Dall-E, Midjourney, Stable Diffusion (etc) - who's playing, and with which?
Out of context: Reply #32
- Started
- Last post
- 824 Responses
- Beeswax2
Abstract stuff is easy but I'm trying to generate specific images based on reference photos or my illustrations and MJ fails big time.
it can draw 2 cats but it fails to draw the 3rd cat
It doesn't understand things like "orange cat" and creates a cat and an orange or a cat in the shape of an orange.
prompts like "orange grove, orange orchard, orange garden" create a garden but in orange color.
it can't create a simple prompts like "cat hiding under a tractor" cat is always on top or inside the engine.
I used keyword weights but that throws everything off even more.
I've spent hours of fast render credits and had to upgrade subscription. Still no tangible results.
I used it once to create a mood board for a client, that's the best real life use case for me so far.