Show HN: I open sourced the QR designer from my failed startup | Hacker News Reader

Show HN: I open sourced the QR designer from my failed startup

github.com

711 pointskoch73 comments

My designer is somewhat special, if I do say so myself, as it allows you to put arbitrary designs in the middle area of the QR while still being totally scannable.

Open articleView on HN

Thanks for open sourcing this.

Reminds me of halftone QR codes http://vecg.cs.ucl.ac.uk/Projects/SmartGeometry/halftone_QR/...

I've been looking for this for years but did not know the right keywords to search for. Saw an Anonymous/Guy Fawkes styled sticker on CCC in 2018 or something, on and off I've been looking for how it was made. Now I know.

Semi related, one of the most useful functions of chatgpt I’ve found is describing something you don’t know the word for. It’s helped me find stuff many times that I would never have found otherwise just by knowing what to search for.

thought that often also works with just a search

Nice! I also found a quick write-up - with a link to a js implementation:

https://backdrifting.net/post/016_halftone_qr

https://github.com/fangj/Halftone-QRCode-Generator

Very cool! Years ago, I was also inspired by https://research.swtch.com/qart

I've been looking for that for quite a while, thanks for the link

Such a lovely and easy to read webpage

What product was your startup building, how did having a fancy QR designer benefited your product? Was QR designer the product? Why did it fail and what are your plans next?

Great questions!

Basically I was working on a link shortener. However in addition to a link shortener that took you to another page, I wanted users to be able to create a mini website that could host apps. Like you could create a small page that had photos of your business and a menu/ordering app and an "IM with staff" app etc.

The fancy generator was kind of an added bonus, I always thought the QR codes I'd see around were so dull and unidentifiable! There was very little in traditional QR codes themselves to make them "human-scannable", so to speak. A QR codes is essentially a sign; it should have some information in it that you, a person, can parse. A little logo in the middle is weak at best.

It failed because I didn't sell it. I needed to sell it. I needed to talk to people. I needed to like cold call people, restaurant owners without websites. Annnnd I always came up with an excuse not to.

As far as next plans, I'm working on something else these days - https://markwhen.com

This sounds interesting and if you're really "giving up" on the idea, perhaps you could "open source" everything. People could take it and make a "restaurant website in a box" that could deploy directly to GitHub pages, even.

hasn't UberEats, DoorDash and their ilk already done all of this all while being super helpful to the point the restaurants were totally unawares it had been done?

They’ve helped in some cases but in others it’s entirely confusing (or nonexistent).

Restaurant owners around here seem acutely aware that they're grabbing 30% of their revenue and actively routing customers away from the restauranteurs' own sites. Yes, there is room for competition.

I've noticed the same in the Netherlands anecdotally.. delivery services hated with a vengeance, yet restaurant owners convicted to using them nonetheless. (There was also a TV documentary some time ago, but about The Fork, and how Amsterdam restaurants resented them).

All restaurants hate TheFork. Same with Booking (which is from NL, coincidentally) and similar sites. There's no way to compete with them, and by being an aggregator, they have massive benefits in discoverability. Customers are just going to open their page to search, instead of using their search engine, so you have to be on their site, but they are filthy pirates.

>markwhen

you are building crazy cool stuff

Thank you!

> https://markwhen.com

Wow this is also awesome! Some programmers are truly creative people haha

Thank you!

Looks great. I'll be keeping an eye on it.

It's a hard sale, restaurant owners are fickle and cheap, not the easiest to sell to

For a good reason... Profit margins at most restaurants are razor thin.

> It failed because I didn't sell it. I needed to sell it. I needed to talk to people. I needed to like cold call people [..] Annnnd I always came up with an excuse not to

I can identify so hard with this it hurts...

build it and they wont come

If you can nail the timeline thing, I'd pay. I think about this all the time and it seems like something that should exist. I'd kind of like to use it to look at life events like "what week was I on vacation when I was a kid." Google Photos is great for this but I'd like to add my own entries.

I have similar thoughts, which is why I'm working on it!

https://markwhen.com - very cool. however, If I could share with you, I would see the value in following case: if I could connect my calendar(s) to it and see what is going on and overlay it with the data here in comment. Use case is both - for retrospective and for planning (for example if you're preparing the meeting and don't want to share content just yet, or jotting something for time in-between meeting what to do, etc)

markwhen.com looks neat! What tech did you use to make the editor?

