History of CAD | Hacker News Reader

History of CAD

shapr3d.com

160 pointsnill065 comments

I've been reading through a PDF on my Kindle Scribe which I downloaded from the original site --- it's a great overview, and most importantly, includes citations and references to the various papers where folks shared their research/progress.

These days of course, one would just get _The Essentials of CAGD_ and similar texts, or read: https://pomax.github.io/bezierinfo/ or watch: https://www.youtube.com/watch?v=jvPPXbo87ds

Highly recommended for anyone who wants an understanding of how modern CAD software came about.

For my part, I have found it invaluable in working on my current project as described at: https://news.ycombinator.com/item?id=43159236

A PDF of the submission? Would be grateful if you had a link; I can't seem to find one on the website.

Is it up on archive.org?

https://web.archive.org/web/20230325080337/https://www.cadhi...

if not, drop me an e-mail at my username here @aol.com and I'll dig it out.

found a PDF copy here: http://images.designworldonline.com.s3.amazonaws.com/CADhist...

> Lemon felt strongly that timesharing systems were the wave of the future and that SDRC should focus on providing software on these systems rather than license its software for a one-time fee. Initially, the company sold ANSYS and NASTRAN on a timesharing basis using computers operated by U. S. Steel.

SDRC did SaaS in the 1970s, before SaaS was cool.

> In September 1994 the company announced that it would be restating its revenues and earning for 1992 through the first half of 1994 to include a $30 million charge relating to sales discrepancies in its Asian operations [...] The company immediately terminated Tony Tolani, a vice president and general manager of SDRC’s Far Eastern Operations.

I was going to say they cooked the books before it was cool, too, but that's a much older trick. They claimed they sold all this software while in reality they just dumped it in some warehouse at the Cincinnati airport. Then proceeded to sell their shares before shit hit fan. SEC link with more details: https://www.sec.gov/files/litigation/litreleases/lr15325.txt

That is to say the downright corrupt and shady sales and management team almost killed the company. It recovered eventually and ended up being bought by EDS

> While the software’s geometric modeling capabilities were improved it was the package’s user interface that really impressed me. Icon menus typically were only two levels deep compared to five and six levels in other systems [...] I wrote in the April 1995 issue of EAReport: “I-DEAS has the best interactive user interface available today for mechanical design and analysis.

I agree. Its interface would look odd and clunky by today's metrics and might even been considered unusual at that time, but it was really nice to use after some practice. They seemingly put some effort into its design.

I remember working with SDRC I-DEAS API in the early 2000s - it was well documented and "clean looking" compared to something like PTC's Pro/TOOLKIT. Interestingly, it was also CORBA-based, and it worked quite well over LAN.

I worked on I-DEAS briefly, on the FEM part, but not the UI, so never got the chance to talk to the UI team. It was a wild time. Windows NT was taking the CAD/CAE world by storm, especially with much cheaper graphics cards available. We still supported SGI, AIX, HP, Sun, but the writing was on the wall already, and there was a break-neck pace to port everything to Windows.

> They claimed they sold all this software while in reality they just dumped it in some warehouse at the Cincinnati airport. Then proceeded to sell their shares before shit hit fan.

Now crypto bros can just nakedly pump and dump. We live in interesting times.

I recently learned that there is a free community edition of Solid Edge: https://resources.sw.siemens.com/en-US/download-solid-edge-c...

According to the History of CAD Solid Edge originally came from Intergraph. Many packages/kernels seem to have changed hands over the course of decades.

Here's a neat paper about Boeing's use of CAD and CAM in the 50s and 60s. I particularly like the story about their first computer-made part drawings. There were no printers, and plotters were too inaccurate; but they had CNC mills and they realized "if you can cut in three dimensions, you can certainly scratch in two." So Boeing's first computer-made part drawings were engraved on sheets of aluminum.

https://inria.hal.science/hal-01526813v1/document

Good counter on John Walker's "The Autodesk File" https://www.fourmilab.ch/autofile/

Seeing as David Weisberg's "History of CAD" is trending today I submitted "History of Unigraphics" by 3 of the original Unigraphics 7 dwarfs.

Edit: Unigraphics X SDRC became the system we know today as Siemens NX

It came out at the end of 2024 - these guys must all be around 80 by now.

https://news.ycombinator.com/item?id=43169194

I'll have to work through the 650 pages.

I use CATIA v5 every day, first released in 1998. It's sharp and crusty and huge, but does everything (with the right license).

Crusty it is. The fillet tool STILL totally sucks after 27 years, and I STILL frequently find myself needing to define surface geometry, fillet the surface, and use that to split the solid rather than just directly filleting the solid, in order to get the same type of fillet intersections that you get when you machine the part.

