It’s always fun to see how various prompts yield different results in Neural Blender. I wanted to make an image of tiny people swimming in a bowl of ramen as if it were a pool, but this turned out to be more challenging than expected. Simply typing in “ramen swimming pool” doesn’t combine the concepts into one, it merely places them side-by-side in the same image.
Although we get some cool vaporwave vibes, especially in the pixel art form, it’s not the outcome I’m looking for. To get around this, maybe adding ‘people’ in the prompt will help. Absolutely not. We get hilarious nightmare fuel instead.
What I was aiming to produce was more along these lines:
Efforts to combine these last two, using the prompt “tiny people swimming in [noodle/ramen] soup pixel art,” results in funny but sub-optimal outcomes. It’s more of a pixelized photo, not the cartoony pixel art I’m looking for, but it does get better as I continue to run the prompt.
Changing up the prompt to “an illustration of swimming in ramen soup” produces some interesting new results.
What about adding ‘pool’ to the prompt? Generating “an illustration of swimming in a pool of ramen soup” brings us back to square one; we get lunch for an athlete. What about “an illustration of swimming in ramen soup pool?” It’s a little closer but they are still treated as separate concepts.
Okay one last chance. I could do this all day but it’s time to bring this experiment to a close. Let’s see what “an illustration of people swimming in ramen soup bowl” brings us.
This isn’t exactly what I had in mind, but it’s as close as I’m going to get.
For even more interesting results, check out my AI Art page.