XSI was originally from Softimage. Softimage sold out to Microsoft, which had no idea what to do with a high-end product other than to make it run on Windows NT. That ended SGI's reign in Hollywood; between 1998 and 2001, studios moved off SGI onto PCs with lots of memory and graphics cards that rapidly got cheaper.

Microsoft sold Softimage to Avid. Avid wanted Softimage because they also had a good non-linear editor, Softimage|DS, which was a threat to Avid's business. Avid used to sell furniture; they built computers with lots of accessories for handling video into wooden desks. These were called "editing suites", and cost upwards of $100K. Softimage could do that on commodity hardware.

Avid had no idea what to do with the 3D product line. It was a product Avid was stuck with, and it had its own team in Montreal. In 2008, Avid sold that to Autodesk.

Autodesk had developed its own 3D system, 3D Studio Max, which is still around. They bought Alias/Wavefront Maya from SGI when SGI tanked in 2004. That's how Autodesk ended up owning the 3D animation business.

> That ended SGI's reign in Hollywood;

Big studios didn't convert to NT, they converted to Linux. SGI ended their own reign by not being able to compete with Nvidia and AMD/Intel. Even now all those studios use csh since that was the default shell that came with Irix.

There was a short lived start of a shift to NT in the 90s before Linux became established/respected enough to squash that idea.

Even SGI itself tried not to get left behind: https://en.wikipedia.org/wiki/SGI_Visual_Workstation

SGI made a few NT computers as an attempt to sell PCs which of course didn't work out due to not offering much except for a name an inflated price. Linux was coming around at the time, but SGI wouldn't have been an early adopter of Linux by any means since Irix was one of the jewels of the company, and helping adoption of Linux would have helped people transition to commodity PCs.

NT was a part of the transition for much smaller groups or individuals since they could buy a dual processor pentium II/pro and put a decent amount of RAM in them along with an Oxygen card that would accelerate OpenGL.

Large studios had a huge amount of infrastructure built up in pipeline scripts and workflow, proprietary software that might have used POSIX APIs and GUI libraries that weren't on windows, and big network file systems that tied things together.

I was working in the industry at the time, and my feeling was that while a few big studios went with Linux the vast majority of studios (especially among the smaller studios) went with NT.

Every big studio and the majority of medium studios converted to Linux once Houdini and Maya were ported. At the time the industry was dominated by five big American studios and 3-5 big London studios, all of which went the Linux route, not the NT route. Maybe you worked on multiple small studios that jumped on windows, but that wasn't the norm nor was windows the driving force is providing a smooth path from Irix.

By buying Softimage, Microsoft wanted to develop a high end product, to demonstrate to the world that its OS was powerful enough to compete with high end, software specific machines, at a lower cost. In that regards, I believe they succeeded.

I remember Softimage XSI, that was a beautiful system and I loved working with it back when I was playing around with some 3D stuff. They used to have great deals on pricing so it was also much more affordable than a lot of the other professional systems, but still seemed to provide a lot of the same functionality.

I was really sorry to see them go.

Thanks for the history lesson; I had no idea Alias|Wavefront was a division of SGI.

In 1995, SGI bought Alias and Wavefront, each with their own animation systems. Impressively, SGI got their people to work together and produce Maya. But it was close to the end of $10,000 to $20,000 graphics workstations; around 2000, the gamer graphics cards became good enough to run the high end software.

SGI thrashed around for years trying to find a niche. They bought Cray, for supercomputers. Didn't work out. They bought Intergraph, which built expensive CAD workstations. Dead end; PCs could do that. They got into servers, but 1U commodity servers crushed them. They made a big commitment to the Itanium, Intel's attempted successor to x86. Failure on all fronts. SGI was in the expensive computing business, and expensive computing was over.

That's what worries Autodesk's Carl Bass. Autodesk was originally, in the early 1980s, the price leader in CAD. The competition was selling high-end workstations bundled with a CAD package. The original big achievement of AutoCAD was cramming large drawings into a 640K DOS PC. It was kind of clunky, but way ahead of manual drafting. An AutoCAD setup was originally about $1K of software and $3K of hardware.

Autodesk has been trying to come out with low end mass market products for years. But there are only so many design engineers and architects. Years ago they had Autodesk Kitchen Designer, for laying out cabinets to fit. Now they have Autodesk Homestyler[1], a phone/tablet app which takes a picture of your room, builds a 3D model, lets you add furniture, and provides photorealistic renderings. It's free, supported by sales from the furniture for which it has models. It's a great piece of technology that's not very successful. IKEA has a competing product, which of course only has IKEA items, and that's more successful.

