A comment on an old project made me rethink how I use AI for art

Back in 2020, I was really into making AI art with early image models, just typing in simple prompts and seeing what came out. I posted a piece online that I thought was cool, and someone replied, 'It's neat, but it feels like you just asked for a generic fantasy landscape. Where's your hand in it?' That stuck with me. They were right. I was just a passenger. So I started feeding the AI my own sketches as a base, even bad ones, and using the output as a new layer to paint over and change. Last month, I finished a piece that started as a pencil drawing of my local park, ran it through a model, and then spent hours digitally painting on top of the result. The final thing actually feels like mine. Has anyone else moved from just prompting to a more mixed, hands-on workflow with generative art?

3 comments

Log in to join the discussion

Log In

3 Comments

dylanwilliams5mo ago

That comment about being a passenger hit me hard too. I had the same thing happen with a portrait series last year. Just typing "cyberpunk woman" over and over. Now I always start with a photo I took or a quick 3D block out. Lets me fight with the AI over specific details, like the shape of a building or how light hits a face. It's messy but it finally feels like making something.

2

jade2715mo ago

My friend got a similar comment and now she uses AI to fix her shaky hand sketches.

1

grantthomas5mo ago

Starting with your own work makes all the difference! It keeps you in the driver's seat.

9