For a book cover I played around with an art generator to produce a simple image, just silhouettes of two people. I had to iterate and tweak my prompt and settings a lot in order not to get 3D images, non-silhouetted ones, some totally different than what I was after. One even hallucinated an extra body part. It took a long time and nothing useful was produced. So I gave up and told my daughter to try it, and she rendered the prompt I texted her in less than an hour on her iPad. One more iteration and we were done.
And it pleases me greatly to credit my kid instead of that stupid software.
One thing few articles about AI art discuss is how the ability to verbally specify a picture will affect artists themselves, besides the IP issues. IMO it will make many people call themselves artists, flooding the market with images that haven't taxed their issuers creativity and technique. Making art will become the same as wandering in an art gallery without the exercise.
When the bottom of the market for lazy images falls out, some "artists" may try to sell them as NFTs or in other novel ways, for there are sure to be be fans of such crap, some of whom will realize they can do it themselves. Making them will be a lot more fun, but only marginally creative and the practice will become a time-waster. Sitting around watching an AI render images will suck up even more screen potato time than searching the Net now does.
I am glad there is some pushback to the creative IP theft, and especially the new obfuscation tools. LLM Generative AI will have a prominent place in our future, and it will take some lawsuits to try to reign it in.
For a book cover I played around with an art generator to produce a simple image, just silhouettes of two people. I had to iterate and tweak my prompt and settings a lot in order not to get 3D images, non-silhouetted ones, some totally different than what I was after. One even hallucinated an extra body part. It took a long time and nothing useful was produced. So I gave up and told my daughter to try it, and she rendered the prompt I texted her in less than an hour on her iPad. One more iteration and we were done.
And it pleases me greatly to credit my kid instead of that stupid software.
This is great to know. See a lot of examples of stolen AI images compared to originals at https://www.nytimes.com/interactive/2024/01/25/business/ai-image-generators-openai-microsoft-midjourney-copyright.html (paywalled).
One thing few articles about AI art discuss is how the ability to verbally specify a picture will affect artists themselves, besides the IP issues. IMO it will make many people call themselves artists, flooding the market with images that haven't taxed their issuers creativity and technique. Making art will become the same as wandering in an art gallery without the exercise.
When the bottom of the market for lazy images falls out, some "artists" may try to sell them as NFTs or in other novel ways, for there are sure to be be fans of such crap, some of whom will realize they can do it themselves. Making them will be a lot more fun, but only marginally creative and the practice will become a time-waster. Sitting around watching an AI render images will suck up even more screen potato time than searching the Net now does.
I am glad there is some pushback to the creative IP theft, and especially the new obfuscation tools. LLM Generative AI will have a prominent place in our future, and it will take some lawsuits to try to reign it in.