Codemirror & vue. Chances are if you're using a code editor in a web app it's either Microsoft's Monaco [0] or CodeMirror [1], which is maintained by Marijn Haverbeke

[0] https://github.com/microsoft/monaco-editor

[1] https://codemirror.net/

Not even cold call. Cold pitch. Find restaurants with shite websites. Build them better ones. Walk in and show the QR code to the owner. Say you've built a better website for the restaurant. Have him scan it and see his restaurant much better presented, on his own phone. Say the first 3 months are free and then $200 a month.

"I needed to talk to people. I needed to like cold call people, restaurant owners without websites. Annnnd I always came up with an excuse not to."

If this is something you absolutely not wanna do, then not do it! Look for somebody who can sell, but almost impossible to find somebody.

Better built something, that you don't have to sell at the way you don't like. Speaking from own experience. This will be hard enough, but cold calls are nealry impossible for a one man show, especially with a low price product.

First, you did something. Nothing can take that away from you. So be proud of what you did. Next, you learned a valuable lesson that will come with you into your next projects. Learning to "sell" is absolutely critical for so much of life - not just tech startups.

Finally, this is not the end. There will be more ideas and more opportunity. You're already on to the next thing. Keep moving forward. It's not how long or short the game is, or how quickly you move. It's how long you stay in the game. Keep playing the game.

If it failed because you didnt sell it then maybe sales isnt your strong point? Find a salescperson to work with seems an obvious solution to this problem

markwhen.com can be useful for people who are looking to generate Gnatt charts and do not want to wrestle with excel. There can be many such use cases like a quick plot, a quick pivot table, a quick Gnatt chart, A fancy looking table for pre-sales guys & proposal writers to render them as part of the proposal. You should aim this tool at people making lot of proposal PPT.

If anyone is still hanging out in this thread I just fixed images so they work locally now

https://github.com/kochrt/qr-designer/commit/3ad0a4edea33d17...

http://robko.ch/qr-designer/

This is really cool. If you fixed the image upload, removed the 25 character requirement (less would be fine), and made the link customisable in the site, I think you could have a really neat product/business. I'd definitely send it to my startup's founder, we're already using QR codes and these would be way nicer than a plain one.

I went ahead and got most of your requests in - except for the image uploading, but you can now edit the url and it can be 25 characters or less [0]. The rest will be padded with spaces but browsers seem to handle that fine based on my 3 minutes of testing. Updated the live site too [1].

[0] https://github.com/kochrt/qr-designer/commit/55d4de3f856828b...

[1] https://robko.ch/qr-designer/

Nice, looks great. Sharing it around, nice job

Did the images too

https://github.com/kochrt/qr-designer/commit/3ad0a4edea33d17...

> Compared to other QR designs, this does not take advantage of error correction that QR codes use, but rather deterministically turns some pixels in the code pattern on or off without affecting what the QR code is pointing to.

Can someone explain how this is done? Is it adding characters to the URL that somehow browsers will ignore?

I assume there's a bit more to it, but QR codes inherently support a "mask" that tells the scanner parts of the image to ignore. So you could do at least some of that with the mask. It's not real flexible, so there's perhaps other tricks also.

I was under the assumption that the mask was just a short static pattern XORd over the data?

It is, the QR code spec details a few specific masks that are available. The one which produces the best QR code is chosen (best in this case means most clear/easiest for a scanner to read. So, no large blocks of white or black).

Wikipedia has a pretty good visual of this. https://en.m.wikipedia.org/wiki/QR_code#Encoding

> In order to maximize designer space, the url that the QR codes links to is expected to be exactly 25 characters long

Seems like a cool QR code, but also limited for general use.

For shorter URLs you could pad it with unused query strings

I could imagine this being part of an offering that includes an appropriately customized URL shortener (or if necessary, lengthener).

Yeah like the other replies said, it's meant to be part of a link shortener; or maybe you have a short domain that you can use short urls with. Like surprisingly enough I could fit a link to my github page in there as plaintext (although all caps)

I went ahead and made it so you can put your own urls at 25 characters or less [0], and updated the live site too [1].

