Hey people! Our OSS project https://github.com/lost-pixel/lost-pixel has been very welcomed on Hacker News so we pressed on and built a cloud version of it! Presenting it here today for your feedback and as always would be happy to chat & discuss Visual Testing with you!

The "getting started" directs me to give permissions for Lost Pixel to "act on my behalf" on my github account. I don't feel comfortable with that.

Github has poor user permission features. Apps often have no choice but to request for broad permissions even when they don't need them. Github really needs to improve on that.

We have improved our FAQ section to make sure it reflects why we need the permission. We have really tight github integration and creating status checks on github is impossible wihout GH permissions. We ask for the bare minimum!

One quick comment is I had to go to your github repo's readme to actually see an example in action. You should have a similar graphic showing the scanning right in your landing page head.

Good that we have it at least somewhere discoverable! Thanks for the feedback, we will improve on this!

What would actually make you understand that immediately?

I made one of these for Power BI that highlights visual changes in dashboard versions, and TIL it's called Visual Regression Testing.

Thanks a lot for the feedback, we should do better job at explaining the benefits of the solution to the people not familiar with it! We will improve on this

Can you speak to (the lazy among us) how this compares to the existing visual regressing testing solution[0] currently offered by Chromatic, the company behind Storybook? Thanks!

[0]: https://www.chromatic.com/features/test

We are different in a way that we offer holistic solution not really bound to storybook. You could do lots of other visual tests with lost pixel e.g. make tests of your react or vue pages or even crazier, test your results of playwright and cypress tests!

Was there something else just like this on the homepage yesterday, or did I see it elsewhere? I recall seeing some kind of AI tool that did auto testing to see what UI aspects of a website broke, if any, after new code is pushed