I had to make it clear in the title itself that the project is vibe coded to adjust the expectations. It still has known and unknown bugs.
I decided to create it for fun. It is fun to work with the AI to solve problems that I have no idea how to. And by solving I mean delegating the solving part to the AI. I have never worked so much with canvas on the browser, and I still do not because it is the AI working with it LOL
When it comes to the image editing aspect of it it is simplistic because I myself am not an image editor and don't have many creative friends to help test. What I do have is some friends who are almost like memelords, and to edit meme templates it has what is needed, an eraser (to copy and edit someone else's memes lol) and a text tool.
It also has some basic shapes drawing capabilities and some standard filters and image effects that are lice. I like using the text and adding an outline to it when editing memes or captioning an image, so it works for basic use cases.
But that is not the most interesting part!
Now comes the nerding out part.
I've said I am building this for fun, so the "product" side is a little bit neglected, I need to improve the usability and functionality and fix bugs, but where I actually have fun is in the infrastructure side.
Picsuba has what I consider very advanced and clever CI/CD pipelines
It is a monorepo, it has many services, most are not ready for release but they exist and they will feature single sign on, so I think it is quite important that builds are safe and properly built and tested. The pipelines do not yet include smoketests or anything but I am planning to add those.
And most interestingly it has per PR environment deployments. As a monorepo yes I know it can get computationally expensive to have branched environments but so far it works and at most I have had over 20 PRs each with their environments and almost 10 containers per branch, and the VPS handled it all. Granted, if I had real customers and users there is a risk their experience could suffer, but when that problems arrives I will think about scalability
For now it is just very cool to delegate tasks to multiple AI agents, have them creating their own PRs and the pipelines auto deploy their branches so that I can test how it all works in the end. As you can see multiple services are deployed per PR so I can really do E2E and integration tests myself to ensure little to no regressions and that it is good enough for humans (AIs can't properly judge some usability quirks).
The next steps definitely are related to adding testing and I hope to be able to finish cloud functions for uploads. Right now everything is handled on the browser only.
If you are willing to, give it a shot and let me know what bugs you've found and which usability issues you've encountered, there are many and hopefully by getting new people to test I can address and focus more on the product side of it.
As a quick note: the bug report was made recently, if you use it let me know because that too needs testing. On my tests it seemed to work properly but I thought it would be worth noting, in case someone files a bug report over there I don't want to risk missing because of a half baked bug report feature.