Week 49 - We become what we behold
While the world marches on towards syncopation, I’ve been heads down, preparing Scanner for open beta.
TL;DR
Shaping our tools
I’ve jumped over the first tiny hill on the way to product - shipping to a few folks, most of whom have provided really useful feedback, and some of whom have captured great scans of things and places I’ll likely never see in real life, but can walk around in VR now.
The next cliff awaits - that’s right folks, it’s open beta time! or, very nearly.
The usual conversations around business model have been thrown around over a few beers recently, and I’m deciding whether a service model makes the most sense, or whether I can make some kind of revenue off this as a direct product. The simplest plan of attack is to make Scanner paid (and super-cheap during Beta) and see what kind of interest I get before I decide. Lots of interest but no purchases will mean re-examining the way I expect folks to use the app. No interest at all will be a great sign that it’s time to move on to the next Evryway prototype! ;)
It’s still not quite ready yet - I spent a good chunk of time this week trying out some “improvements” to the voxelisation, and they work great under certain conditions and really badly under others. I have put in some confidence and quality filtering on the pointcloud data and improved the texture projection, which is a good start.
There’s still a million things I want to do, but I think I’ve narrowed down the list of things I really have to do to about 10 items, so I’m hoping to have something ready next week that’s good enough for general consumption.
Our tools shape us
My favourite (/s) news of the week was that Google have added texture projection into their Tango 3D Reconstruction library. I’ve not had a chance to try it out yet, but it’s possible (and indeed probable!) that the addition of this functionality to the API will render everything I’ve been working on obsolete.
Ain’t it always the way?
I’ve always known that someone else would create this tool (and probably to a much better quality than I have). It’s pretty galling to see something so fundamentally powerful - which has been missing from the core capabilities of the device for two years - added into the API without any prior notice. It’s part of Google’s black box libraries, so I can’t see the source for how it’s working to improve my own efforts, and if it’s decent, then that’s great for everyone else who wants to write a scanning tool, but annoying as all hell for me after spending a good few months making this work from scratch.
Still, nothing ventured …
I await results from other folks trying this stuff out with a-bated breath.
Voxelisation
I was intending to spend a chunk of time describing my efforts at improving the voxelisation, but I’ll leave that for next week - it’s still ongoing, and I need to take some pretty pictures. If you like dots, lines and cubes then you’ll love this.
Next week …
Next week I’ll be publishing, or damned, or both. I hope!