A magical recap

Happy Holiday season everyone!

What a year it has been. About 12 months ago we started preparing our Kickstarter campaign, which successfully launched in February 2017. A lot has changed since then. In total, it’s been more than 21 months of ups and downs and I’m incredibly happy to be a part of this awesome magic community with our little baby: Snapcardster.

Milestones of the Kickstarter Campaign 2016/17

In the beginning, we started out as a group of magic enthusiastic students, but the project quickly evolved into a serious case. Company foundation, software development and fundraising was suddenly on our daily menu. Every angle needs to work out, but with the right team around, everything seems to be possible. In the following I’d like to dive into some things we had troubles with:


1. Get things started as a rookie

After participating at the Prototyping Week in March, it was Harm Brandt, one of the founding members of the non-profit OpenCampus.sh that saw my eyes flashing when I told him about my idea. With the “Fast»Forward” program, Harm supported me to hire the students Nicolas Jessen and Malte Delfs in the very first months of the project and the team of OpenCampus, including Alexander Ohrt, Eliza Rottengatter, Eva Charlotte Koschinsky coached us to pitch, evaluate and find the perfect market fit for our adventure. Still, there were thousands of unanswered question marks in our plan.

From left: Malte Delfs, Peer Oke Richelsen, Nicolas Jessen

Frank Bock, from CORONIC Software & Security and Peter Schottes from Eisenschmidt Consulting Crew helped us to get the business strategy right and advised our team for three months. As young students it was the perfect assistance you could ask for.


2. Build a working Prototype

We teamed up with the supportive institute of Multimedia Information Processing, a research group of Prof. Dr.- Ing. Reinhard Koch at the CAU Kiel to tackle the computer vision challenges with three bachelor theses. Julian, Christian and Niklas wrote their thesis about our project and scored solid grades. Special thanks to Prof. Dr.- Ing. Reinhard Koch and Johannes Brünger who coordinated and supported the group.

From left: Peer, Julian, Malte, Nicolas, Christian

Later, Malte, Nico and I quickly built a mediocre prototype, just to realize that scanning cards is a bigger challenge than expected. In today’s software development, especially when it comes to visual computing and machine learning, the focus lies more on data than algorithms. The Problem: We had no data at all. While there are beautiful resources for magic developers such as mtgjson.com and scryfall.com they couldn’t provide the data we looked for:

real life images of cards, with all its errors of bad lighting, bends, different angles and low contrast

We had to start crowdworking. The idea was simple: We made a Kickstarter campaign to raise small initial funding (6.000€) and asked the community to send us correctly labeled images of magic cards to support us. As a compensation, each supporter received Snapcoins based on the rarity of the card (yeah, it was a non-crypto token back then).

Oh boy, we received a lot of data. Next up: training a machine to understand magic cards is similar to training a human. A manual process called supervised learning teaches the algorithm: “this is an island, this is a forest” etc. We spend countless hours of verifying the submitted data or corrected the information if it was incorrect.

In the end everyone in the team was incredibly happy that this stuff we made up in our head about crowdworking really worked out. Thanks to everyone who participated! You guys rock!


3. Fundraising

Seriously. Magic: The Gathering? What’s that? It’s not like the venture capital scene has any idea of what’s going on in the market, nor do they care to understand. We’ve heard the word “no” more often then playing against Miracles.

In the end, after putting in the work of fundraising, pitching and networking we were able to find seed capital from a swiss investor in mid 2017. The entrepreneurial company (Snapcardster UG) evolved and renamed into a limited liability company (Snapp.ai GmbH) to enhance the vision and grow the team size. Finally, we had the resources to start working on the iOS beta, enhance the scanning quality and improve the product.

In November, 8 months after the Kickstarter, we released the very first closed beta of the long awaited iOS app thanks to Julien Großkrüger who put in the work, day and night to rebuild the app from scratch for iPhones.

Still, the current beta has its bugs, problems, unfriendly UI and is far from being called mature. But after months of intensive testing, we have a pretty good idea of what to fix in the next upcoming weeks.


4. Spread the word

What’s the goal of building one of the best apps for magic if nobody knows about it? In the end, everything comes down to the market entry.

For this matter, we’re very proud to present you both Team #Snapcardster and our Pro Tour Team.

