Disney Open Source

I'm also curious to know the use case and am reminded of the "Do you need a blockchain?" flowcharts [1][2]

[1] https://pbs.twimg.com/media/Cn2zMbTWYAAQA6i.png

[2] https://media.licdn.com/mpr/mpr/shrinknp_400_400/AAEAAQAAAAA...

Just tossing out wild guesses/ideas: Theme parks maybe? Or tracking/trading movie ownership? Contracts? Points systems for games?

Disney Dollars possibly.

Disney probably doesn't need a blockchain, but they do a fair bit of research. Considering Disney already has a relationship with its customers, a centralized system works best. In fact, this is what they're doing now with their wristband payment/identity system. Considering they just invested $1bn into this, I suspect they're going to be using it for quite some time.

https://www.wired.com/2015/03/disney-magicband/

Thank you - fixed. Just deployed the Org side github pages and missed that.

Check out the video. In the example, rendering of an animated model goes from 0.25fps without it, to 4fps with it using CPU rendering, or up to 40fps using CUDA for the example model....

I'd say that is large enough to have the potential to fundamentally alter how people work.

I'd say a 1-2 order of magnitude speedup is very significant.

Kind of - it depends on the artist you talk to. OpenSubdiv utilities have been part of the major 3D packages for about a year or two now, and it's true that they can offer some significant speedup, well-needed in both the games and per-rendered CGI industries where high-poly subd models can easily go up to several million triangles.

Besides the speedup, there is a very interesting feature of OpenSubdiv - local edge creasing. To influence the general creasing or smoothing of a subd mesh, an artist usually has to add more or subtract geometry in an area, respectively - whereas OpenSubdiv gives you the alternative option to assign a crease value to an area of the source mesh which influences the apparent creasing of the resultant mesh in that area without the artist having to add more geo. It works some of the time: not every artist likes using local edge creases but for those that do, it can be a very big help, considering the the occasional difficulty in adding new geometry to a mesh that flows correctly and doesn't accidentally ruin the rest of the mesh.

Huge chunks of modern computer graphics production rendering infrastructure is based on open codebases.

I don't get the snark. The visualization studios (Disney, Sony, etc) work on open source is a fantastic example how open sourcing codebases can be a win-win-win situation.

Cool, I'm glad, but still always be careful.

Meanwhile, I don't really have access to a lawyer and I'm curious:

1. Should I trust a license which is not distributed with the source code (i.e. on the hopefully official website)?

2. Should I trust a license which is in a sub-directory? My initial assumption is that, only that sub-directory is licensed.

I don't think it changes yazaddaruvala's point. What happens when ptex website goes away and disney lawyers decide there was no proof?

If by "well-defined rules" you mean normal "Copyright Law", then yes. But it extends no protections or permissions beyond what copyright law already gives you.

As far as how tied to Disney they are: the project lead and initial sole developer (others chip in quite a bit now) is an employee of Disney. My impression is that he's open to changing things but not if those changes interfere or complicate their deployment at Disney. So I'd say definitely Disney-prioritized rather than a straight up community project. Affiliated projects like AutoPkg are community driven and somewhat tool agnostic.

I think what you should use depends on your available time, inclination and mindset. I meant for this to be short but it ended up being ridiculously long. Sorry about that. Hopefully it helps save you and anyone else the time required to get a broad understanding of this space. tl;dr at the bottom.

I wouldn't bother with Server App. It's complete crap. Maybe useful for >10 macs but I'm honestly not sure if that's even true given all the ways it can totally shit the bed on you. Not well maintained. Probably best to think of it more as a reference implementation for 3rd parties developing an MDM solution.

If you have time for the learning curve, want to apply the configuration as code/automated testing mindset and want/need to save money go with the open source tools. Don't expect to be able to commit the commercial versions to source control. Munki+Reposado+Margarita+AutoPkg for sure:

Munki: Nothing touches it terms of the core/engine's package installation abilities. Repo is configured by the admin/tools like autopkg so it's just a static repo that can be hosted by any web server. Customizable self-service app that's very similar to the app store app.

