Image Input

Overview

As IA becomes multi modal, accompanying text prompts with images will be a common input to expose the insights contained in the image.

Unless a model is pre set for an image interpretation task, some text prompt will usually need to direct the AI's interpretation of the image asset.

User Archetype

Users no matter their technical ability are mostly familiar with images & visual files in the age of the internet. Couple this with how visual people are & how rich image formats are, there is a large amount of data contained in our existing images.

While the level of detail and sophistication will vary, the ability to attach & query an image will be a ubiquitous for al user types.

Considerations

Anytime we provide a file upload we want to consider core metrics of size & resolution.

In terms of resolution, especailly when providing the image for analysis, a minimum resolution should likely be set to set some threshold for a useful image resource. This goes hand in hand with file types, where formats like .gif aren't general capable of high resolution & therefore aren't as useful as JPG or PNG file types.

Aside from general file upload considerations your particular design context may need more sophisticated targeting for the image resource. Do your users need to identify a focal point or specified area within the image for the AI.