Hacker News new | past | comments | ask | show | jobs | submit login
3D Modelling in a Browser (vectary.com)
170 points by redindian75 50 days ago | hide | past | web | favorite | 72 comments

See also: Clara.io [0], Onshape [1], Lagoa (bought a few years ago by Autodesk) [2]

I must confess I struggle to see the long-term market for these services - I was working on a startup in the field for a while, and eventually decided it wasn't worth continuing.

The only thing these services really offer above desktop equivalents (which Autodesk totally dominate) is the collaboration (like Google Docs for 3D). But do you really need to collaborate to create 3D? Maybe to review/annotate etc. but to actually collaborate during the process of creating an object/scene?

To put it another way, are you going to convince your entire art team to leave behind the toolchain they have used their entire career, just so they can 'collaborate'? And if so, do you think Autodesk are going to let that happen? (Obviously not, as they bought in to the field a while ago).

[0] https://clara.io/

[1] https://www.onshape.com/

[2] https://architosh.com/2014/11/autodesk-reportedly-acquires-c...

I’ve used Solidworks for 15 years, and I think Onshape is close to as good for modeling, and now much better for collaboration and version control. Collaboration is great, but version control and configuration management is the killer feature of online CAD tools. These also facilitate typical collaboration which is at a part level more often than part level. The existing tools rely on a lot of discipline and are very hard to set up and maintain, and generally requires people assigned full time to maintenance (or full time to manually handle it in the filesystem). On the other hand, Onshape does not allow engineers to invent their own workflows.

For a software analogy, the existing tools are like sharepoint at best and cadfile_v2_final(Mike).prt at worst. Onshape (and I assume other) online tools are like github.

It's kind of incredible how big a pain in the ass the SW license and PDM servers are. Definitely requires full time IT support and small riots are likely to form when they try to upgrade them during a busy period, especially if someone suggests using an SP1 version.

For lightweight & quick conceptual work, even final deliveries most of the time, nothing beats a browser based available-everywhere tool. No files to sync, same tool to both view & edit the files.

I used Industry heavyweight Photoshop for 15+ years, Sketch for 4 and dropped everything for Figma[1] after playing with it for 2 days. Now my UX comps, UI sketches all available EVERYWHERE. All inspirations, rough concepts all stacked away in any browser I have access to. I move to different machines a lot (Macs/Windows). Used to carry around a USB stick (synced to Dropbox from my home computer).

Now it is nirvana - No syncing, No dropboxing or figuring out where my prototypes, v1, 'final-final-v1-1-edit-final' files, & source files are. Not only the files are available, but the viewer to view those files are right there.

[1] www.Figma.com - Browser based UX Mockup/Prototype tool.

That is a compelling argument. Be careful with the terms of use though, they can hide a few surprises. Example: in Vectary's terms of use[1] I interpret the section 4.6 as "backing-up your data is your responsibility. If our servers burn, we might not be able to recover your data and we won't be responsible for it".

That being said, the limits of Vectary's services seem to be clearly stated, so one can adapt to them (by using some storage drives from OVH, Digital Ocean or any such provider to back-up the data, for instance)

[1] https://www.vectary.com/terms-of-use/

I use Tinkercad[0] every single day, for rough mock-ups through to complete exported STLs that I then 3D print. I have Fusion 360 but find it faster/easier to whip something together in Tinkercard unless it has any complexity to it, and generally there has been nothing 3D printable that I can't do in Tinkercad, leaving Fusion 360 more for larger project design (stuff I'm going to make in mixed materials, like a complete robot).

I also love that if I get an idea in bed or when away from my PC I can use my laptop to log in and mock it up, or work on ongoing projects there.

As a maker/tinkerer/hacker I actually think Tinkercad (and all the related web based design tools) are one of the most exciting advancements in this hobby, along with 3D printers and DIY CNCs.

[0] http://tinkercad.com/

I'm 3D modeling on OnShape right now, and it's just because I can't be bothered booting into Windows to get Fusion. That's pretty much why I stopped using the latter, and only use OnShape. It's convenient.

How about casual use?

I know 3DS Max very well (at least I used to... Hopefully the synapses are still firing) but it's a zillion dollars for a licence and I probably need to use it maybe once a month.

I work mainly in Unity and 3D modelling is not my main focus. I would use a 3D package for some light editing and as a glorified file converter.

Nobody makes software to do this at a reasonable price so I usually head to clara.io

Definitely check out Blender -- they released version 2.8 a month or so ago and its miles ahead of where it used to be especially when it comes to interface.

