How Blender started, twenty years ago | Hacker News Reader

How Blender started, twenty years ago

code.blender.org

184 pointsunwind40 comments

Blender is a really fantastic software, from a professional, a educational, and a tinkering perspective.

I started using Blender in 1998 or 1999 back when it was still closed source (around version 1.6, I think). Back then I also used to use 3D Studio Max and it was a far superior software. Then, NaN went bankrupt and managed to raise enough awareness to collect the required money so that the investors could be paid out and Blender could be released into the public as open source. In the following years, Blender has continously grown in feature-size oftentimes coming quite close to what professional packages offer.

From a professional perspective, one can get a very solid 3D Rendering and Modelling solution for free that offers enough features to get a deep understanding of the whole subject. This makes it easy to learn enough to be competitive when it comes down to searching for a job.

From a educational perspective, Blender is a great teaching tool. The animations that Blender can generate (with particles and hair and bones) look good enough to amaze young kids. Based on this amazement, one can explain the math behind it, or the code behind it, to make computer science and math look more practical.

From a tinkering perspective, Blender is fantastic. The core is written in C/C++ but a huge part of the application is scripted in Python. This makes it very easy to have a look into the source, extend the app, and gain a better understanding of the ideas behind 3D Rendering and modelling. Case in Point: When working on an iPad game recently, a friend and I decided that we did not want to build our own level editor. Since one requirement of each level was the ability to edit nurbs curves, we decided to simply script a lightweight Blender Plugin so that we could use Blender 3D as the level editor. After some initial problems it went fluidly, and we quickly had a great working level editor with an extensible interface, all scripted in Python. If I were to work on another game and if there weren't already a good level editor for the kind of levels I'd need, I'd go with Blender again as it is really easy to extend and already has functionality for lots and lots of use cases.

All in all, Blender is - in my eyes - one of the big success stories of open source software. Especially because there's no big corporation behind it. Open Office has/had Sun, Linux has IBM, RedHat, and many many others, but Blender is really being build by a community of seperate entities and few corporations.

I'm also from that time, it was lovely to see Blender grow. I still remember being online one afternoon (yes, in those times being online was something rare), just to download Blender 2.0 when they implemented huge interface changes and the first game engine.

Congrats Ton and all!

Blender represents both the triumph and tragedy of open source software. I've been vaguely following its development since the late 90s, periodically making attempts to learn it, and failing until I finally made a more serious push earlier this month. Once one has gotten over the very steep learning curve, it's an amazing piece of kit. On a feature-by-feature basis, it handily outcompetes many commercial packages, and once one has learned the keyboard shortcuts, I can see how the interface would actually become a pleasure for daily users.

However for beginners and casual users, the UI is a disaster: stupidly opaque and pointlessly obtuse. It is obtuse in ways which aren't necessary to support the power-user workflow, but seem to exist for no other reason than to maximise the "WTF?!?" factor for new users.

For example: object-selection is done via a right-click, requiring the user to break their muscle-memory from literally every other GUI program they've ever used. WTF? And where are the contextual menus? And why are the all-important keyboard-shortcut tooltips scattered so haphazardly through the UI?

Making things worse, the defaults and documentation assume a three-button mouse and a full keyboard with a numeric keypad etc., and the most basic navigation functions -- panning, zooming, dollying, orbiting -- all rely on inputs from these devices. Working on my 13" Macbook Pro, it took me about 15 hours of banging my head against the software until I found a combination of settings which allowed me to pan and zoom etc. using only my trackpad. (I had asked the Blender forums for help, and the advice I received was to buy another computer). Now it's working quite well for me, but only because I was willing to endure a few days of pain to get there.

3D software doesn't need to be this difficult. In my day, I've been highly proficient at 3D Studio, Lightwave, Cinema4D, Form•Z, and various other platforms, and none of them embody UX anti-conventions the way that Blender does. Heck, even basic things like its click-drag mechanics are just wrong (click-drag to initiate a drag, at which point the object or control point becomes "sticky" to the cursor, even after it's released; a second click is required to end the drag), requiring the user to comprehensively retrain their muscle memory in order to use the software. WTF?!?

