News for hackers. Stuff that matters.

IuMaker

I considered running the vectorization before showing any of the elements to the user, so the user would be editing everything as a vector. I didn't go with this approach since the results of the vectorization weren't consistently good for all the images. But I'm aware that the vectorization step is the step that requires more work, the problem is that it already has received most of it.

The main options for the vecotrization are: - Straighter lines / smoother curvy lines - Less noise / more details [Advanced] - Number of colours: simpler vector / more details [Basic]

I even looked for deep learning approaches to vectorization but they tool a really long time to run: https://arxiv.org/abs/2306.06441

Thanks for testing it, I knew that the advanced options didn't work on iOS, the problem is that they work for Desktop & Android (maybe it's a problem with the browser). I was hoping that it was only in my wife's phone, but I haven't been lucky.

The vectorization is probably where I've spent most of the time regarding creating a UI to tweak it. I already implemented two vectorization libraries, the results aren't perfect, but they are the best I could get with what I found. It would be great if anyone know a better technology (thanks in advance!). Great idea, using color palettes could fix the issue of having to tweak too many colors, I will put it in my TODO list!

Show HN: My Take on Logo Design: Leveraging Diffusion AI for Entrepreneurs

https://diffusion-logo-studio.gradientinsight.com/

Hello Hacker News community,I wanted to share a project born out of my own entrepreneurial journey – Diffusion Logo Studio. If you're a founder or entrepreneur, you've likely experienced the challenge of getting a unique logo for your small projects. Well, I've been there too, and I've got something that might help.What is Diffusion Logo Studio? Diffusion Logo Studio is the result of my journey...

( Read More... | 4 comments | 3 points | on HN )