Also free, cross-platform and super powerful. It also has a brand new real time rendering engine called "Eevee" thats really impressive.

> To put it another way, are you going to convince your entire art team to leave behind the toolchain they have used their entire career

Yes. Because they will start out using these tools. Having experts be able to oversee novices is very powerful, esp w/o having to transport files, load and diff, etc. An expert can hop into someones session, show a technique, do a quick critique and move on.

CAD tools generally have pretty good lock in. So that certainly helps with the long term the prospects.

I’m not going to convince my team to move over from the tools they use. But the engineering team that makes their jobs irrelevant because they bid by 50% lower through increased efficiency will.

Rendering is computationally heavy, so a browser is not ideal. However, offloading simulation, file storage, version handling, etc to the ‘cloud’ has boundless potential for productivity and efficieny increases.

gpu does a favor to this kind of rendering?

Solving the model is usually CPU bound while rendering is usually done by the GPU.

Yes i agree. There are some exceptions. You have Fusion360 (also autodesk) which is a slim version of Inventor. It is web based and used by lots of professional people. It also works quite similar to Inventor. So switching isnt that hard. But to go from 3dmax or maya to an entirely different ecosystem. I don't see that happen anytime soon.

From a hobbyist perspective, I gladly welcome the onset of cloud-based tooling. Most of these options (OnShape, in particular) have generous free tiers. I used SolidWorks in a previous profession and generally was very happy with it, but in no way could I afford a single seat license for "play."

I'll plug our product here as well https://3dc.io

I've recently started using 3DC and find it to be a wonderful creative tool. It's super intuitive and easy to use on the go, on a mobile device. I look forward to using it more in the future.

Nicely designed and has potential.

Missing (couldn't find) many useful and necessary components (e.g. explicit sizing, rulers/measurement, pivot points).

I am a Mechanical Engineer and we use SolidWorks for CAD. I was surprised to find there exists no 3D modeling software (Nurbs / Polygonal based) that I can model using accurate dimensions. 3DSMax, Maya, Modo, C4D, Blender, Houdini, you name it. I understand that they’re not doing parametric modeling such as SolidWorks, Inventor, etc. but there has to be a way to merge both modeling paradigms. I was honestly shocked at this fundamental lack of capability.

I wanted to accurately model a product for visualization and there is no way to specify dimensions. Everything is done by “eye” in the visualization industry. When accuracy is required, especially proportions, manual element dimensions are typed in during creation of primitives.

I am used to the extreme discipline and constrain based modeling in engineering. Does anyone know why visualization folks don’t care about dimensional accuracy?

In blender at least you can specify the size of any primitive, measure the length of any edge, measure between any two points, move a vertice, object, or face by a specified amount etc. so this is just not true for blender at least. You can also annotate any dimension easily with a dynamically updating annotation. What you can't do easily (unless you're prepared to do some serious python scripting) is parametric / constraints based modelling because it's not a parametric modelling program. I'm not sure what else you'd want.

As to why visualisation folk don't care about dimensional accuracy, we do when necessary, but unless we're designing for 3d printing there's no point in worrying about things like clearances and tolerances since we're not making a physical object. Also, especially if we're doing arcvis, the demand is to knock out large quantities of models very quickly, not produce a perfect replica. I'm not going to worry about the exact dimensions of a piece of complex equipment or furniture I'm modelling if I need to produce a fully textured model in under an hour.

I’ve seen countless tutorials - Eames house to Braun products, even full furnished houses on YouTube. No one cares about dimensions. It bothers me so much that it should be fundamental to designing anything - I would argue, even dragons.

I think it’s the culture and people involved in 3D modeling that do not have the discipline - at the risk of sounding snobby. If customers demanded a feature, modeling software companies would supply those features.

“Ball park” dimensioning is like nails on a chalkboard.

There are really only so many modeling kernels.


> Does anyone know why visualization folks don’t care about dimensional accuracy?

Maybe because the intended output is more about proportion than absolute dimensions?

Besides, you can specify dimensions in the programs you mention, Blender, Modo etc. See here for Vectary: https://www.vectary.com/3d-modeling-learn/dimensions-how-to-...

That’s the surprising part: These software packages are used for both, proportional modeling of dragons and animated characters to architecture and industrial design visualization. The latter, requires absolute dimensions.

Maybe I missed something in your question, but in every modeling package, even online ones, you have always been able to numerically specify the size, position and orientation of every object in your scene, with a precision of several decimal points. No need to rely on your ‘eye’ at all.

