Tools to explore the World of AI-Generated Images

AI Image generation for Meatsuits

BadPirate

--

In this post, I’ve compiled a list of tools that have been helpful in experimenting with AI-generated images. There are more complex tools if you want to build your own python environments, but I’ve left this list to be just those that are easily used and accessible.

Image Generation

DALL-E, IMO, best translation of complex text entry into images. Also has good “In Painting” and “Out Painting” for an online tool, allowing you to “re-roll” or re-imagine portions of the image. It has some trial tokens, but is a paid API.

DALL-E, Elephant Jumping on a Trampoline

https://www.midjourney.com/home/

MidJourney generates amazing photo realistic and art photos, and has the best generation that I’ve seen. It has a number of free tokens, is used through Discord, but eventually you have to pay a bit.

For best results add --v 4 to your prompts.

A Robot Artist, covered in bright paint, in her NY artist apartment — v 4

If you have an M1 Macbook and want to generate images easily (and locally) DiffusionBee is an easily downloaded app that uses your own computer to generate images (DALL-E and MidJourney both use web based hosting). You can also download different .ckpt models (Training models) allowing for more variation and less restriction than the online models (See HuggingFace).

Elephant jumping on a trampoline: OpenJourney (DiffusionBee)

Improving images / Upscaling

--

--

BadPirate

Kevin Lohman, Software Engineer, Father, Story Teller, and former US Navy Sailor (who never set foot on a ship)