I've been playing around a lot with Stable Diffusion recently, and I thought it would be fun to showcase this previously impossible use case: business logos that are generated daily.
The implementation itself uses Depth2Img (although ControlNet might be the next step) to generate an image from the logo, as well as some post-processing/masking to clean it up.
Great point. I might need an entirely separate landing page for the artist/general audience vs the prosumer type landing page that currently exists.
(If you’re interested, the gist is that custom models allow for completely distinct “styles” as well as unique characters. For example, if you wanted to generate art in the style of Monet, you could train a custom model in that style)
It’s a great point. I had been using BunnyCDN to optimize the images/serve as webp, but there are a few on the model preview page that I definitely need to shrink further.
I got pretty into Stable Diffusion soon after it came out. Like a lot of users, I tinkered around with different ways to run it, going the usual route of running on my weak local machine, then going on to runpod, then implementing my own custom solution.
What I came up with worked pretty well for me, so I created a site that allows users to upload custom models and run Stable Diffusion “in the cloud”.
I launched in early December and it ended up being more successful than I expected. I just got to $700 MRR, which I’m definitely happy about after years of side projects making exactly $0.
Great work! I just wanted to point out that I'm seeing "Plans start at $xx.95/month" on the homepage. Wasn't sure if that's a typo/something you meant to replace.
Thanks for checking it out :) And I agree with you - it's definitely a small use case. I think it would be most helpful for companies with a small dev team who want to implement a favorites list, but may have other, more pressing things on their backlog.
It's an idea that has been in my head for a while. I like the idea of small feature-based APIs that can be used without requiring much dev time.
If they like it enough to sign up/give you their email, what’s preventing you from sending them an update? Or sending them a simple questionnaire?
I would be careful when considering an upvote a positive signal. At the best, it’s someone showing they are just on the line of being interested enough to sign up. At worst, it’s just empty validation for an idea that won’t work.
Messaging apps like Signal or WhatsApp are not providing content in the way that you describe. They are just carrying messages from one user to another, and would be rendered useless by this bill.