Funny, I'm French, and I know people who developed a bit of it.

If you really push CATIA hard, you can get it to crash in French, or at least crash with a poorly translated error message.

"The Document cannot be unloaded because he is dirty" was one of my favorites.

I like the "Click OK to terminate" error. As though it wants to make sure it has my permission to crash.

Yeah, the whole button for user to click "Okay" for something which is abjectly _not_ acceptable has _got_ to go away.

Please replace it with "Understood" or "Acknowledged".

Or the not too uncommon "dismiss".

I keep a PowerPoint with screenshots of bizarre CATIA/ENOVIA error messages. It's up to 8 slides now, completely covered with errors. A few other highlights:

"As it was too important, the number of workspaces has been reduced."

"Primitive Value is not an AggrInstance"

"Instead of directly transforming an import, we recommend you apply the Add Position contextual command onto the solid. Then the required GSD transformation must be applied to the axis system of the Positioning Set. Do you want to go on with the transformation or quit the command and follow our recommendation? (Yes/No buttons)"

"A problem occurs during Process."

"Error: Error Stack is empty."

an empty dialog box with options of "OK" or "Cancel."

"An error condition has been detected, but no error information."

While performing a Surface to Surface analysis: Status: 104% complete, 1193hr 2min 46sec remaining.

"Cannot bitblt"

"Technological Package found use appropriate engine to handle it. It can not be saved as a File."

Me too. But I really looking forward to it.

I don't seem to find a dead tree version of the book. I don't know how much work it entails, but maybe the profits of such a sale could go to the good cause they were advertising?

Maybe it's the right opportunity to ask if you know a good online service for printing/assembling of big documents.

>Becoming a book publisher wasn’t on my bingo card when I started Shapr3D. Yet here we go! For now, it’s only available to select customers—but if there’s enough interest, we’ll make it accessible to the public. The book spans 860 pages and weighs approximately 3 kg—just as heavy as CAD is.

Become a select customer of Shapr3D and their CEO István Csanády might send you one.

https://twitter.com/istvan_csanady/status/188829861216722566...

Hmmm, no mention of Teknicad? sadface

I seem to be the only person (along with 3 others who were my cow-orkers) who ever used or heard of Teknicad , running on Tektronix Unix workstations and terminals :)

One thing about this history is that it kind of light on the development of CAM, Computer Aided Manufacturing, which was co-developed along with CAD, and in some cases was actually a predecessor to CAD.

I recommend "A Possible First Use of CAM/CAD" by Norman Sanders as a short history of CAM in the 1950s/1960s.

https://inria.hal.science/hal-01526813

Excellent point, and interesting paper.

Arguably though, CAM is a totally different subject, though as you point out, inextricably intertwined with CAD in their origin.

Wasn't it a pretty straight-forward lineage of: G-code is developed, there are a couple of commercial apps which write to it, then when the EMC software is release and Mach3/LinuxCNC/Grbl become prevalent there was an explosion of different software options?

Just not sure what one could say about that which would be interesting.

tangential : I still can't get over the idea that shapr3d license is subscription only. You cannot buy the software. This means that your whole work/workflows ceases to exist if you not continuing to pay. No fallback option.

Unfortunately, the printing industry rolled over for Adobe and set a precedent.

An alternative to Shapr3D might be: https://www.plasticity.xyz/ which has an interesting license and licensing model (I bought a Studio license at launch, but haven't used it since).

Fortunately, FreeCAD is markedly improved, and for folks who want something light-weight and intuitive, I would recommend trying: Dune 3D as discussed here previously:

https://news.ycombinator.com/item?id=37979758

https://news.ycombinator.com/item?id=40228068

https://news.ycombinator.com/item?id=40228257

https://news.ycombinator.com/item?id=41975958

When I saw this thread I wanted to toss a comment in: If you used FreeCAD prior to 1.0 and abandoned it because editing past operations was an exercise in frustration-- give it a try. It now seems to be as robust in editing as solidworks or onshape is.. and is now quite usable.

I just started using CAD last week, but as I understand:

- Plasticity is so expensive because it uses one of the major kernels (SIEMENS Parasolid). Most of the licensing fee actually goes to them. You can find the Show HN post with a much cheaper price before he changed kernels.

- There was a firm called Ondsel that was trying to improve the UX of FreeCAD by selling a fork. It recently folded. One of their employees is trying the same as AstoCAD. The idea is that you fund the development of FreeCAD, and you get the improved binaries before the work lands upstream.

In the world of CAD, Plasticity is only expensive when compared with free tools. The “Indie” version is $149USD, and yes a lot of that cost apparently goes to paying Siemens for a Parasolid license. Shapr3D is also Parasolid based, along with NX, Solidworks, Onshape and a bunch of others.

