Show HN: Pixibot – Slack bot that makes text in posted images searchable | Hacker News Reader

Show HN: Pixibot – Slack bot that makes text in posted images searchable

pixibot.co

88 pointsrgbasin14 comments

Thats neat, just installed it to test. We sadly don't post much stuff that needs ocr'ing, although the occasional screenshot here and there is useful to ocr. Sadly I don't see that we would ever get over the 100 uploads a month limit.

In an ideal world this wouldn't even be necessary. This just speaks to the sad state of image format adoption and workflows. Every browser supports basic SVG. SVG text can be read by screen readers, indexed by search engines, found in the page, copied, resized, machine-translated, and everything else you expect from text. Why are we still taking raster images of text? I think the onus should be on the OS and browsers to implement snipping tools that can save in SVG. Regardless I still think this is a cool project for making that lost information accessible.

As much as I would like to have this, in addition to your point about browser support, there are reasons I can think of from the top of my head:

There isn't widespread support in image editing programs to simply add text to a jpg/png/whatever, then save the result as an SVG.

Most people don't see any motivation, as most people don't have a lot of text on images.

A lot of images with text are screenshots or photographs, rather than with overlayed text.

Upload services for images usually don't support SVG.

Yeah unfortunately I have to agree that the workflow just isn't there yet. (definitely a chicken-egg aspect to it) I can kinda see why, editing SVGs is completely different from editing raster images. (the former is pretty much code) The only reason I can see why image hosts don't support uploading it is because SVG has had some security issues. (don't know if this is still a factor) As for your point about most images not being text, I agree, but SVG does support embedding raster images so it would only be a small cost that yields all the benefits mentioned.

May I ask, what OCR engine does it use?

It makes use of an ocr.space API

Thanks. That is the one at https://ocr.space/ocrapi ,or?

is there more information? mobile version of website shows a giant logo, and if you scroll down it shows a static image of a slack channel [1]

Using Android/Chrome

[1] http://imgur.com/n5f6bAG

Thanks for pointing that out... this link will help with the missing gif!

http://www.pixibot.co/static/img/pixi_intro.mp4

Hey! I just fixed this yesterday for Wishy.gift (https://news.ycombinator.com/item?id=13159528).

You need `muted` for videos to autoplay in Chrome/Native browser on Android, elsewise it just won't play. Try it out :)

edit: Oh, btw. Pixibot looks slick! Very cool application of OCR.

Hey thank you!

It might be a better UX to post a snippet since those get folded to a short preview.

Hey Ron. You're on Product Hunt. https://www.producthunt.com/posts/pixibot

Thanks Faisal! Mind adding me as a Maker? Just sent in a request

Reply on news.ycombinator.com