Here’s something a little different. I haven’t written or recorded anything for this week so I decided to share more of the “behind the scenes”. Lol. I started playing around with AI image generation around a year and a half ago. Since then the technology has come quite a ways. If you take the time to learn how to talk to the tools that create the images, you can really hone it into specific styles. What can come through with experimentation is sometimes also very interesting to look at.
This is a couple of hundred of images that I generated for use on my site. It’s kind of picked randomly for this video and I think there may be a couple of duplicates on accident in there. And of course the odd fingers and hands thing here and there. But weird hands never make it into anything final. Its just a fascinating thing that still happens sometimes. 7 fingers? Why not.
I aim to generate images that I personally feel have a bit of something in them. I do try to think about how others will experience them, but I also tend to fall back on what I personally like. So what you’re seeing was selected from probably thousands (?) of images that were generated and discarded. I think at least a thousand.
I learned through a lot of experimentation how to get a nice, older, 35mm film look pretty consistently. That film grain look you'll see more in movies from the 80s and 90s maybe. This is combined with a touch of Art Nouveau texturing and flair throughout. There are other styles that are woven in a bit, but those are the main one’s that create a consistent kind of theme throughout a lot of what I have used on the site and here.
The AI image generation tool I use comes with Chat GPT. It’s called DALL-E. It’s pretty fun, but it really can take a lot of experimentation and time to discover styles that you can create consistently and that you also like.
The music on this was supposed to go with Hypnokink. I made it specifically for that and then it didn’t sound right so here’s a chance to use it. I normally use just Garageband on Mac and then fix the sound in Final Cut on Mac so that the voice and music sound right together. More recently I have been expanding to see what AI can add.
For this music I have been experimenting with UDIO which is an AI music generator. It’s a fun free one. I got something that sounds good from that to start with as a foundation. Then I built upon what the AI created and aimed to make it feel more like a complete song. I used Garageband and Final Cut for that. So it’s not just cheating and using AI generated music. It still takes work and creative input.
Then I have also been using a language model called Claude Opus which is really good as a creativity helper. I use it kind of like a really smart thesaurus that has helped to expand my vocabulary a lot. It also has helped me to build language within certain themes like on Extravagant Velvet. For that one I focused specifically around language that is “elegant, opulent”.
Sometimes it will generate an entire wall of text that just has one word somewhere in there that is the one I’m looking for. Like “luminescence”. It’s a different way of writing that is more like having access to words on a palette like painters have paint for consideration.
I’ve found that you can use these tools to just be dumb and say, “computer make this happen”. But the reality is that in order to use the tools well, you actually have to expand your language. You have to understand what you’re talking to the thing about or you won’t have the words to get anything meaningful back from it. I think it’s pretty fascinating how people have picked them up as tools for creativity before a lot of other uses.
Karen
2024-07-23 03:27:16 +0000 UTCJulaine
2024-07-08 18:03:53 +0000 UTCHolly Holy
2024-06-12 00:44:15 +0000 UTCLy Aamutähti
2024-06-11 05:30:33 +0000 UTCAnna MarieArteem. Moye
2024-06-09 23:56:39 +0000 UTCSue
2024-06-09 07:57:28 +0000 UTCDuchess
2024-06-09 05:08:32 +0000 UTCJulaine
2024-06-09 04:27:01 +0000 UTC