That is my understanding as well.

Curious what CAD program (programs?) you are using and what sort of work you are doing and how you are approaching it.

I'm using Fusion 360. It seems to have the best community support.

I work at a big company that likes to claim ownership of anything I create while I'm on contract. Fusion is free until you earn more than $1k from your making, but I don't expect to earn anything from it while I'm employed.

This video is a little slow, but it does give you an overview of the UI:

https://youtu.be/8VFOVxdi4Qs

And this one teaches you enough to start making things:

https://youtu.be/xJihYVwkCqw

I used 3ds Max in high school and have been a lifelong user of the Adobe suite, so I just needed enough to know where the buttons are and what the core workflow looks like.

I've got a few projects going.

I just moved into a new apartment, and I'd like to sculpt some hangers - one to mount a speaker on the TV, and a pair to hang my old Rock Band guitars on the wall. I think I'm gonna do the engineering in Fusion and then sculpt the aesthetics in Tilt Brush.

Speaking of Rock Band, I've discovered a graveyard of neglected plastic instruments in the game room at work, as well as an open source Arduino/Raspberry Pi firmware that will allow them to work with any game, donglefree. I'm designing a new USB faceplate and brackets to make the new boards fit the existing standoffs, so I can refurbish them.

https://santroller.tangentmc.net

Finally, my new kitchen island has a frustratingly short drawer. I cut the back off it and bought longer rails. I'm designing spacers and jigs to help me keep it square and level as I reinstall the extended version.

Have you considered how Autodesk arrived at that level of community support and how the license for Fusion 360 has changed over the years? (and is subject to change in the future).

My comment from a Reddit discussion on Fusion 360 license changes:

>Translation:

>We have extracted all the beta-testing which we wanted for free out of naïve users

>and are now taking our game over to the other field where we can charge admission.

https://old.reddit.com/r/hobbycnc/comments/itw4d7/autodesk_i...

None of those events have had any impact on me as a hobby user of Fusion 360. Sure, it's subject to change. The software is a business for them. I know a lot of people got worked up over this. I do not see any alternative that is as cost-effective and functional.

Honestly, I'm more interested in getting these projects made than in "what if"ing the future. I have a hard time imagining my needs exceeding what I expect to remain in the hobbyist sphere, and I have the ability to transfer my skillset to another tool when the time comes.

I'd rather spend a month trying things out with the free thing with lots of reference material than go off on my own path spending hundreds on some indie software that I might only use a few times.

They have a neat program, but its really hard to defend their pricing model. I gave it a shot and its just way too expensive given the functionality. Yes, they are constantly improving, but the price for me is just too much considering alternatives.

Read through this discourse topic: https://discourse.shapr3d.com/t/why-there-is-no-plan-for-hob...

I don't understand why they do what they do when it comes to pricing.

There's an interesting cultural divide here. Probably a majority of mechanical and electrical designers still think it's risky and weird to use software that doesn't come from a vendor. Even though they experience occasional outages when something goes wrong with the license billing and payment. The electricals are a bit further along, I think because they're closer to the programming world.

I think it’s more that management in this space is very traditional and from a business perspective there’s not much downside to a subscription. The individual engineers don’t really have a choice in cad package anyway and frankly, the alternatives still suck in mechanical space.

Not to mention, file formats between packages are sort of interoperable, but often the design history cannot be transferred. It’s not like software land where if a vendor goes down (and the code is reasonably structured), an organization can mostly replace it with substitutes.

Just wanted to say that i found out about shapr3d here on HN through a comment a while back. And– though it does sound dramtic– it changed my life. It got me into CAD modelling without any experience. Which i wouldn't have tried on a computer, but on my ipad, it was so easy and intuitive to get started.

Since then i've designed so many things. From weird furniture ideas to replacement parts for my vacuum up to the mot important 3d models of a device-prototype. It's super fun, intuitive and i think in general the most expensive software subscription i pay for, also hands-down the software that impacted my life the most (as a consumer/user).

Have you tried any other CAD programs?

If you have access to a computer, could you try Dune 3D and write up a quick comparison? It seems the easiest-to-use of the opensource options.

Interesting, didn't know about dune3d. I can't try today though. What are you expecting from my comparison? If i don't forget, i can come back here in a few days if had time to try dune and tell you about my experience.

I am just curious how Dune 3D "feels" to a person accustomed to Shapr3D --- the former is notable among opensource applications for having a touch/trackpad-friendly UI, which is a defining characteristic of the latter.

Have you tried Shapr3d? The tablet UX is entirely different from any other CAD program. You are asking to compare apples to oranges.