The Blender community, unfortunately, seems to largely be a self-selected cabal of people who have survived this pointless hazing ritual and are determined to make sure that everyone who follows in their footsteps does the same. Although a small minority of the community "gets it" and is making earnest and well-thought-through proposals about how to fix the situation (see: https://docs.google.com/document/d/1ScPMbHv8WRCU2znB7IU2l-W9...), they don't seem to have much traction, and are continually derided / ignored by the rest of the community. Thus even the very simplest lowest-hanging fruit -- such as enabling "select with left mouse button" by default (it's already an existing setting, buried deep in the preferences where no noob will ever find it) -- aren't implemented.

It's a sad situation, because with a bit more polish on the UI/UX, Blender could be a genuine player in the 3D industry, rather than an intriguingly powerful assemblage of missed opportunities.

I realize you have strong feelings here, but what you are saying isn't really a tragedy. Blender is hard to learn, so are 3D graphics in general. You can accomplish a lot but it's difficult to do so. That's it. not a tragedy. the community is pretty helpful too. Every artist community is pretty critical when presenting your work, blender is no exception but Iv'e gotten help at various times over the year in forums and IRC with sure the occasional snark (that's everywhere).

I bet more people ask for feature requests of big complex commercial software and don't gain traction either. People get used to the UI and come out of the woodwork when noobs want to change it.

Another comparison, without having owned it but just through seeing and hearing demos, Maya is barely usable for projects without 3rd party plugins (my understanding anyway). How is that a good setup for a new user?

I know that 3D graphics aren't trivial to learn, but the tragedy with Blender is that it is arbitrarily difficult. The difficulties aren't intrinsic to learning 3D. A lot of people -- both beginners and professionals from other platforms -- are turned off Blender when they can't even figure out how to do basic functions like select an object or pan across a view without watching tutorial videos. That's really unreasonable, and I'm sure that Blender loses a substantial portion of its potential users due to completely unnecessary UI anti-conventions like this. It wouldn't be difficult to fix: simply enabling "Select With: Left", "Orbit Style: Table", "Zoom to Mouse Position", and a handful of other existing defaults would make Blender vastly more approachable out-of-the-box. I'm sure that something as simple as this would produce a measurable increase in its rate of adoption, perhaps on the order of 15% or 20%, without compromising the advanced functionality in the slightest. Other measures such as consistent tooltips across the UI and contextual right-click menus would give another significant boost in adoption.

The tragedy of Blender is that the community is highly responsive to feature requests from established users, and has developed some superlative advanced capabilities as a result -- but is blind to the needs of beginning users. If Blender were a commercial product (maintained by people who think not only in terms of features, but in terms of market penetration) then these kind of simple issues would have been addressed a decade ago, and the good efforts of the Blender developers would now be reaching a much much wider audience.

My biggest regret was that they haven't invested more resources into BGE (Blender Game Engine). They had it within their graps, and could have easily stolen Unity3D's thunder, as well as probably gotten funding somewhere.

They just didn't listen to market, imho.

Admittedly, some licenses should've been changed for that to be true.

True. The BGE already worked great, including Python scripting and Windows export back in 2000. It was really impressive to edit in Blender, press a button, and run the game.

BGE wasn't ignored because anyone on the dev team wanted to sideline it. In fact, many many people really loved it and hoped it would go somewhere. But it is an open source project. The stuff that gets resources is the communities that show up and work on it.

If you want BGE to get better, join the team and make it better.

I think that you'd see a stronger BGE if Unity hadn't become free for personal projects. Unity and blender work pretty well together and Unity just became really popular so suddenly aspiring game devs just weren't begging for blender. I'm convinced of this, but curious if others have comments about it.

I love blender, but as a person with a startup in the 3d printing field (shameless plug: bld3r.com) my one regret is that the learning curve for modeling is so steep. Is there a forked version that is a standalone for 3d modeling? If there isn't, there should be.

Blender is the only program i know of that is open enough and powerful enough to spend a long time learning, but for noobs in 3d printing, it's absurdly disheartening. I'd hope we could develop a version easy as solidworks, even if it were uglier. If i'm totally out of the loop on open cad software, do please tell me what i'm missing.