[1] http://www.homestyler.com/

Interestingly, SGI seems to be doing alright with their supercomputer business, such as UV. The market certainly shrank on them, but what's leftover doesn't look like commoditized or cloud computing will be able to compete it away any time soon.

I agree they made poor acquisition choices, but not that "expensive computing [is] over" -- it's just not the consumer market people thought it would be.

To be fair, it's not the original SGI anymore. In 2009 Rackable Systems (a cheap commodity server company) bought what was left of SGI and renamed their company SGI.

So, it's not quite the same lineage.

The new SGI does seem to have some nice gear, but I don't know how much secret sauce there is. I've never used it, but it looks pretty commodity to me. (Not that there's anything wrong with that).

SGI bought Alias|Wavefront BECAUSE of Microsoft.

When Microsoft bought Softimage, SGI freaked out, they feared MS would pull something awesome and make everything switch to x86 for editing.

Then they bought Wavefront and Alias separately, and forced the two to merge, Wavefront had a cool suite of editing tools that was award winning in the movie industry, Alias had a competing suite, the suites had lots of stuff unique to them, they merged the suits and created Maya (example: some of binaries that ship with Maya came from Alias, another ones from Wavefront, the GUI if I remember correctly was designed by Alias, the scripting language and the file formats by Wavefront, and so on...)

Autodesk then just reaped the mismanagement of MS and SGI

Just wanted to point out that 3ds max used to be named "3D Studio MAX" and grew out of an even older DOS software called 3D Studio. Autodesk bought it somewhere around when it got ported to Windows NT (maybe 1997?).

Autodesk really has grown mostly by acquisition.

3D Studio was developed by the Yost Group and published by Autodesk pretty much from the start, they bought it outright after 3D Studio MAX R2.

>They are a huge player with many expensive software suits. But they are not very innovative.

Compared to what? "Disruptive" social sites and Uber for X?

There is orders of magnitude more computer science innovation in Autodesk than in the average "unicorn".

"There is orders of magnitude more computer science innovation in Autodesk than in the average "unicorn"

I was under the impression that they've bought most of their marketable innovations. I realize this was not a fair comment to current employees there - yes, the software suites are impressive and Autocad reseachers have produces very nice things. I would still categorize them as a huge corporation with vested interests rather than an innovator in themselves.

Now, offering stable products is a good thing for many people. Owning DWG gives them an underhanded advantage though, and slows down the evolution of computer assisted design as a discipline, IMO. The DWG situation weights more in the anti-innovation camp more than a hundred clever Maya plugins.

DWG is considered a "standard" while effectively it's just a black-box binary blob. The prevalence of the DWG format means that either new players in CAD need to licence the RealDWG suite from Autocad, they can use third party reverse-engineered libraries like Teigha or they can hack their own. Teigha has been at it with several hired developers for a decade and they still lack bug-free support for the format. The fun part? Autocad can very well deny RealDWG to anyone they don't like. How's that for a market space for disruption.

It's nice to see there are new things happening in CAD world like OnShape (https://www.onshape.com/) that is not yet bought by them :)

As someone who works with AutoCAD nearly every day, I would love it if they opened up the DWG format. There are quite a lot of tools that AutoCAD produces to customize AutoCAD though.

Interestingly AutoCAD was originally written in lisp and a lot of the drawing objects are represented (or can be accessed) as lisp lists in AutoCAD's own lisp dialect (AutoLISP). It's pretty confusing for anyone not familiar with lisp though and fairly limited compared to other lisps (no macros, as far as I could tell, though I'm sure someone could probably find a way to implement them)

I doubt that Autocad was ever written in Lisp.

Autolisp was added to Autocad when it already existed. They took the free XLisp (written by David Betz in the early 80s) and integrated it into Autocad. The XLisp of that time was a small Lisp dialect written in C.

Some years ago the old Autolisp implementation was replaced with a different (but compatible) Lisp implementation, which Autodesk bought. That was then called Visual Lisp.

It's confusing, all right, and I don't think it makes for maintainable code.

Example: http://www.lee-mac.com/5pointellipse.html

> I was under the impression that they've bought most of their marketable innovations.

They did, but once you buy it, it's yours? The large companies buying up smaller ones are essentially funding the research and innovation.

Not when their goal is to lock-out all the competition in the space, which is something Autodesk is very interested in.

That seems like the opposite of innovation to me.