Yes, I did try it --- got part of the way through the tutorials and decided that the UI wasn't appealing enough to overcome the price/license.

Spotted one typo, the reference to Hanchette in the section on Matra Datavision should be Hachette.

That section could also mention that the Matra software is now opensource.

Why are you telling us?

I skimmed, so maybe I missed it: Is there anything in there about OnShape?

Could be a particularly interesting follow-on to the section about Solidworks and Jon Hirschtick.

No, that is too recent a development.

The book was published in 2008 and the PDF I'm reading doesn't seem to have been updated.

Rather a shame it wound up at a commercial site rather than say https://www.wikibooks.org/

Is there an overview of which software is using what kernel?

In the open source world there are 2 (maybe 3) geometry kernels that handle NURBS and can produce STEP files. Open Cascade, which is GPL now but had a commercial origin. It's quite good and is used in FreeCAD, Salome, KiCAD (EDA), Horizon EDA, Dune 3d, and probably several more. And then Solvespace which is the only user of it's own bespoke kernel which we are still trying to make more robust. This one is also limited to a certain subset of NURBS constructs but is still quite useful.

Open Cascade may be the most difficult piece of FLOSS IP to recreate outside the Linux Kernel. In fact I'd say it'd be harder to replace given the number of people with interest and the background to work on these two different type of software.

Thank you so much for all the work you and your colleagues do to bring solvespace to the world.

I believe it has the potential to be one of — and probably already is in — a class of super impactful, all time classic pieces of software alongside the likes of Firefox, Gimp, Inkscape, et al.

Thank you so much for that comment. How you see solvespace is why I started contributing to it. I saw it as fundamentally different and very cool, but it had a lot of shortcomings that I felt capable of addressing. Unfortunately I haven't given it much time lately (personal happenings). I wanted to put out version 3.2 about 2 years ago, and it's been "almost there" ever since. We will get there though. Comments like yours really help when its volunteer work! Thanks agian.

Agreed, though I would put it in a more rarified category of "Opensource software which has innovated in terms of interface in a unique and meaningful fashion." --- other tools I put in this space are:

- LyX

- pyspread

- ipe

Will geometry kernels ever be fully robust, or is there still a holy grail to be discovered here?

>> Will geometry kernels ever be fully robust, or is there still a holy grail to be discovered here?

WRT robustness maybe not. Even the commercial ones tend to fail in certain situations. I'm not sure if there's a widely desired holy grail people are looking for and I don't even know what that would look like. I might recognize it if it appeared, but maybe not.

BTW for FEA I think a major new thing is available in Altair products the last couple years.

What about one of those SDF code CADs but with FreeCAD's sketcher as the UI?

I'd imagine extruding a sketch into an SDF object is possible, and you could just only allow sketches to reference stuff in other sketches, or the bounding box cube of a sketch extrusion.

They'd need some better tools for referencing stuff between sketches but other than that, it seems like it's mostly all there.

And then at the end if you wanted a STEP I'm sure someone out there has got to be working on some kind of mesh to STEP thing that properly preserves design intent.

I am pretty sure that this is a consequence of the mathematics for modeling this being _very_ hard, and a robust (and performant) solution requiring as-yet undiscovered/unpublished mathematics.

Solvespace only seems to have simple STEP export, Open Cascade can import STEP as well.

>> Solvespace only seems to have simple STEP export...

And it will probably stay that way. STEP is a huge specification and solvespace only supports a small subset of the features of it. In particular it doesn't do complex NURBS curves or surfaces, it only supports rational Bezier curves and surfaces up to degree 3. It does support trimmed surfaces which result from boolean operations and which most other OSS NURBS libraries do not handle.

Some day I'd like to get it to import what it can - particularly its own output - but that's a very low priority since it would be quite limited.

BTW gcad3d has fairly extensive STEP I/O with its own GPL3 C code (so we could start with that). I'm still not sure what that tool is mostly for - it seems to be more for CAM simulation? but I'm not sure.

I would use the STEP I/O library from Open Cascade (or BRL-CAD), they are closer to how we expect someone to read and write STEP models, neither are really tied to the parent CAD system. The one in BRL-CAD does have problems compiling a feature of the latest STEP standards.

One thing missing from the subject of the thread was any history of the exchange formats.

https://en.wikipedia.org/wiki/Geometric_modeling_kernel

"Kernel developers section"

Thanks!

How does Shapr compare to FreeCAD?

It uses a robust, commercial CAD kernel, and has an interface optimized for touch/stylus (originally it only ran on iPad Pro when an Apple Pencil was available).

Similar programs to consider include Moment of Inspiration and Plasticity for commercial, and Dune 3D for opensource.

Reply on news.ycombinator.com