Starting the OpenStreetView.io project

Hello!

I’m starting the OpenStreetView.io project, a collaborative 360° street view imagery project, in which I will put a significant amount of energy. I believe that street view photography is something necessary, more than just a fancy addon (e.g. see that story about the Strava users).
I believe that until now, there was 2 major hurdles in any street view initiative:

  • Lack of affordable 360° cameras
  • Lack of scalable storage in a non-profit environment

I believe that recent technology will overcome these 2 hurdles:

  • 360° cameras are becoming affordable. I recently acquired a 360° camera, and the quality is quite good! In a place I tested, the quality was not as good as the quality of Googe imagery, but better than Bing.
  • Decentralized cloud storage systems are coming: StorJ, IPFS. The idea would be to run our own instance of a decentralized cloud storage, with people donating hardware space, and let part of monetary donations be automatically redirected to storage donors (StorJ/Ethereum/…).

All this project would remain strictly on a non-profit basis, i.e. no VC funds or whatever.
This project would remain strictly on 360° street/path view photography, i.e. no traditional pictures.
I need to look into it, but imagery licence would stay as close as possible to the spirit of OSM.

The very first hacking is being viewable at www.openstreetview.io, code being at github.com/nand2/osv-frontend.

The short term goal (end Q1) is to be ready to take contributions, and to have a part of Paris (one or two arrondissements) fully mapped.

I will keep posting updates over here.

Cheers!

Great job you are to do.
Are you doing this by yourself ? Alone ?
Is this like https://www.mapillary.com ?
It will be hard to contribute what/ which to chose.
I am using mapillary in order to contribute openstreetmap.

Thanks!

Yes I am aware of Mapillary, and I have two reasons why I’m not contributing there:

  • I am focusing only on 360° pictures.

  • The pictures we contribute are not really free and openly accessible.

In the Open Source business model, we have usually in one hand the open data / source code, and on the other hand the companies selling services on top of them. Each parties are separated. For example, OSM/Mapbox, Drupal/Acquia, …
But in the case of Mapillary, there is a conflict of interest : the open data, our CC-BY-SA pictures, are stored in cloud storage paid by Mapillary. That’s nice of them, but then it’s not really openly accessible.
Mapillary is a VC backed startup ($1.5 million invested by Sequoia). And Sequoia will expect that money back, ideally multiplied by 10, later on. So if, let’s say, another company want to provide services on top of our contributed data, there will be a conflict : Mapillary will make them pay for our contributed data (current prices are $99, $999 and more), and could even theorically deny them access if the services the company provides threaten them.

With OpenSteetView.io, I aim at making a independant, non-profit, scalable repository of street view 360° pictures. If companies, even Mapillary, then use that data to provide paid services, no worries, that would be awesome. But at least the data would really be free, and free from any conflict of interest.

Why was it not done before? Until now, the main hurdle was the lack of affordable storage for potentially thousands/millions of terabytes. But new technology is now emerging (StorJ, IPFS) that will let users contribute hard drive space, and even pay them for electricity costs with moneraty donations (StorJ, ethereum). So we now are in a position of storing that data in a non-profit.

Finally, yes I’m starting as a one man operation : I am going to fund myself and dedicate 2 weeks per month on this project, while working at a regular job the 2 other weeks. I will definitely be looking forward at contributions later on :slight_smile:

NICE !!! I hope you get enough support :slight_smile:
P.S. : just my opinion ; why not another name instead of “openstreetview.io” (sounds like googie, and maybe “confusing” for some people), like for example “openview” or something else :slight_smile:
Also , there is kickstarter to raise some funds ; https://www.kickstarter.com/learn
:slight_smile:

Thanks :slight_smile:
Yes agree on the name (close/confusing to OSM & GSV); will maybe change later on but for now I’m focusing on the code itself.
Ahah yes kickstarter, but I don’t need any fund for now! One scenario I could see that is if I would like to offer lending of 360° camera for free for top OSM contributors – I would buy let say 10 cameras, and give them for short periods to willing volunteers (in exchange of a bond) so that they can play with it.
Otherwise, time is all I need, and time is what I bought myself.

OpenStreetView is infringing both trademarks if it is confusing. Since you acknowledged that it’s confusing, you are acknowledging that you are infringing the trademarks, and have opened up yourself to a lawsuit. Sorry to bring bad news. I advise you to change the domain name immediately, and stop using osv.io.

Who makes these 360° cameras?

few “suggestions” ;
osmcam
osmview
osmsight
osmorama
opensight
etc… :slight_smile:
here you can check if domain names are already taken ; https://instantdomainsearch.com/

EDIT : on the other hand, i found this recent article ;

… or “osm360.io” ??

Ok ok, this plus that, I will make a name change, will need to think about it…

The camera I am using is a Ricoh Theta S, ~€350, which has the advantage of having an API, and being a second generation (learnt from the first one). And doing video as well. But I can’t tell for the others.
There are also plenty of Kickstarters, but as always, be careful with those…

Thanks for the suggestions! But I think I will stay away from putting the “osm” word as it would induce that is has been started by the OSM foundation, which isn’t the case. So I will need to be creative a bit…

I’ll be back to coding in ~10 days, for a 2 weeks overdrive, and I hope to be able to finish a quick 3D print to put my camera on my helmet + a TI sensor bag to capture orientation of the helmet + the android phone triggering pictures and storing GPS loc :wink:
Or maybe I will first prioritise first the script that will take as input the GPS track, a serie of 360° pics, and an optional serie of orientation data, and then output the pics linked to the OSM data.

@nand interesting project. However besides the name problem the other legal issue is privacy and you are going to have to address blurring at least faces, number plates and potentially whole buildings in one way or another.

Simon

Yes! At the beginning, I will very probably follow what the big street view imageries are doing, i.e. blurring face, plates, etc, because they probably do it for a reason.
Then, when the project actually start to live, I will take time to figure out why exactly are the regulations/issues and finetune the privacy settings. It will be an interesting legal area to discover :slight_smile:

On others news, just added the AGPL licence, and will split the projects in separate projects (JS frontend lib, backend data crunching, website), and put their appropriate licence, once I get back on the project in a little more than 1 week. Can’t wait to start some data processing with the OSM data!

Are you aware that the Ricoh Theta has orientation and compass data? It’s in the EXIF data of each image, but the compass is rather low quality, as in many smartphones.

Thanks a lot! Indeed I digged a bit a found something about it. Tried, but indeed at first glance the data seems not very high quality… the reason I guess they didn’t make it public.

There is one big problem with panoramas: if quality is good, then loading time of a single image is significant. You can chceck it here: http://im.pwr.edu.pl/~polowcz/ZamekCisy/ That was a reason I’ve stopped making panos few years ago. Nowadays there are techniques of loading pixels in some order, which allows to display bad quality image really quick and improve quality later (dynamically during loading), see here: http://www.viewat.org (select any pano and zoom in) - there are also better solutions.

I can already give access to my original Ricoh Theta S pictures I made for Mapillary (as filipc).

Thanks philippec!

On the project iself, unfortunately, the time I though I had secured to be available is no longer available… already… sad face… :cry:
Still, the code stays in github and do not lose hope to gain time again in the future; and meanwhile hopefully someone will fork and take over : this is a huge project, especially the StorJ and decentralized-storage-funded-by-donation part!

Thanks all for your feedback you made on the project.