WebGL Studio | Hacker News Reader

WebGL Studio

webglstudio.org

320 pointsdesp88 comments

This blows my mind. I've got bugs with a carousel today and people are building this. So impressive.

it is just a matter of time, I've been working in this project fulltime for three years

As soon as you start trusting other people's code and reusing things (eg, don't write your own carousel) you can start building cool stuff on top of other people's work. This editor is standing on the shoulders of litegl.js.

Other people? Javi Agenjo is the author of both WebGLStudio and litegl.js :)

Is he? I didn't realise. Talented guy.

Litegl grew as the requirements of webglstudio grew. I encourage my students to do not use lubraries so they have full control of ehat they do. Thanks for the compliment :)

The readme says litegl is a fork: https://github.com/jagenjo/litegl.js/tree/master#readme

Evan Wallace should get some credit. Evan created that fantastic WebGL Water demo.

http://madebyevan.com/

https://github.com/evanw

True, without his lightgl library this project wont exist, but the first thing I did was to strip all the things I didnt like and improve from there. There is hardly any function in litegl that remains unchanged from lighgl.

This is amazing and feels very fluid. I think it is part of a web-based software trend that will surprise a lot of people who think of the web at best as an interactive document display system. Another app that has a very snappy UI like this is http://OnShape.com CAD software, which is full featured 3D solid modeling without any plugins (all JS and WebGL).

WebGL allows you to build completely custom UIs, and these are well suited for 3D apps (which typically use OpenGL on the desktop, so transferring knowledge is trivial).

However, between custom OpenGL UIs and HTML-based UIs, there is an entire spectrum of native apps (mobile and desktop) that make use of platform conventions and standard functionality. This space is difficult to cover with web technologies because you have to simulate the look & feel and make the site feel like an app.

Many people have tried to build "just like native" toolkits from WebGL and DOM, but it's pretty tough. WebGL is very low level for this, and DOM is much too high level for many things.

As a user I have to say it doesn't matter me if your app looks native or follows HIG. That's just a convention used to make apps behave predictably and smoothly. If you can accomplish that with your own sensibilities that works too.

The common user interface/HIG idea makes less and less sense as software begins to do more and more things.

IRL, we don't expect toilets, hammers, bulldozers, shotguns, and surgical scalpels to have the same user interface features, and it would suck mightily if they did. Unfortunately, as those things get computerized we wind up getting the same crappy touch-screen menus and whatnot bolted on to them (imagine what using a hammer with a touch screen interface would be like).

I don't know the answer to how it could be brought about, but we definitely need some radically thinking in UI design. 3D libraries seem to at least provide the tools to make some new tools.

Those real life tools have a far simpler interface than, say, a mail client. As software becomes more complex, HIG increase in importance, because there is more to learn and HIG enables us to transfer knowledge across apps.

The actions you perform with a scalpel (or even a bulldozer) are far more complex than sending email, so arguing that they have a simpler interface doesn't really support your case. Quite the opposite, in fact. I would argue that they have simpler interfaces because those interfaces are actually optimized for the task, rather than being some jack-of-all-trades compromise.

But by having a shared convention, we avoid having to relearn a new one for every app.

You really wouldn't mind if the app used control-p for delete, and control-z to quit?

Now you mentioned it also passed my mind to scrap DOM and use my own gui, lots of advantages but im worried it will slow down the development. Indeed all the timeline editor is a canvas. Its worthy to consider it.

Doing your own GUI affords a lot of advantages in terms of how you structure your code. As soon as you decide to render every frame you can take advantage of immediate mode techniques that work very well in editors IMO (unfortunately, many of the examples of using and implementing these online are not very good and imply restrictions that are not inherent with the technique...).

That said you seem to be far enough along that redoing the GUI would not be a good idea really.

Also the DOM is a fairly okay retained mode UI, so if that is your preference I would just use it.

But i have stumble in so many problems thanks to the DOM inconsistencies. more time spend in DOM problems than webgl problems

In many ways WebGL is akin to a plugin. It's effectively replaced what plugins like Flash and Silverlight offered in the past by bolting on a new piece of functionality to the browsers.

The difference is the browser vendors are in control of the implementation that will make it slightly more secure, but it's still a long way from the HTML/JS/CSS environment people though would replace plugins.

OnShape is amazing, built with Angular too :)

That's great, it looked already good in 2013 when it won best paper (http://gti.upf.edu/webglstudio-gti-paper-goes-viral-at-web3d...), I'm happy to see it reached a 1.0 release.

