Playing with AI models
AI models are fascinating, I've been building a few sites as well as extensions, etc that use chatGPT, Dall-e, stable diffusion, and others to accomplish various tasks.
One of those sites is Chef Brainy which uses chatGPT and specific prompts to analyze a list of ingredients a user enters and returns up to five recipes that use those ingredients, I've used this a few times myself to help plan dinner.
Another new site is myRoomAI which uses stable diffusion to take an uploaded image and generate how it would look with a new layout based on prompts people have entered.
myRoomAI has a few improvements and features coming, including adding a "lookbook" where they can add multiple rooms and arrange into a gallery they can share as well as being able to share their generated rooms if they want to or leave private. Currently, free users are limited to 5 uploads a day, and paying users can upload unlimited uploads a day, lookbooks will be for paying users only to start out, and will also feature bulk upload features.
Similarly, an update coming to Chef Brainy shortly will be to let users save their generated recipes into their own "cookbooks" to come back to later, as well as letting users leave a thumbs up or down and suggest edits.
Another niche site I've built is fixMyPhotosAI which uses the same framework I set up for myRoomAI but geared towads restoring photos. The result there have been interesting, and I look forward to adding more features for users to use as well.
Each of these are simple use cases but still handy and have been interesting to build on, they've also let me get back into next.js features such as NextAuth.js and integrating it with Directus, which I had moved away form for a bit.
I'm also using the new app folder for both as well, which has been interesting to build in.
There are a number of various uses for AI models and I can't wait to see where things go with them as we dive deeper into it.