Reposado: Let's you do pipeline/canary testing/one-some-many deployments of your official Apple updates since they can rarely mess things up on their own but more frequently break some software.

Margarita: Web UI for managing Reposado, only needed if you want/need that or just don't want to deal with figuring out how to manage/automate it.

AutoPkg: This is so amazing you'll want this even if you use something like Casper but it was designed with Munki in mind. Using recipes it downloads updates for software from the vendor's websites/repositories, repackages as necessary and can be configured to automatically add those as updates for your existing packages. So you can get security updates and force their installation long before your average laptop user bothers to update. You can do pipeline/canary testing/one-some-many deployments using your preferred CICD tool (Jenkins seems to be popular for this). Recipes are maintained in a community repo so even if a vendor changes their site and breaks things someone has probably noticed before you and provided a fix.

Key differentiators for these: Open Source, configuration management as code, automation, testing, CICD, etc.

Puppet, Chef, Ansible, Salt or GitHub's Boxen (Uses Puppet but has an opinionated layer on top): If you want more configuration management and distributing it via packages through Munki isn't your cup of tea use one of these in addition to Munki. Key differentiators: Open Source, configuration management as code, automation, testing, CICD, etc.

If you want easy as pie and very SMB oriented take a look at JAMF's Bushel. It's clean looking and subscription based and probably the least amount of effort overall. It's cloud-only I believe, so if on-prem is a requirement look at the next one. Not sure if you can do custom packages or much in the way of config but you can definitely distribute Apple App store applications. Key differentiators: modern, easy to use and quick to get started with. Auto-magic enrollment of devices you buy (as long as they are through the same account). Distributed App Store apps.

If you want more control and a decent GUI (particularly if you have a decent number of Macs) take a look at JAMF's Casper. Warning: The trial is limited to 14 days but you can purchase a 45 or 60 day trial+2 or 3 days of training for a mere ~$4K or $6K. Training I think is provided by a VAR/Partner. No you can't use it after that unless you buy it. Expect a hard and high touch sales experience since this is more or less enterprise software. There's a cloud version and on-prem version. On-prem will require some upkeep (MySQL instance to manage, java app that runs on tomcat, etc.). You should use AutoPkg with Casper too even if it's a bit annoying to get setup. Key differentiators: Big community with a lot of answers and advice that are often universal enough to be useful no matter what you're using. Very enterprise-y so good if you need that (contract, support, feel good sales pitch to management). Easy to use for junior admins or support people. Easy self-service that customizable per-person/group. Auto-magic enrollment of devices you buy (as long as they are through the same account).

Note: The JAMF sales people might bring in a rep from Apple's Enterprise Support team during the sales process to talk about the benefits of Casper (they use it at Apple) as well as (surprise) pitch their enterprise support services and a package called enterprise connect. Enterprise connect is essentially a workaround for bad UX/bugs in Apple's directory services support that they want you to pay for. When the Apple rep told us that you'd be silly to use open source tools we asked about Google and Facebook and he replied, "Yeah... well they probably think they're pretty smart! Haha." Backpedaling ensued and he said, "Well of course they're smart, they just have different problems." When pressed about the problems Casper couldn't resolve for them he said "scale" after conferring with another person at Apple.

There are others but either I don't know enough about them to offer any insight or they aren't worth mentioning for various reasons.

tl;dr:

Munki/Reposado/AutoPkg: If you can afford the learning curve and need/want the DevOps tools/automation/philosophy.

Bushel: If you want easy, no hassle and automatic device enrollment, SMB.

Casper: If you want a GUI, enterprise features and automatic device enrollment, easy self-service, have a decent number of Macs.

Thanks, I'll write to you soon. edit: :)

I dunno, I prefer to eat duck over dog or mouse. ;)

Trademarks are only within a field. If Disney wanted to open-source something and trademark its name, they'd have to get a new trademark

> Why should random companies get to legally print knockoff Mickey t-shirts just because enough time has passed?

Because our Copyright laws are all sorts of outdated and ridiculous in various ways.

