and top and its width and height. Nothing is saved.
Options:
- Things to find: up to 12, for example "Company logo, signature, address block". Press Enter after each one.
- A question about it: optional, for example "Describe the layout".
- Page: for a PDF, which page to look at.
Ask KimtAI: "Look at this and tell me where things are."