GPL v3? Huge mistake. Permissive Open Source licenses are clearly winning in the intellectual marketplace. GPL v3 is particularly problematic. See for example [0].
Well, the main software components are GPLv3. But there's also hardware released as CC-BY-SA, some modules as Apache v2.0.
I guess you are thinking of it as a library.. In that case you can argue that a copyleft license would restrict the possible adoptions of the system. But this is a framework composed of many modules, most of the time you will interface this system with your own stuff using agnostic protocols, so.. what would be the legal problem for a closed-sourced business to use it?
Most businesses don't want to touch GPL code, particularly GPL v3. By licensing it as GPL v3 the team is basically saying "no one but students and hobby programmers will ever touch this code."
That's kind of the type of user that this project targets.
The idea is to put on the independent creator's hands the power to accomplish motion capture in a flexible and inexpensive way.
The fact that the license will probably keep big companies away doesn't necessarily means that the project can't succeed. Just to name some examples on the non OS environment:
-Slic3r, the tool that virtually started the domestic 3D printing revolution. AGPL license, 8 years of heavy development, 91 contributors and more that 6000 commits on github.
-Blender, the powerful 3D creation tool. GPLv2, 25 years, now receiving good financial support from many top-tier companies and from the community.
I don't know if you took a look at the current Low-cost mocap panorama. It's an interesting environment where many changes are happening. We felt like there was a place for a community driven project.
It is tiresome when people use misleading statistics.
In the context of GitHub, MIT license is the most common license. The average number of contributor for a GitHub project is also 1, and the average number of downloads are also 1. A very large portion of all projects are empty except for a license file.
In the context of Debian and downstream distrubution like Ubuntu, GPL is the most common license. GPL family of licenses cover about 80% of all packages. The average debian package looks very different from that of a average github project, has significant more contributors and users.
So who is winning the intellectual marketplace? Debian and downstream of Debian is responsible for about 67% of all web servers. The wast majority of companies in the world has chosen to run servers which primarily has GPL licensed software inside. Operative system with primarily Permissive Open Source licensed packages are a small single digit percent of the web server market.
There is no need to resort to ad hominem attacks here, this is HN and we generally try to stay above that level.
Do you anticipate the network effect dynamics of a motion capture library to look more like an operating system (Debian model) or a specialty library (Github)? I would expect it to look a lot more like the Github data, so the fact that Linux is for historical reasons tightly coupled to GPL licensing, while interesting, is probably not a good model for predicting future contributions to a newly GPL'ed motion capture library like this one.
I anticipate that a project with 8 paid members to be more similar to a package in debian than a average project on Github. A package in debian goes through a curating process. It is not a one-click registration but involves making sure the package work, is relevant enough to gain sponsorship, and is mature enough to be stable for a 2 years release cycle.
We are not comparing the linux kernel with this library, and I am not sure what prompted the comparison. There would not be 80% GPL in Debian and downstream if the kernel was the only package that with GPL. 80% GPL mean 80% of all 30 000 software packages in the repository.
In this context I also find it relevant that Chorodata seems to have two teams, half that focus on building this tool and an other that work on real-time performances and installations that I assume is not done for free. That implies the model that this motion capture library seem to follow is similar in spirit to open core, where the revenue does not come from the tool itself but from work that is enabled by it. Such project, new and old, usually go with GPL.
I thus expect GPLv3 to be a quite suitable license for this kind of project and it is a license that many artists that use Creative Commons understand. Share the work as much as possible, invite collaboration if you want but retain some control.
Dude. Are you the one that flagged my initial comment? That's some pretty serious thought policing to think that anyone who dares question the sanctity of the holy GPL deserves to be flagged for their sins. We can have legitimate disagreements about best paths and data analysis but reaching for the flag hammer just because someone disagrees with you is quite simply fucked up and beneath someone who is actually able to articulate their opinions as you are. Perhaps GPL now advocates for free as in speech, unless that speech disagrees with RMS, in which case flag it and lock up the dissident who dared question whether using the GPL might be a mistake. Profoundly unimpressed.
There is no need to resort to ad hominem attacks here, this is HN and we generally try to stay above that level.
A comment get flagged killed on HN when a larger number of people report it, usually because it introduce flamewar topics without having something genuinely new to say.
Profoundly unimpressed with the the hypocrisy around free speech evidenced by the GPL advocates. I had hoped you would fire back to say it wasn't you and that you disagreed with the flagging. Your "clever retort" however is the reply of someone who is pleased that opposing viewpoints are being shut down and hidden from the world. Perhaps one day you will have the confidence to believe and understand that your ability to articulate for your points is powerful enough that you should not just allow but encourage opposing viewpoints to be heard, so that you could use your facilities to convince others of the right answer. You have that capacity. Blocking the opinion of others from being heard is not worthy of you.
