The Struggles Of Picking Models (A Small Rant)
By cocoayoc
Tags: ai image, ai generations
Comments (6)
paulie
I feel your frustration with a model giving you what you want once, then seemingly never again. There are a potential myriad of reasons. Prompt wording, maybe too many loopholes in the prompt. I think @Cheinia had a blog post dealing with the ai adherence vs. its own interpretation of what you prompted. You might try running your initial prompt through an enhancer. Check that the result has the items you want and give it a try. One thing I've found when I use an ehnanced prompt is that the results stay relatively consistent within any particular model
DavidP
That is frustrating. Also for me is being able to choose the model that will deliver closest to what I imagine. The variation between models is always interesting, in all aspects of the prompt, and don't have one model that consistently wins.
EternaSky
This is nicely done
JustAsh
I feel the pain of this post. It's extremely frustrating to feel like credits were wasted on a result that didn't work out.
mim86
I know exactly what you mean. It becomes a fight in your head between quality, over credit budget.
King
@DavidP we will release a new feature call apps-in-app, which allow users to become designer to design a niche that does a job well, which is model agnostic. i.e., to other users, they know what it does, and they use it to achieve what it aims. it is model agnostic, straight forward. Higher first2value, less learning curves.