Using Generative AI

[David Eichler]

Is there a way I can use generative AI to create objects that are similar to, but not identical, to other objects in a scene, without having to go through trying a lot of verbal descriptions until I get something sort of usable? Yes, I know I could try copying and warping existing objects into different shapes, so they don't look quite the same as other similar objects, but I am wondering if there is a way that AI can make this process quicker, to get something usable.

[digitaldog]

Yes, there is a load reference image (yours or others) that can be used for the generation of AI.

[atiqursumon]

Sure thing! Generative AI can help with that! Apps like Stable Diffusion, Midjourney, or DALL·E let you whip up similar objects just by using prompts or reference pics. You can even use the image-to-image features to change existing stuff without needing to dive into heavy editing.

[plugsnpixels]

On the Mac you can use the free DiffusionBee app for image-to-image, text-to-image, etc.

Here is a quickie where I modernized a 1910 image with a basic prompt.

You'll want to try different source models, styles and input strength.

There are tons of free apps to do this sort of thing.