Why would you need a standalone version for modeling? That would be like having a car sans everything except the engine. The 3d viewport works closely with almost all the other views, if you don't need a feature, just hide it. Once you grasp the basics of how to manipulate/transform vertices the curve is flattened immensely. The only tip I could give is to learn the keyboard shortcuts. If you still insist, you could build your own version of Blender and omit many of the other things it comes with now by default, the game engine, cycles etc.

In a word, clarity.

I realize it would be redundant, that's not the point. I'm hoping for 3d modeling software that is free, and easy to start with. Maybe blender isn't the solution to that problem, but i've spent well over two years using blender, and i honestly have no idea what half of it does.

There are some helpful "blender for beginners" video tutorials, but not many, and i've looked.

In trying to advance open 3d printing, this is a huge barrier. Modeling software is the future, and if we want any independence from stratasys we'll need a powerful, easy to learn, modeling program. Otherwise we'll be overwelmed by proprietary software that only works with proprietary printers, e.g. makerware slicing program.

2 years and no progress? I think you are doing something wrong. I have been using Blender for a long while now and there are parts of it that I see every day and don't use or even know what it does. It's a generalist's package, it can be overwhelming but just focus on what you want and don't try to grasp every bit of it. A beginner's series I would recommend is http://cgcookie.com/blender/cgc-courses/blender-basics-intro... and some useful resources can be found here http://meta.blender.stackexchange.com/a/43/12. I have a friend on fb that makes amazing 3d prints using Blender and a couple years ago he was fairly new to it, here's his blog, http://generativemodelingproject.blogspot.com/

I'm familiar with this course, the amount of time dedicated to modeling is not very large, but the overview of the controls is pretty decent.

The reason i'm not good at using blender is that when i hit a wall... it's a pretty insurmountable wall. It's much easier to just switch programs. If i need precision, openSCAD, if not, sketchup, both of which are obviously inferior.

The stack exchange post seems very useful, i'll look into that, thanks.

If you already understand the process of modelling, it's really nothing more than learning some keyboard shortcuts. I'd say that the modelling process itself - learning how a modeller thinks about the mesh and approaches various problems - took me far longer to learn than any particular quirks that Blender has.

There was a discussion recently about making Blender's UI "more friendly", which seemed to revolve mainly about coming up with a pretty icon for each operation and making left-click the select button. Pretty superficial stuff really.

Actually one of the premier packages for 3D does exactly this, Lightwave 3D. It has a separate exe for modeling. At least it did back in the day.

Lightwave is commercial and as someone who has used it in the past, you will be falling into the same learning curve pit.

There's Wings3D

I'll look into this, thank you.

"...the learning curve for modeling is so steep"

I agree Blender has an intimidating UI (most 3D apps do). There's been a recent campaign to improve the Blender UI that has (not surprisingly) met with mixed views from the Blender community.

In my view, Blender has the same problems with UX as many open source projects: you can't design by committee, but everyone feels they should have a say in the UX. As soon as you create a dedicated team to shape the UX, it creates so much resentment. (Ubuntu is another example of this).

Here is a Youtube video of a presentation at this years Blender Conference (Oct 2013) on improving the Blender UI. The Q&A at the end of the session is a bit depressing: the audience seem resistant to change and feel that making Blender easier to use would mean "dumbing down" the app.

https://www.youtube.com/watch?v=6aIA2LaB2Iw

OT: If anyone is interested in trying Blender or need a place to get solutions to problems, http://blender.stackexchange.com/ is a great resource.

I remember way back when, how 3ds max was something like 600 megs and blender could fit onto a floppy disk. Some really amazing stuff. Now blender seems to be 60 megs, but I imagine comparative things like maya are probably far in the gigabytes now

The size of a default Blender installation _now_ is ~214MB. Zipped, it's abt 40.

But how much of that is bundled assets vs actual code bloat ?

Definitely a steep learning curve, but like vim/emacs, it quickly becomes second nature.

