/ic/ - Artwork/Critique

>>6235258
Yes but to elaborate, people practice figures or portraits because it contains all fundamentals and is a good test of skill.

But, AI doing “floating” portraits or figures means little by comparison. For example, the study mentioned in the article below prompted “a spoon in a cup” and the results looked like professional product images but a cup “on” a spoon look like picrel. This is because AI doesn’t have understanding, it’s only copying patterns, which they exploited with simple relational prompts instead of complex ones.

https://www.unite.ai/is-dall-e-2-just-gluing-things-together-without-understanding-their-relationships/

And according to the study, they might abandon text to image entirely due to it’s inherent limitations for general AI….