For this project, I wanted to experiment with horror because I know AI can have trouble with violence and graphic content. Surprisingly the AI was able to generate all the prompts without any issues, although I didn’t dive too deep into any seriously graphic stuff.
I gave the AI a few different prompts that experimented with perspective, lighting, and body shape. It took a while to get all the components of the images right, but after lots of trial and error, I managed to slowly get everything right.
Something I noticed during the process was that the AI did much better at generating more accurate scenes when I wrote prompts in simple but excruciating detail. I gave the AI one prompt that was similar to how a book author would describe a scene and it seemed to give me the best results from a single prompt.
Here is the link to my presentation…