It used to be flash based if I remember correctly -- at least the 2013 version.

Nop. But i have done 2D editors in flash before

Good memory. Its surprising how from the first version to the first release you need so much time

Do they have permission to use the WebGL trademark in the name? Khronos has quite some info about that, but I don't see immediately how their use fits into the terms. https://www.khronos.org/legal/trademarks/#webgl

Having read through the Trademark Guidelines[1] it would appear that they may have overlooked it. I can not find any attribution on Github or within the web app itself. [1]https://www.khronos.org/files/legal/khronos-trademark-guidel...

yep, its a placeholder name, I will change it as soon as a nice name comes to my mind.

"Decahedron.js". From the three-dimensional variant of the decagon, the symbolic shape of a spider web.

Mmh, yes, sorry, I'm bored. :)

Too long!

You could always go with the short "d10.js" if you want to add another level of abstraction.

Something related to light or rooms seems to be the norm (Blender, Studio, Cinema, Lightwave).

How about "Photon Studio", "Sunshine 3D" or "Lightbulb"?

It is not a 3D modeling tool, rather this is a Game Engine. Thus Unity, Unreal Engine, etc is the more proper naming style to follow.

https://www.youtube.com/watch?v=aaZTSDfT1o4 Light or rooms?

Hahah those names also came to my mind, as frame, glow, lux. But havent decided. Thanks!

Awesome project! Congrats to the author for such clean code base too!

Maybe slightly off topic, but one of the most important takeaways from this: you don't need a bulky JavaScript framework to build complex applications. In fact, you may be better off without using one.

I'm sure there's a very narrow sweet spot where a given JavaScript framework can make you more productive. But if your application is complex enough, like this editor, the framework will probably work against you. And if it is simple enough, it may be better to not use the framework, so you don't grow dependent on it and find it hard to remove it if you need to grow the app.

Instead of the framework imposing the structure and constraints, a team can decide on the structure, agree on the constraints, and build libraries helpers according to that. Way more economic than the alternative: working around buggy and/or undocumented framework features, trying to catch up to constant changes and "upgrades", fighting against the JavaScript marketing machinery trying to decide which of the "Top JS Frameworks" to use, etc.

exactly what I think

I have to say I feel very proud that this comes out of Pompeu Fabra University.

I learned so much there, so many amazing people.

Great work guys!

merci :)

If you want to create 3D models and animations for WebGLStudio, you can do so via:

http://Clara.io

Not open source, though.

To be clear WebGLStudio is a game engine, rather than 3D modeling tool -- thus WebGLStudio and Clara.io are not competitors.

THREE.js has a similar editor:

http://threejs.org/editor/

As long as I like that editor, it's not even close to this one.

The main diference is that I created the engine to fulfill my editor needs and not the other way around. Also not having legacy allowed me to recode lots of parts without problem.

This being said I love the idea to stop at runtine any application coded in threejs and have the freedom to edit it from the console using the chrome extensition.

The THREE.js editor isn't very good or well developed to be honest. It really feels like a proof of concept more than anything intended for real use.

This seems much more polished in comparison.

This project is amazing, especially considering that you have been working on it on your own for 3 years. Lots of perseverance!

A similar project (albeit not open source) is Goo Create by Goo Technologies: http://goocreate.com/product/

It is being used by several companies in pretty large productions (including games, ads, educational projects, etc) both for desktop and mobile devices. You should really check it out.

Full disclosure: I work for Goo Technologies as a Frontend developer.

I know goo create, great project, keep it up!

Looks amazing, but is there a demo scene I can load to test it out?

There are some. Check the videos to see how. You need to login first as guest

Very cool. Seems better than the competitors (playcanvas, three editor, ...) by some amount. Being 100% open source including the editor is pretty huge IMO

The big questions for me are: What model formats does this support? How well developed is the animation support? How is the performance in large scenes and on mobile?

Good questions. It supports collada which is the default ascii format por scenes, i managed to export complex scenes with animation without problem but there are some cases that do not work.

Performance could be improved, im aiming more to flexibility but i managed to have hundreds of objects with independent materials at 60 fps. For mobile i want to create a lightweight rendering mode because the current one has too many cases to consider.

The next release will be focused in bugs, and the one afterward in performance.

This is really great and I would not know where to start building this. The only thing that I would like to have is the use of "modern" JS tech like ES6 and a build pipeline... Surely this can be added later as well.