Because that's what the US Constitution requires [1]:

"To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries" (Emphasis added)

[1] https://www.constituteproject.org/constitution/United_States...

>> Why should random companies get to legally print knockoff Mickey t-shirts just because enough time has passed?

Because after a while is supposed that the investment paid off. Why should a single company keep Mickey or any other IP right hostage forever?

The purpose of copyright law is only to "promote the arts and sciences". Note that it doesn't say anything about the individual creator.

There are two factors determining how much value a society can derive from the body of art created within: the amount and the distribution. (ignoring "quality" because it's unaffected by the following).

The quantity created and the distribution it gets are somewhat conflicting goals: If you maximize for distribution (all works are public domain), the amount that is created may be low because artists aren't rewarded financially / can't concentrate fully on their art etc.

Currently, the compromise is to protect a work of art for 70 years after the creator's death, assuming there are very few people who'd say "oh, I'd love to write the great american novel, but I fear my great-grandchildren won't be able to collect royalties on it, so I'll go be an investment banker instead".

Arguably, 70 years is too long to be optimal. Personally I'd say something from 10 to 20 years is probably enough, or maybe different terms for different categories like newspapers being protected for a week and movies for 30 years.

Because it does not benefit society in any way to grant them a monopoly on Mickey t-shirts. Society as a whole is better off when there is market competition. Monopoly is only granted to encourage the initial creativity. There is also significant public cost in protecting the Mickey monopoly, such as courts and police.

Your talking about a company that owns at least 50% of it's success to the i.p. of characters it didn't create. Cinderella, Sleeping Beauty, Pinocchio, Aladdin, Beauty and the Beast...i could go on. I don't think anyone would argue that the films they made, based on the work of others whom they did not pay, contained substantial original creative effort and that they deserve the success they gained from those films due to the significant changes they made. The great hypocrisy though is that Disney would not have nearly the success they have today if Disney had achieved then what they are striving to achieve today.

"Why should random companies get to legally print knockoff Mickey t-shirts just because enough time has passed?"

Because that was the deal on copyright. Protection in exchange for release into the public domain. Extending copyright is a really shitty way for them not to stick to their end of the deal.

Alembic was actually mostly Sony originally - ILM were mostly just a name on the ticket, and ILM started using the format because it was much more Katana-friendly (random access and lazy-loading) than their current format, and they were starting to use Katana at the time.

The first version of Alembic was pretty much just Sony's internal geometry format renamed.

Site is still terrible :/

It's calculated of some predefined averages that are way lower than actual market rates. Most H1B's I know are making pretty much market rate or within 5-10% of market rate though. How do they find people to work for about half I have no clue.

I believe it is based on the average for the country, which is much lower than the average in certain areas, like California.... But I may be wrong.

FTA "H-1B immigrants work for less than American tech workers, Professor Hira said at a hearing in March of the Senate Judiciary Committee, because of weaknesses in wage regulations. The savings have been 25 percent to 49 percent in recent cases, he told lawmakers."

More precisely even H1Bs are also required not to replace domestic workers, but somehow I must have missed seeing Disney execs being charged?

What happened was HCL and Cognizant were going to provide IT consultant work for Disney, but the consultants were based somewhere else other than Disney that had a lower prevailing wage.

Sponsoring employers are supposed to report the working areas on the LCA applications, and the USCIS is cracking down hard on the tech consultant firms trying to exploit that loophole this year.

It's for their job title; Disney is happy to pay 150% of an Associate Junior Technician's $30k average salary compared to 100% of a Senior Systems Administrator's $150k.

That's hyperbole, but it's the same idea.

They hired contractors. Tata et al. are experts at this stuff.

Alembic was originally Sony's thing, which ILM started using in their early Katana days because their own file format wasn't good enough for use with Katana (random access and lazy loading).

Sony gave it to ILM to start using, and then open-sourced it.

I'd quibble that while ILM and Pixar employees are technically now Disney employees, it's not as if they work for Walt Disney Animation Studios directly, so it's not as clear cut as you make out.