I've been participating in an ML art community for about a month and it's been immensely fun. It's one of the more welcoming and supportive online spaces which is a breath of fresh air. Just some random thoughts:
Firstly, it's made me view art in a surprising way. I will now intentionally seek out new artists keeping an eye out for good prompt fragments. Prompt fragments are a real currency in these spaces. Sure, an image or batch is fun to look out, but trading prompts is viewed as having much more worth.
Tools like Clip Retrieval[1] are genuinely fun to play around with on their own but are also great tools for evaluating prompts before committing to generating an image.
There are some challenging open problems like creating human faces consistently. It's not exactly clear if the model is capable of the task, or what the boundaries are. Either way, it's still entertaining to try.
Firstly, it's made me view art in a surprising way. I will now intentionally seek out new artists keeping an eye out for good prompt fragments. Prompt fragments are a real currency in these spaces. Sure, an image or batch is fun to look out, but trading prompts is viewed as having much more worth.
Tools like Clip Retrieval[1] are genuinely fun to play around with on their own but are also great tools for evaluating prompts before committing to generating an image.
There are some challenging open problems like creating human faces consistently. It's not exactly clear if the model is capable of the task, or what the boundaries are. Either way, it's still entertaining to try.
1. https://rom1504.github.io/clip-retrieval/?back=https%3A%2F%2...