There are some features in ES6 that could improve the coding but nothing that is really game changer for performance, but ES7 will have SIMD support, that would be amazing

Is there any doc or readme for the geometry format? I've downloaded the source, and am reading, but some breadcrumbs to guide one would be great.

Totally right. I want to publish a doc where i store all the important info so people can mess around.

My binary format is similar to the WAD format, it is a way to store a json with the typed arrays encoded in chunks

Thanks for using a right-handed coordinate system.

Hahaha. Just using the OpenGL default

This is really nice! how did you do the UI?

I created my own js library called litegui. Check it out at the same github account although no documentation is available right now.

Thanks, this is great! Shame I'm not in Js/Web anymore. I like your TV interface on your site as well. Keep up the good work!

Is the author here? What license do you plan to use?

It says "Released under the MIT License" at the bottom

Im here but my license knowledge is limited, thats why I just applied the MIT which I think is the one that fits better

MIT is a fine license, but it worries me when I hear this. I've seen too many people who make assumptions about what copyright offers (and doesn't). It sucks, but it's a good idea to spend some time getting familiar with licenses so that you don't offer something that you don't intend. Also, if you are using 3rd party libraries, you need to make sure that they are compatible with the license you choose.

Just friendly advice from someone who has seen things go badly before ;-)

Your project is fantastic. And, releasing the source code as free is the best thing you've actually did. This would be that start of it!

Impressive.! Lets Hope we play the next gen 3D Games in browsers

Very impressive.

Great job! Keep up the good work!

Brilliant work!

Wow

wow!

how do I create 3d models with this? needs some video tutorial but this is very cool.

btw how do you create applications like this that rivals desktop apps?

this is not a modelling software, you need to have 3d models and import them here.

People think there are too many diferences between developing for desktop and developing for the web, but there arent, it is just a matter of time

Even though the rendering performance is very snappy and efficient, to me this fails on the most important thing for this sort of software - UI. Can you imagine someone totally new to 3D modelling opening this thing and just going for it? It's completely unintuitive and yet, because it's so accessible that should've been the primary concern.

But it is not a 3d modelling software, it is a 3D scene editor, like Unity. Improving the interface is the easy part, for me the important aspect was to be sure that all the important features are there, so now I can focus on improving the experience.

Exactly. 3D engine !== 3D modeling tool. If you are interested in a WebGL-based 3D modeling too, give http://Clara.io a try.:)

BTW we should figure out a way to connect Clara.io to WebGLStudio as it can be a source of content, both models as well as 3D animations (Clara.io has a full keyframe editor.)

Nice suggestion. Worth considerin

Is there any 3D modeling software that is intuitive enough for a beginner to jump in start creating 3D scenes? The barrier to entry and learning curve of 3D modeling has always been high and steep.

Sketchup

ZBrush and Wings3D. Completely different, but both are super easy to start with.

I dont know about Wings3D that is probably casual tool. But ZBrush? Are you sure? ZBrush is notorious for its crazy learning curve. I mean sure you can do something with it instantly but to make something meaningful is hard.

Besides you should compare it to game engines like unity or ue4, its very differend thing than 3D modeler.

Wings3D is not a casual tool. It is a fully featured 3d modeling application that does that very well. It works on the box modelling principle. ZBrush is really easy to get into and make something instantly like you say, but yes it's really difficult to learn. I started with Wings3D and still use it.

UI is following standard practices in 3D applications from what I've seen so far. It's intuitive for someone used to those apps. Those apps aren't something you dive straight into though. It has a learning curve, but it is logical.

Yep. I did t want to reinvent how these tools work. Most of the ideas come from cinema4D after effects and unity

My first thought was Nuke :) Maybe because of too much use, in my case.

The object/component system looks like a Unity3D ripoff. Even component names ("MeshRenderer") are unchanged from Unity.

Also it is interesting that it's written in JS without any higher-level technology like TypeScript or Dart.

A little bit off topic, but every time I try any kind of similar application (this, a CAD, a PCB editor) that is implemented to work in web browser, I lack feeling of "stability" of a desktop app with all the files stored locally - I would rather use a desktop app with inferior functionality.

Component based applications have existed for a long time but obviously there are choices made in webglstudio that tried to appeal to unity users, just because unity is the defacto standard.

About the stability of desktop apps I agree with you, but sooner or later that will have to change, so thats what we are pushing.

> The object/component system looks like a Unity3D ripoff

Component based systems have existed since 1968, way before Unity:

https://en.wikipedia.org/wiki/Component-based_software_engin...

Reply on news.ycombinator.com