Team #Snapcardster

Read More

#Sponsored Players

Pro Tour Team

Read More

#Affiliated Players

Being a part of this awesome community with Snapcardster means, to give something back to the active players who craft this experience. I’d like to credit Andreas Petersen for his incredible efforts to build and grow the professional scene surrounding us. Thomas Enevoldsen, Michael Bonde from Team #Snapcardster and Christoffer Larsen from Team Revelation winning Grand Prix Lyon was my personal highlight of the season. After starting 0-2 they grinded to the top with fourteen consecutive match wins and won the Grand Prix Lyon 2017.

From left: Thomas Enevoldsen, Christoffer Larsen, Michael Bonde

5. Sharpen the Saw

The Next Web just wrote a perfect article about the current state of apps.

Apps aren’t dead, they’ve just evolved […]

[…] you only get one shot to impress your consumers.
The Next Web

Tl;DR: Apps need to look and feel mature to make a dent. Our first and only priority is to provide you the highest quality app for our most favorite hobby: “Magic: The Gathering”. We want to make sure to get everything right, before publishing our first version one point zero.

Until then, I invite you to download the open beta on android or leave us your mail address to become a member of the new closed beta on iOS.

Special thanks to everyone who already participated in the beta and gave feedback. You’re the reason that keeps us running!

If you want to follow the progress, make sure to head over to Twitter, Facebook or follow us on Instagram.

Thanks for reading and happy holidays!

Signing off,

Peer

P.S: We’ll be on vacation over the holidays and return with full speed in 2018. Invitations and support inquires might take longer than expected. Thanks for understanding.

Android Update 0.6.3: New Scanner

Hey dear Snapcardster User,

we’ve listened to your feedback and made some changes to the scanning function. At the moment the scanner is not always recognising the correct set, however we integrated an easy to use edit form to speed up the process.


Old Scanner

Too many steps

The idea behind the old scanner was to “scan first, edit later”. The problem with most scanners, including ours, is that they sometimes have a hard time recognising the correct set. To prevent adding wrong cards to your collection we thought it was a good idea to edit sets manually. We quickly realised that this process takes up way too much time.


We screwed up. It takes too many steps to add cards


New Scanner

Edit while you scan

The new scanner features …

(a) an offers toggle to instantly add your scanned cards to your offers-list,

(b) a foil toggle to add cards in foil,

(c) a timer to add multiple cards of the same name,

(d) an edit button to make changes to the scanned cards,

(e) a “+10%/-10%”-button to apply a percentage to the recommended price.

The interface also shows the latest recommended price while scanning. This way you can just scan cards to check out the price, without adding the card to your collection.


Add cards faster with a simpler UI

We’re confident this update takes the scanner in the right direction. Our vision is to build the best magic card scanner on the market that recognises every card attribute except the condition. To recognise the set we’ll be introducing a new scanning algorithm in the near future. Until then you need to edit the set manually, if it’s not the right one.

If you find any bugs or if you have a feature request, feel free to:

Send Feedback


We’re listening!

 

Signing off,

Peer from Snapcardster

Update 0.5: Introducing Offline-First

Heelloo, dear Snapcardster User.

I know … it’s been a while since our last update, but we haven’t been lazy. With each update, we receive lots of new feedback from you, regarding common bugs, feature requests and more. So first off: Thank you! <3

Introducing: Offline-First

With todays update to v.0.5 beta we’re introducing an offline-first approach.

From this day, you’re able to add any magic card to your collection, without internet. Once you leave the magical basement of your local gaming store your app will connect to our cloud service and sync with the newest prices from CardKingdom.com (new in 0.5) and TCGPlayer.com.


Adding multiple cards

Rebuilding the core structure also allowed us to add multiple cards. 

This was one of the most asked feature requests and a function, that is mainly used to add playsets (4x) of cards or lot’s of basics for a new deck.

We hope this feature will speed up your collection management and deck-building up.


International Currencies

With this new price- and data structure we also added all kinds of new currencies from all over the world. Accessibility is what we care about.

Btw. we’re looking for translators. Hit up with Miri (she’s the mastermind behind our instagram and twitter) if you want to support us with your translation skills.

Anyway. I hope you like this update. This one has been a rough one, but I’m confidant it was the right step and we got it done. As always: you’re invited to report any bugs & feature requests.

Send Feedback


Signing off,

Peer from Snapcardster