Month: January 2025

News article here: https://www.theregister.com/2025/01/27/deepseek_suspends_new_registrations_amid/

Two possibilities:

1. Deepseek is lying and they just oversubscribed (you might have noticed they went viral with the excellent R1 open source model I have been using for the past week already on SmartPoi, Magic Poi and more)
2. Someone is DDOSSING the heck out of them because they are jealous. Is it the US govt? OpenAI? Random hackers?
   

Annoying

Whatever the reason (and I really want to know so I can know who to be angry at!) it is affecting my work.

Thank you so much to my Patreon supporters for their contributions, the money is already in the Deepseek API account so I’m just waiting for access to be restored(!) before I can continue.

I did get some stuff done.

Did I mention that SmartPoi is now (almost) ported to the ESP32? The only thing that is not working is the LittleFS uploading, and that is really an Arduino IDE problem – I am trying to find a work-around but currently you can use it (just fire up the SmartPoi-js-utililities and upload some images). Check out the ESP32_2025 branch – but maybe wait a month (end of Feb?) by which time it will hopefully be merged into main (the whole thing is now working on ESP32/8266 with the same code base!!). Recommended: ESP32 S3 Super Mini (available on Aliexpress). This is the actual chip that Magic Poi will run on! Don’t get it confused with the ESP32 C3 Super Mini though – that one is single core only (and half the price – I have some C3 here and SmartPoi works on there too.)

For this new ESP32 branch I have even re-enabled the UDP streaming (which you can try using the old SmartPoi Android app). The new Android app for uploading and switching pics works as well. I haven’t tested UDP yet but if you use a dual core ESP32 S3 the WiFi runs on a separate core to the display code, which means it might actaully work now (I disabled it due to stuttering images).

This is hard work

Once again I appeal for support! Please help me by joining Patreon (it doesn’t cost anything to join) and subscribe to a support option if you can.

Support me on Patreon.

I just got here, what is SmartPoi, Magic Poi and DeepSeek?

SmartPoi project – WiFi connected POV Poi

Magic Poi – Internet connected upgrade to SmartPoi (work in progress)

DeepSeek AI – Right now the world’s second most powerful large language model when it comes to coding. I use this extensively to develop open source apps faster. Cheaper than ChatGPT but not free – hence the Patreon 😉

UPDATES:

Sign up for our update alerts:

Well 2024 was a good year, I feel like I got a lot done – mainly with finalising the SmartPoi project enough to be usable for December (it worked, yay!).

What is coming this year?

TLDR: doing the same thing as last year only better, faster – I just got distracted by some monkeys this month.

Right now we have one prototype with a lot of issues to address. The factory gave us the wrong buttons for example and one trace wasn’t connecting for some reason… Anyway I disabled the buttons (there is a web interface after all) and so we have something to test on while getting the next circuit ready.

So yeah I think for me it’s about getting the poi working seamlessly, and improving the user interface – I thought the buttons were going to be enough but after looking at all of the options (Multiple Timelines, Sequences, sync with friends, brightness etc) I was clearly being a bit optimistic. I recently saw someone came out with POV poi that has a screen on it (!) – is this what we need? Sounds like a battery hog to me so probably not but anyway I am looking into it.

With the web interface we still need timeline sharing, so that has to be first priority (it is directly related to the poi buttons, though – my plan was to sync with a button press originally). Also more ways to share images and maybe rate them or something.. I am rambling here but you can see there is a lot to look at.

A bit of a delay..

I have been a bit distracted so far this year, not just with the circus tricks – I got hired by a university to help with some research related to Monkey Detection (another speciality of the Circus Scientist). Never thought I would get paid to do this, but apparently South Africa is not the only country that has issues with monkeys conflicting with humans in urban environments. So I am currently working on a Machine Learning (Image Recognition) app for research purposes.

I also have a bit of a deadline this year, as my main “experimental” server running Ubuntu 20.04 is going to become obsolete in July and I need to transfer everything off of there onto a new one well before then. I know you can upgrade in place but some of the stuff relies on old libraries etc so requires a re-write anyway (or just to be thrown away) so I’m taking each deployed project and transfering to a new server. The new Alpha version of Monkey Detector is already on a newer version of Ubuntu Server thankfully.

So yeah

Give me a month (February is traditionally very slow for live entertainment here) and I expect to be posting more frequent updates to the Magic Poi project. Looking forward to getting version 0.2 out and moving closer to the Product Launch

PS:

I really plan on getting an actual product out by the end of the year, even if it is just a kit version. Please quote me on that, this thing has gone on long enough. I am ready, are you??