That's a pretty hasty generalization. As someone who works for a product at ADSK that has frequent collaboration meetings with our biggest competitors, I can point to at least one significant example of innovation via connectedness. Our philosophy is that our customer is going to pick the tools that work best for their specific workflow; since as a larger company we can't support every niche market, the best thing we can do is help people connect the dots between our software and their other vendors.

Niche markets are fine and good since they are not encroaching on the biggest mainstream user markets.

"the best thing we can do is help people connect the dots between our software and their other vendors."

Yes. And the worst thing Autodesk has done is try to inflict costs to this "connecting the dots" - namely, denying realDWG support for those they consider a strategic threat. Since there is Teigha this is not financially an unsurpassable problem but it's a pain in the ass for anyone who can't use realDWG.

Autodesk is in a position to wield leverage through DWG, they have the financial incentives to do so (and have done so) and for a company this is quite understandable.

Construction industry software is so ready for disruption.

A hint for any ambitious software engineers with a penchant for linear algebra - grab a few beers with some buddies who are trained construction engineers, ask to observe their work for a day with any software they are using, and observe how simple the principles underlying most CAD software is and how obnoxiously expensive and low quality most of such software is. Grab the construction engineer buddy, a computer graphics engineer and an applied mathematician. Start from a. performance b. quality c. shareability. Rule the market.

Yeah, on the gamedev side there's never been proper Collada support from Maya, it's very much low on the support totem despite being a pretty solid standard for tooling interop.

BIM software like Revit is pretty clever and would be difficult to build from scratch. AutoCAD is horrible but a lot of the horribleness is difficult to get rid of while providing the same functionality. So, I'm not sure I agree with your take.

They bought Revit and have somehow managed to regress it. Before owning Revit, and therefore only owning a small percentage of the nascent BIM tool marketplace, they drove the creation of IFC classes to ensure interoperability between BIM platforms. Now they practically own BIM, support for this has gone.

Auto desk are a vile company. A serial monopolist that make any of the othe offenders look tame. I'd love to see them go out of business. That said, I hope those that have lost their jobs find new ones soon.

I was commenting on "most CAD software", not just Autodesk products. Autodesk is not a sexy company (monopoly or otherwise) but I really wonder whether DWG/AutoCAD's key market, architectural drafting (as opposed to architectural drawing) is ever going to be a rewarding activity for anyone.

I've got years of diverse experience in the field of architecture and software, and Revit seems like a very clever toy but not a substitute for endless labour over details. Good architects draw every brick (or equivalent), and/or they also work in close experimental collaboration with the people fabricating the building elements. I'm thinking in the first case of Caruso St. John http://www.carusostjohn.com/ drawing every brick and in the second of a practice like Grimshaw http://grimshaw-architects.com/ using lost-wax casting (investment casting) to make components. The trade or craft knowledge involved means there are effective constraints which cannot even be expressed in Revit (individual bricks are not really an option in Revit, and you definitely wouldn't want to design a metal casting using Revit... It's an Autocad job, or better, a job for something more expressive like Maya or Mudbox (or whatever it's called) or whatever tool suits your aesthetic. You can't get to Rodin from Autocad.)

> They did, but once you buy it, it's yours?

Autodesk certainly doesn't think so, at least with respect to people who purchase copies of their software.

They also do a lot of interesting work via Autodesk Research (including on some things you might not expect, such as synthetic biology): https://autodeskresearch.com/

Publication list: https://autodeskresearch.com/publications

Compared to SolidWorks, SketchUp, and even parts of MS Visio. The UI and meta-data handling is very cumbersome and every time we use it we think how much better some parts could be. As with many mega-corps they tend to follow or buy the innovations of others, eventually.

Hmm, I tried to use SketchUp a month or so ago, after being a proficient AutoCAD user but not touching it since around 2005. I'm curious what parts of the UI are more usable in SketchUp than AutoCAD. For 2D drafting, I got up to speed fairly quick in SketchUp but it became apparent to me that in terms of drafting productivity, SketchUp in 2015 couldn't really compete with AutoCAD circa 2005.

Sketchup is designed for architectural sketches. The modeling grammar is nice for some operations but Sketchup is not really a well rounded modeling package. They are trying to move into construction.

IMO, main advantage of Sketchup over Autocad are a free API to access the model (although the domain model is not entirely well designed) and a fairly limited context of user interface operations (which makes the learning curve lowish).

The most critical difference though, is that Autocad supports actual analytic shapes while Sketchup handles only boundary representations (i.e. polygon meshes). No real cylinder for you! This may or may not be a problem.

