▲ 148 points
back
45 comments
It's nice at least a few dead Google products get open sourced instead of deleted outright. I have to imagine this is only possible because it's pretty much entirely a client-side desktop application that doesn't rely on Google servers for much.
Exactly why building apps for the cloud is a big downside.
I don't think "Cloud" matters here. Apps are not created from vacuum, they may link to proprietary code (a lot), not to mention infra dependencies.
E.g., a Google Reader like thing:
* Google internal library dependencies * Internal storage systems like BigTable * Borg configurations * Service dependencies: group management, access control, telemetry....
Basically, the closed enclosure of a service is just too big to open source.
Those are essentially what I meant: Google building on Google infrastructure makes their shuttered web apps nearly impossible to open source.
An app you download and install from Steam that runs on your PC... pretty straightforward. They have to remove the SteamVR library, but it's available for you to plug it in yourself.
Even client-side apps can link to lots of proprietary dependencies or require internal tooling that a company is not willing (or sometimes just not staffed) to opensource.
Yep, looks like they put in a non-trivial amount of work replacing proprietary libraries in order to make this possible: https://github.com/googlevr/tilt-brush#systems-that-were-rep...
there was an effort at one point to write internal things on (internal) google app engine, which would make open sourcing more straightforward, but I don't know how far that got.
It's the other edge of the monorepo/standardized tooling sword. You get nearly unlimited access to the work of people who have come before you, but separating your work from the internal commons is no small task.
I suspect most Google projects would be non-trivial to open source (even without legal concerns). Building software with Google's internal tools is very different than what's popular in the open source world.
It cuts both ways: Many open source tools would be difficult to use at Google, because they presume open source tools and package managers that don't have equivalents in the Google codebase.
> that doesn't rely on Google servers for much.
My understanding is that it used to interact with Google Poly. But a couple of months ago it was announced that that is shutting down. (See https://news.ycombinator.com/item?id=25277615 for the HN discussion about that.)
Yeah, Poly was like the content store, but it wasn't fundamental to the app's functionality.
On the other hand, cloud software is often more convenient to use, allows for syncing on multiple devices, and is more powerful due to having access to the computing resources of large data centers. So you can't really say that building cloud apps is only a downside. Tilt brush just happens to be the type of application that doesn't overlap much with these benefits, making it easier to open source. Also, cloud software isn't much more difficult to open-source, see self-hosted open source applications like NextCloud: https://nextcloud.com/
> Also, cloud software isn't much more difficult to open-source
What's the open-source, production-ready, 100%-API-compatible implementation of AWS SQS or Google BigTable?
Genuinely curious, because I've been down these roads and they're not simple.
I have to imagine this is only possible because it's pretty much entirely a client-side desktop application that doesn't rely on Google servers for much.
Tilt Brush models could be shared through Google's "Poly" 3D asset library, but they're shutting that down too.
There is a good discussion of this also with ama from one of the developers live on Reddit :
I just want multiplayer Tilt Brush so more than one person can paint the same scene at the same time.
Amen! I wonder if such a thing might happen now that it's open sourced?
Already being discussed on the Discord for Icosa (the open replacement for Poly)
EDIT - apparently someone has already got basic multiplayer working: https://twitter.com/RendeverHealth/status/135427119807556403...
Gravity sketch does this in their paid plans
This is awesome. I had so much fun with tilt brush, but felt that there were just a few things I'd change. Maybe I will now after I get done with all the other things I'm doing...
I might actually be able to use it (with these modifications) to help me with some of my other projects :)
I was curious what they deleted, so I clicked through to the build guide, and then I got distracted by their trademark conditions (you can't call the software you build "Tilt Brush", which is weird). Tilt Brush's trademark info page then contains this gem:
> Don’t use the term Tilt Brush as a verb, or use it in possessive form.
Oops. That just makes me want to Tilt Brush even more.
It's not that rare. Mozilla does the same thing with the Firefox trademark.
I remember Debian getting caught in the middle of that, and awkwardly having "Iceweasel" instead of "Firefox" for a while.
You should do so, in your own personal build of Tilt Brush.
Several original devs and various other interested parties are over on Discord discussing future plans: https://discord.gg/GsnAKqwK
Hope they plan to do the same for Google Blocks too, their low poly VR 3D modelling software.
Seeing how Google took on VR now makes me frustrated when they enter any new tech. They put so much effort in right at the beginning then backout out almost instantly but crowded the market to prevent other companies building these tools. Should have been clear for anyone back then that VR was a very long term play and didn't really become viable at all for 95% of users until stand alone headsets could be made.
Nice, although with Google not maintaining it anymore I wonder if there will be enough community interest to spin it off as a real project rather than just an archive.
Already about 10 forks. One with a binary build available and a WebXR port in progress.
There's a very passionate community around Tilt Brush.
Nice! I was going to spend the 20 bucks on this on Steam a couple weeks ago. I'll wait for an open source binary to be created and run it that way then!
Open sourcing it is good, but without a formal successor it looks like the community will need to settle on a given fork for future development.
The developer most willing to continue active development and support usually wins. No reason multiple devs won't eventually merge their efforts too.
Yeah, it's just potentially a pain for people who paid for the Google version which y'know, even got discounted like two weeks ago and frontpaged on the Oculus sale!
I hope the Oculus Quest gets a new build of Tilt Brush app (with a different name) but offered at no cost [1]
[1] https://www.oculus.com/experiences/quest/2322529091093901/
This is unlikely to happen on the official store, but will probably be available for Side quest at some point. Your best bet is probably still building it yourself
Always thought this should've been open sourced as soon as google acquired it. The mechanic was copied in other VR games and it would've been a good reference/learning point for other VR devs. It seemed odd for google to charge for something like this when it could've been used to share and inspire new ideas. Glad they got around to it.
I'm actually surprised Google doesn't charge for more of side projects like this... revenue of some kind can often justify continued investment, whereas most of their random free app projects are shuttered quickly because they aren't monetized easily.
Also, a lot of VR projects, Tilt Brush included, had price tags in part because they were built with exclusive deals with hardware manufacturers as bundled launches. It's hard to say your VR headset bundle has a value added deal of a piece of software that you then give away for free.
My very distinct impression from companies that built exclusive VR titles for new headset launches, is that the revenue guarantees involved were big enough for that to be a reason in itself to build a given title. CCP Games built four... Valkyrie was originally exclusive to Oculus Rift and Gunjack was originally exclusive to the Samsung Gear. Gunjack 2 was then exclusive to Daydream. And Sparc started as a PlayStation VR exclusive. Apart from EVE Online, CCP hasn't made any successful launched games except for VR titles... likely profitable solely because of those exclusivity deals alone.
Looks like the community will need to fork it in order to port it away from the proprietary Unity engine and SteamVR, maybe to Godot and Monero/OpenXR? Is there a fork yet?
Support for runtimes other than SteamVR is already documented. It was taken out of the build in a way that's trivial to add back in.
However - I can't see anyone bothering to rewrite in Godot. It would be a huge task and Unity is very popular with a fairly liberal usage licence.
uh...really? i think this is much too difficult to assume it would happen in the first 2 years after open sourcing, much less the first day
It was pretty soon after Synfig Studio (2D animation tool) was open sourced when folks started working on the code.
Tilt Brush is a much less obscure tool than Synfig was at the time, I assume there are lots of folks who would jump on this.
it's too bad they are closing down Poly, the gallery of creations with tools like this, but it is nice they are open sourcing their code.