Working on an AI to generate SVG vector images.

SVGs are awesome and currently unrepresented in the diffusion-based model landscape. We have something that produces pretty great results and we're working on the next version which should be even better.

https://vectorart.ai

Something like this is desperately needed, keep up the work. One thing that I immediately noticed is that all downloads are named "download.svg". A more descriptive name would be helpful.

Edit: Also, copying the SVGs to the clipboard would be nice. Download from the browser still sucks, and with web based SVG editors (like my own one, www.hyvector.com ) people can quickly edit the generated SVGs without having to go through the downloading and uploading process.

Re Edit: great job on hyvector! are you open for some sort of collaboration? any way I can contact you to discuss?

Just wrote a comment on Twitter. I hope I got the right account.

Great feedback, thanks!

Website requires signup.

I'm working on a service that generates SVG vector graphics using generative AI.

This is something current models struggle with but I've always wanted high quality vector graphics for my projects.

https://vectorart.ai

This is really nice, is this converting to SVG from Raster like similar services?

This is a really useful mindset as a programmer, but backfires in real-life as it makes you an anxious person.

I guess you need to compartmentalise into different kinds of 'trust'. The not 'trusting' you do with a computer is different from the trusting you do with fellow humans in daily life. They just happen to use the same word in English.

I trust the computer far more than the fellow humans. The computer will generally give a predictable output for a given input. "fellow humans in daily life" .... not so much.

Human variability and unpredictability can also be a source of creativity. But I understand you completely

I hope you never have to cross a street, or get anywhere near a car with a human behind the wheel.

Its terrifying; sometimes when I am crossing the street people accelerate, sometimes they slow down, sometimes they stop at red lights, sometimes they drive through them.

That's why I don't have a driver's license

Yes, but alas other people do.

> but backfires in real-life as it makes you an anxious person.

If I was to point out this is an approximation, or a tautology (it is only true to the degree that it is true, which is not (necessarily[1]) 100% of the time), would it make you anxious? And if so, do you think it wouldn't be possible for you to learn [1] a new approach so it does not make you anxious?

While achieving absolute certainty in code correctness is often impossible

Funnily enough, I program in real life and have also been anxious lately.

Clearly this is another example of Sora in action

I wonder what the prompt was!

hey, what's down for you?

updated, what do you think?

ouch

Thanks everyone for your feedback and support. I've been adding GPUs as fast as I can to meet demand and some of the earlier performance issues should be much improved.

Thanks for surfacing this. I'll work to improve this to reflect reality - your prompts and images are private and owned by you.

Reality is whatever is in your terms of use agreement. Who wrote it?

AI, of course

Haha, I wonder if we’re ever going to see a litigation over a EULA, where the service provider is trying to nullify some duty based on the fact they used AI to write the agreement which swapped Provider and User in a sentence.

Great era to be EULA hunting!

I'm sure it will happen. However, I'm also fairly confident that whoever is the first to try it will be ripped to shreds by the judge:)

Those AI lawyers play hardball