Just did one of those things that makes you go “huh”.

OpenArt hasn’t really impressed me with fantasy renders.

So far, I can only see how to use a single model in a render. So you do one character-model and describe the rest of the scene. Here Sarah battles an ogre.

I’m not real pleased with the Ogre. It looks plastic to me. And there isn’t really much action or “battle” to the scene either.

OpenArt has an AI feature that lets you enhance your written description for a render.

Actually, I changed the terrain. The rest of this is the same description, just “enhanced” by the AI. Sarah looks less like herself (is it taking more liberties with the model? or just one of those things?). And seriously, that armor is described as “full plate”. Sure, maybe the Hooters version.
But this does show some improvement. It encourages me that I might be able to get more use out of OpenArt.

For comparison

This is the same description just cut and pasted over to Co-Pilot. I did Have to change the character description from “Sarah” to “blonde warrior woman”, but otherwise no changes.

I much prefer the sense of action and drama in the Co-Pilot/Dalle-3 render. Sure its over the top, but that’s what I’m doing!

2 responses to “Playing with AI”

  1. Zeno Avatar

    That first Ogre is… “something”. That’s for sure. I feel like when you train an OpenAI model it trains your character on top of some bare-bones baseline model. So it can only add things with your model that the baseline understands. That’s why I’d like to be able to combine multiple models. Train a “paladin”, train an “ogre”, and then have them doing something together.

    Like

    1. atcDave Avatar
      atcDave

      I’m okay with only one model at a time in battle. Maybe not thrilled, but okay.

      But I really want to be able to show the models, the couple together in a range of story and drama related scenes.

      Like

Leave a comment