How it works
Just tweet a black & white photo to @colorisebot!
What happened next?
A few minutes later, the bot was swamped with requests in the hundreds-per-minute range. After a few teething issues and working late into the night with Alex, we came up with solutions to the load problems we faced and were able to develop a far more error-tolerant service complete with Grafana dashboard backed by Prometheus metrics.
ColoriseBot is now hosted on Google Kubernetes Engine (GKE) thanks to weaveworks and as Alex writes, Stefan Prodan helped us integrate with Weave Cloud for monitoring and visualisation of traffic between the functions.
ColoriseBot also featured during weaveworks' CEO Alexis Richardson’s keynote at KubeCon. 💯
There is also now a dedicated ProductHunt page for ColoriseBot which got me very excited as this was a first for me.
ColoriseBot also popped up in Le Monde which was another pretty cool moment.
Créé par un jeune développeur anglais, un robot permet de coloriser n’importe quelle photo en l’envoyant directement à son compte Twitter. pic.twitter.com/noAzbYrAH9— Le Monde (@lemondefr) November 21, 2017
However, the highlight so far has definitely been receiving a tweet from none other than @MalwareTechBlog (“The guy who saved the NHS”). It’s not every day that one of your heroes tries out your service!
The other day, ColoriseBot also made it onto BotWiki, “an open catalog of friendly, useful, artistic online bots”. Nice 😁
EDIT 15th Feb 2018 ColoriseBot also popped up in the Family Tree Magazine, the March issue highlights it as “App of the month”. It’s great to see people colourising their old family photographs and unknowingly harnessing the power of machine learning, totally for free.
- Oli’s post covering the machine learning details 🛠
- Alex’s blog on ColoriseBot at KubeCon in Austin, TX 🇺🇸
- Finally, go upvote ColoriseBot on ProductHunt! 🔎
Colourise images yourself!
The code behind ColoriseBot is now available in the OpenFaaS function store, so just head over to your OpenFaaS UI, search
colorization in the Store popup and deploy it. Then you can paste in a url to an image and it will be colourised for you using machine learning and serverless functions. 😎
If you would like to get involved with OpenFaaS then please do, the community is amazing and I’ve had so much help and support from them. One of the most fun parts is always #faasfriday, a great chance to score yourself some limited edition OpenFaaS swag!