[0] https://github.com/kochrt/qr-designer/commit/55d4de3f856828b... [1] https://robko.ch/qr-designer/

Definitely. I would love to be able to customize the design area and get matching amounts of data to be entered.

This is really good, I wish I saw it before I built something similar. Mine is server-side rendered at the edge. It allows any value but still has no where near the functionality as yours. https://github.com/jmcmullen/qr-kit

Very cool - I used a website with a custom qr code generator [1] and some hacky RPA tool about 8 years ago to create custom QR codes for each guest at our wedding. My wife created a wedding logo and we had that in the middle of our QR code - it worked well. The QR code was a personalised URL for each guest's rsvp which used a URL shortener [2] installed on our wedding domain (hosted on a free micro AWS instance).

Was a fun way to do my part for our wedding planning.

[1] https://www.unitag.io/qrcode [2] https://yourls.org/

It’s done in Nuxt, cool!

I have a huge interest in starting using Nuxt on my next web projects. Might even start earlier because of this.

How does Nuxt handle the backend part? Any recommendations?

I am used to how Rails does it by using ActiveRecord and it’s flexible ORM.

Nuxt and the Nuxt team are working on some incredible things in my opinion. I made this using Nuxt 2 which uses Vue 2, I would start with Nuxt 3 [0] which uses Vue 3 and has made a lot of improvements built upon a bunch of other open source libraries they maintain [1]. They essentially have a 'you can deploy this anywhere' model, from traditional node environments to edge runtimes like cloudflare workers.

The backend part was pretty straightforward for nuxt 2 because I could just use express. I think it's a bit different (maybe even simpler?!) for nuxt 3 [2].

I haven't checked recently so I'm not too sure what kind of ORM options they have available. I suspect there would be database plugins regardless, if it's not baked into Nuxt.

[0] nuxt.com

[1] https://unjs.io/

[2] https://nuxt.com/docs/guide/directory-structure/server

Eh, not a big fan of QR codes with images in the middle. One of my favorite things about QR codes is their extremely generous error correction capability, which is greatly reduced when you instead use that valuable space for a logo. It looks nice, but makes the code less scannable than an equivalently sized one with larger pixels or more error correction bytes.

Just a warning to users: light-on-dark QR codes can not be parsed by all QR code readers, even though they look nice.

sounds ridicules, please share some evidence

https://github.com/bennyguitar/BTCDonationViewController/iss...

https://www.qrcode-monkey.com/6-reasons-why-your-qr-code-is-...

Very cool stuff, thanks for open-sourcing this! Might incorporate it into Dub , an open-source link management tool that I'm building: https://github.com/steven-tey/dub

The tool is very cool, although at first sight it looks to me like the perfect side hustle rather than a startup… may I know what was your napkin business plan for such a startup?

I guess you put a lot of love in the tool. Freeware, abandonware and open source is sort of a tragedy of the commons for the IT professionals (yes HN, I know there is a global benefit, just no one seems to talk about the cons).

i have a lot of people unable to recognize QR codes as scannable. making it more picture-like would reduce usability?

Why not use the image to provide a better "scan me" visual?

Very cool could you share what features people wanted from QR codes? I'm looking into ways to improve https://t.ly/qr-code-generator

I really like the tool. I used some similar desktop apps few years ago, but they missed the simplicity of your solution. Almost a 1000 GitHub stars within one day. Really impressive.

Man I really wish I had access to this functionality ~ 5 years ago.

Thankyou for open sourcing this! Always nice to see stuff like this. I will definitely pass this onto some underclassmen.

Does the readme's last example have sections of one code embedded in the empty space between another code's sections? That's pretty cool.

I did not get the bottom one to scan on the last example with two different scanners. It might just be invalid/=.

The add image feature is not working for me.

Ah sorry about that, I'm not surprised... I think I uploaded it to a server first but now that it's all shut down there is no server for it to be uploaded to.

https://github.com/kochrt/qr-designer/blob/420adea59e617659f...

Fixed

https://github.com/kochrt/qr-designer/commit/3ad0a4edea33d17...

This is so cool. I think I've already got a project idea I can use this for

Why did your startup fail?

The OP explains it here: https://news.ycombinator.com/item?id=36130625

Reply on news.ycombinator.com