I could write a bot that replied to all news stories that mention software under Permissive Open Source licenses and say that such licenses are a huge mistake and GPL licenses are clearly winning in the intellectual marketplace, linking to the study on Debian and Debian market share as proof.
Bots that advocate for either side would not make HN a better place and people flagging it would be the right choice. Not just because they are bots but because the repeated discussion does not advance any persons views or opinion. It is just advertisement noise from advocates arguing that their side is the holy crusade.
Cool project, cheap mocap is a quite an interesting field.
I would have expected a system based purely on such sensors would start to drift and accumulate errors over time. Are there any tricks used to compensate for that?
Thanks!
It is not a trick, it's what's called a Sensor fusion algorithm. They come in many flavors, we currently implement Madgwick's one. Another popular and somehow more robust alternatives are Kalman filters, we'll probably also implement one of those in the future.
They all rely on absolute references (gravity and magnetic field) to build a non-drifting reference frame which is used to correct the integration of (angular) accelerations.
19 comments
[ 2.7 ms ] story [ 34.2 ms ] thread[0] https://opensource.com/article/17/2/decline-gpl
The fact that the license will probably keep big companies away doesn't necessarily means that the project can't succeed. Just to name some examples on the non OS environment:
-Slic3r, the tool that virtually started the domestic 3D printing revolution. AGPL license, 8 years of heavy development, 91 contributors and more that 6000 commits on github.
-Blender, the powerful 3D creation tool. GPLv2, 25 years, now receiving good financial support from many top-tier companies and from the community.
I don't know if you took a look at the current Low-cost mocap panorama. It's an interesting environment where many changes are happening. We felt like there was a place for a community driven project.
In the context of GitHub, MIT license is the most common license. The average number of contributor for a GitHub project is also 1, and the average number of downloads are also 1. A very large portion of all projects are empty except for a license file.
In the context of Debian and downstream distrubution like Ubuntu, GPL is the most common license. GPL family of licenses cover about 80% of all packages. The average debian package looks very different from that of a average github project, has significant more contributors and users.
So who is winning the intellectual marketplace? Debian and downstream of Debian is responsible for about 67% of all web servers. The wast majority of companies in the world has chosen to run servers which primarily has GPL licensed software inside. Operative system with primarily Permissive Open Source licensed packages are a small single digit percent of the web server market.
There is no need to resort to ad hominem attacks here, this is HN and we generally try to stay above that level.
Do you anticipate the network effect dynamics of a motion capture library to look more like an operating system (Debian model) or a specialty library (Github)? I would expect it to look a lot more like the Github data, so the fact that Linux is for historical reasons tightly coupled to GPL licensing, while interesting, is probably not a good model for predicting future contributions to a newly GPL'ed motion capture library like this one.
We are not comparing the linux kernel with this library, and I am not sure what prompted the comparison. There would not be 80% GPL in Debian and downstream if the kernel was the only package that with GPL. 80% GPL mean 80% of all 30 000 software packages in the repository.
In this context I also find it relevant that Chorodata seems to have two teams, half that focus on building this tool and an other that work on real-time performances and installations that I assume is not done for free. That implies the model that this motion capture library seem to follow is similar in spirit to open core, where the revenue does not come from the tool itself but from work that is enabled by it. Such project, new and old, usually go with GPL.
I thus expect GPLv3 to be a quite suitable license for this kind of project and it is a license that many artists that use Creative Commons understand. Share the work as much as possible, invite collaboration if you want but retain some control.
A comment get flagged killed on HN when a larger number of people report it, usually because it introduce flamewar topics without having something genuinely new to say.
See more at: https://news.ycombinator.com/newsguidelines.html
Bots that advocate for either side would not make HN a better place and people flagging it would be the right choice. Not just because they are bots but because the repeated discussion does not advance any persons views or opinion. It is just advertisement noise from advocates arguing that their side is the holy crusade.
I would have expected a system based purely on such sensors would start to drift and accumulate errors over time. Are there any tricks used to compensate for that?
They all rely on absolute references (gravity and magnetic field) to build a non-drifting reference frame which is used to correct the integration of (angular) accelerations.
Something like the kinect can be very powerful for facial tracking. But monocular optical techniques tend to give poor body capture results.
The tracking data was not used for the final rendering, but was a basis for the animation team to work from.
You can definitely see how much it has improved when comparing that scene to the more recent iterations thanos.
The Xbox Kinect was such a versatile, and well-built device. Its a shame that its been discontinued and didn't find much use in games.