IMO, Solidworks was more innovative than AutoCAD in the 90s and early 2000s. AutoCAD followed their lead by making Inventor. Today OnShape is innovating and AutoCAD is following with Fusion 360.

Actually, I believe Fusion 360 was around first. Having said that, trying to learn Fusion 360 was one of the worst experiences I've ever had learning software, while learning OnShape actually made me enjoy using CAD. Once OnShape gets to the level of maturity that Fusion 360 has, I am hoping they will be better at all levels.

I found learning Fusion 360 was painful. I'm dyslexic and need the big picture to understand how to put the individual pieces together. Fusion 360 uses terminology that is not defined anywhere. I worked with some of the support staff and they agreed that they lack this sort of documentation. It seems that some of the people I know that use Fusion 360 just hack around with it till they get it right. I don't have the time for that. Too many projects going at one time. I will ahve to look at OhShape. Thanks.

> Compared to what? "Disruptive" social sites and Uber for X?

I can't comment on CAD, but in the world of 3D graphics I find SideFX's Houdini and The Foundry's Modo to be way more useable, extensible, and feature rich than Autodesk's Maya.

Arguably most of their more innovative stuff like XSI, Mudbox, etc have all been acquisitions and not original IP.

I can't speak to the entirety of their empire, but as far as their rendering command stream: not innovative. Well...maybe 30 years ago. To be fair, that's most of the workstation industry.

Compared to actually innovative CAD companies.

Autodesks MO for the last couple of decades is to just buy out any competitor in the market.

>They are a huge player with many expensive software suits. But they are not very innovative.

In architecture and construction they are by far the leader in Building Information Modeling, so i think it depends on what you put in the spotlight.

They are a leader because everybody uses them and if you need to collaborate (which you do, as architects, engineers, contractors are all separate companies) you need to use what everyone else is using.

Revit has come a long way and it's starting to fulfill it's promise, but it's been a slow road.

Atm there is no real competitor to revit as openbim is not really market ready and all other software lacks compability.

I worked developing a Revit plugin for one of the biggest furniture manufacturers in the US for the past six months and despite the fact that Revit has a really good UX for people who is used to it, the development side is not as good feeling like they have been bringing the same stuff forward for the last ten years which turns their API into a bloated unfriendly piece of software.

Now we are starting to develop a replacement using Configura which is a swedish company working on the same space for the last 20 years or so and although they have a proprietary language and no documentation at all, still makes for a much better developer experience and game changing UX where users can design, render photos and produce movies of their project from within the tool. It's a whole different world.

However, both tools are, in my opinion, stuck in the past since I expect to see some competitor coming up in the next few years bringing the same experience in a web environment.

I rather liked working with their stuff but converting from their ever-so-slightly-different BIM format to standard BIM format was not 100% (I was using a library), wasn't the worst though.

Project cyborg isn't innovative: https://autodeskresearch.com/projects/cyborg

Tools for modelling matter at the nano/molecular scale. Seems pretty far out to me, I get it's just research right now but not many other groups working on this.

What passes for innovative in your universe? Fancy JSON file format? AutoCAD web app?

I believe there is still much to improve in the way computers are used in the design of structures. The default modeling experience in Autocad has not really changed ...since when? Yet, it's one of the largest softwares out there. I think the modeling paradigm is flawed because it originates from a need to fit lots of data into a tiny ram, and the whole workflow is built around this. Yes, there is the APi - but the differnce between a good UI and an API is the difference between Photoshop and ImageMagick.

Project pipelines: Design information is passed between specialists and project offices. Here DWG - being as popular as it is - is used like a checkin of a git branch - it is as well suited for this as a docx document. People manage, though, with a bit of manual project specific work thrown in. No amount of fancy viewers is going to fix this.

People have started to notice that good graphics are available from 'off the shelf' and that handling the basic geometric primitives is not really rocket science. I'm confident OnShape will be followed by a sleuth of new modeling tools soon - and probabably not from Autodesk. They are too busy protecting their entrenched dominion. I might be wrong though.

Fusion 360 is becoming a small revolution among my makerspace.

Which features of Fusion would you consider the most important for your usage? Why is it gaining traction?

It allows people who are familiar with 3D printing to learn how to use the CNC machines without a big investment in single-purpose machining software.

Man, I so miss XSI. It was such a great piece of 3d software.

few of? Pretty much everything in my mind.

Still such a sadness that XSI got bought, they had killer engine integration(treating a game engine as first-class citizen inside XSI).