I shouldn't have left all my Blender books on the bottom shelf of the bookcase where my old cat could pee on them. I need to pick it back up again. Good thing there are so many more resources these days.

Kudos to Blender! I would not have completed my 3D animation degree graduation project without it, years ago, and I picked it back up just a couple months back for video composition work. Looking forward to doing even more in it once my new computer arrives.

Here's an entertaining Blender podcast.

http://www.blenderguru.com/introducing-the-new-blender-guru-...

In early episodes he tried to rally people to fix the UI and make it more intuitive. He got so much crap (why should we change!?) that I think he's given up on that for the moment.

Actually no, what he was proposing was to reskin the interface to make certain things more obvious and consistent. However, it's not as easy to peel a UI off an existing program as non-technical people might imagine. It was an excellent initiative and the effort is ongoing. http://code.blender.org/index.php/2013/10/redefining-blender... and https://developer.blender.org/project/view/12/

Well, I didn't think what he was proposing would be trivial. He was simply trying to make the app more accessible to a larger audience.

Listening to his podcasts made be think the about the way Apple does it with constant UI refinement, and why many open source projects are really lacking.

It was a good exercise to follow that entire conversation Because one could really see a lot of the good reasoning from many sides. Out of it I determined that (I think) the devs and community will concentrate on menus, context menu and better tooltips, as well as giving some keyboard only commands actual menu options. These are things that are low hanging fruit. After that, who knows!

OT: The URL is http://code.blender.org/index.php/2013/12/how-blender-starte...

I've never seen "index.php" in the middle of a URL like that. Very quirky.

The web server is configured to use something like mod_rewrite to pass everything after index.php as a query string parameter to index.php. Many sites use this (though most don't do it so nakedly) to make pretty URLs that search engines and users prefer.

Essentially, the request is equivalent to index.php?Article=2013%2F12%2Fhow-blender-started or perhaps index.php?Year=2013&Month=12&Article=how-blender-started.

They also need the following in the .htaccess:

RewriteRule ^index\.php$ - [L]

RewriteRule . /index.php [L]

Actually, you don't need to edit your server -at least with Lighttpd-.

PHP automagically (or is this expected behaviour?) gets the string appended to /index.php for you! Check out the example at http://private.woutervdb.com/examples/url-query/index.php/We...!

I've made no changes to my webserver configuration. Works on Lighttpd on my laptop and webserver, not sure if Apache and IIS do this as well.

It's called PATH_INFO and is part of the CGI spec ("The PATH_INFO variable specifies a path to be interpreted by the CGI script"): http://tools.ietf.org/html/rfc3875#section-4.1.5

It's a useful technique for routing to controllers without needing 'mod_rewrite' installed.

I really hope that something like Blender Composite Nodes and Blender Python Nodes can replace Quartz Composer (which apparently has been ignored by Apple anyway).

It is a pity that you can't tap into design tools unless you have a Mac.

Offtopic, but I think a huge reason for that is how fantastic the Apple Frameworks are for any Pixel/Vector/GPU/UI stuff. Quartz Composer offers an API to include it into apps, there's CoreImage, CoreVideo, Quartz2D, SceneKit, CoreAudio, and much more.

SceneKit [1] alone, makes it unbelievable easy to create animated 2D/3D interfaces that combine various media types.

Due to this, it is probably easier to create a sophisticated design application for the Mac platform than for any other platform [2] [1] https://developer.apple.com/library/mac/documentation/3DDraw... [2] Based on some reading it seems that Qt is starting to offer similar functionality, especially with regards to Filters/Pixels/Vectors, but I'm not really sure about that.

You may be right, and I'm no expert in this, but that is not a large part of the reason.

I think it was a well designed piece of software by itself - and the node based programming model lended itself well to interface design.

Right now, it is more a case of adoption. For example, Facebook is investing a lot of resources in creating stuff like Origami for Quartz Composer. I'd be willing to bet that if Facebook puts even a small part of that resources into supporting something like Blender, then the fundamental differences at the OS level should'nt matter.

The early source code files from the screen shots are a reminder that big things start small; very small.

Reply on news.ycombinator.com