Check out the source of CropperJS

[felquis]

I was using this library cropperjs, with react-cropper I just wanted to make sure you know these options.

It works great on several devices, has lots of options, I found an issue that I could not use touch to scroll down my page, but besides that it is very good.

Now I'm going to use your library as it was recommended by the former team so, consider this issue an email.

Anyways, better than to give the user the ability to crop the image, is to just understand the image and auto-crop it, I'm not sure if we can use like https://js.tensorflow.org (old deeplearn.js) to do just that in the client-side with no server calls. That would be the perfect solution.

Thank you.

[brizental]

Hey @felquis , thanks for the tips, and nice to know that you are going to start using our solution! I will take a look at the resources you mentined.

I never thought of the 'auto-crop' feature too much, because my initial idea was for the user to mark each point of the cropper polygon. Could you maybe elaborate on that a little further? Are you thinking of something kind of like Photoshop's magic wand tool, where there is a little user input and a little intelligence from the software? Or are you thinking about something completely autonomous?

Also, if you have any problems upon implementation, please let me know :)

[felquis]

Hi :)

I never thought of the 'auto-crop' feature too much, because my initial idea was for the user to mark each point of the cropper polygon. Could you maybe elaborate on that a little further?

It is a trend in technology, to make the software intelligent because most people are not. Any intelligence you add to the software that help the client-side make sure the image is a license or at least that it looks like a license, will already prevent pictures that does not make sense. (nudity, anything not related to a license)

Are you thinking of something kind of like Photoshop's magic wand tool, where there is a little user input and a little intelligence from the software?

Yes. The goal is to remove user input completely (try it the best) and make the code do most of the job.

The ideal is to auto-detect the license, crop it, and if it is okay for the service, give a button to the customer "Is that your license? Confirm YES | NO". If we could not find the license, just say "Please, take a picture {put instructions here} and try again"..

Or are you thinking about something completely autonomous?

That must be the goal, we need to baby step until as close as possible to it in high end smartphones and fallback to a basic cropping in low end smartphones, we may have some problems with browsers support too. I don't know, I'm the dreamer right now hahaha

---

Remember that in Brazil we're going to have a new license model by next year, but the current one will still be used for a while. http://www.cidades.gov.br/ultimas-noticias/5425-carteira-nacional-de-habilitacao-passa-a-ter-formato-de-cartao-inteligente

That is it, just a LSD trip about what is the most ideal experience, hope you consider it as something that can be real if we put enough effort on it.