How to Create Images in DEVONthink

While DEVONthink 4's core focus is still on textual processing, there are times when an image would be useful to illustrate something, or just something creatively fun to explore. With the Pro or Server edition, you can use AI to create images directly in DEVONthink. Here is what you need to know.

First of all, you need an API key for a text-to-image AI provider. A key for OpenAI’s services covers DALL-E 3*. Alternatively, Replicate.com gives you access to some powerful models, including the excellent Flux series and some others. Once you have a key, open Settings > AI > Image Generation in DEVONthink, choose a default model and specify any other recurrent options you want to use.

Now, choose Data > New > Generate Image and a utility panel appears, pre-populated with your defaults and ready for your prompt. Write your prompt, press the Generate button, and see what it comes up with. You can also request an image via the Chat assistant, e.g., “I need a picture of a cathedral like the ones in Portugal”. Drag and drop the image from the chat, Control-click it for other options, or save it with the chat.

On small caution: These services may charge per-image and the pricing may vary per-model. Especially as this can be a fun thing to explore, you can forget the meter is running. So make sure you’ve read through the pricing structure.

For more information, check out the section Windows > Generate Image in the built-in Help or user handbook.