Yes, you can model accurately in these programs but you really have to understand how construction documents are put together. You aren't going to be able to quickly take dimensions from a drawing and convert it into an accurate 3D model with any of the big 3D model packages out there. When I was an architect, I stayed far away from 3DS max, Maya, Blender, etc because modelling in them felt like drawing with crayons. Its the difference between using pencils, rulers, and compass on a drafting board and using paint and brush on a canvas. And that is what's surprisingly lacking in the market, the accurate measured generic 3D modelling program.

> Everything is done by “eye” in the visualization industry.

Im not sure if we tried the same version of blender... Its pretty easy and even intuitive to enter numerical values for dimension, rotation etc.

Yes, I noted that in my post that you can specify numerical values when placing primitives. But there is no full fledged dimensioning like in engineering CAD. As someone said, designing in 3D modeling software feels like drawing with crayons compared to the precision of engineering CAD.

Rhino 3D is kind of ok as a nurbs modeler and has some dimensioning stuff.

I think my point was that nothing is stopping you from achieving the same level of precision as in CAD, but yeah dimensioning is much more useful for engineering tasks and working with meshes arent really suited for that kind of job.

Rhinoceros 3D is a Nurbs modelling software that uses dimensions and is used heavily in the architecture/arch-viz industry. But I agree, its shocking that every mainstream 3D modelling software has no concern for dimensional accuracy.

I hope you saw onshape mentioned in the other comments

Tried OpenSCAD?

Does per monthly pricing turn off a lot of people?

My usage patterns are pretty uneven and - although I know I can unsubscribe/resubscribe - I end up feeling like I'm probably wasting my money when I'm not using it (despite the low price)

It's irrational but there you go.

Does anybody consider a per-usage pricing model? Is there some reason why that is a terrible idea?

Yes, i do not mind paying for quality software, but i want to have control over it. This means payment once (...i am ok with buying new major versions as long as the developers aren't releasing a major version every year or whatever to force people repay) and i get an installer .exe (or whatever) that i can keep on my external HDD for as long as i want and there isn't any need for online authentication/DRM (i am ok with key files as used by Total Commander and WinRAR).

Sadly, as time goes on this is becoming increasingly less common.

I don't like that model for anything I want continued support for (excluding things that are more or less complete like WinRAR or games).

The problem with the traditional pay once (with paid upgrades) model is that in order to continue developing the software, developers need to be paid every month. So a hugely variable revenue stream needs to be converted into a monthly one. Maybe by being very conservative about spending on developer resources to leave a big buffer.

There's also mismatched incentives, since if all your revenue is from paid upgrades it might make more sense to focus on flashy new features even if the majority of users care more about stability and bug fixes.

There's the issue of how much effort to spend back-porting bug fixes to older versions. Users will rightly be annoyed with having to choose between bugs and an upgrade that (they consider to be) not worth it, so the devs probably have to put in some effort. But then with the old model the developers are putting in more work and fewer users have access to new features, and the total amount users are paying still has to be high enough to fund everything.

These days I'm happy to pay for things I use on subscription (as long as the total cost is reasonable).

Maybe a good middle ground is something like the JetBrains model, where a 1-year subscription also comes with a permanent license for the version at the time of purchase, so you can at least open your old projects.

As a user i do not really care about the developer's business plans and especially i do not care on trying to make my purchasing decisions based on guessing what the developer might end up doing in the future. I prefer to act on what is available for me right now and have control on the software. Hence the preference for pay-once, DRM-free software with an installer i can download and keep myself.

And BTW, a developer focusing on flashy new features instead of stability and bug fixes is something that is independent of the monetization scheme. If the users want stability and bug fixes while the developer provides only new broken stuff, then the developer is simply not listening to their users which can happen (and often happens) regardless of if the software is paid once or subscribed to.

TBH I think you are looking things a bit too much from the developer's side.

FWIW i do not mind 1-year subscriptions with a permanent license, although i'd prefer the permanent license to be for the end of the year, not the start (especially if there is a chance that the files you worked on with the last version wont work with the earlier version). It is essentially "buy once, get free upgrades for a year". Assuming of course the other things i mention (offline installer, no authentication needed, etc) still apply.

I wish it was a "If you use it this month, we'll charge you $15". I personally hate having to balance the tools I subscribe to. I mean even if I use something like EAGLE for once in a month, that one time is worth the $15 or whatever it is now. If I don't use it in February though, I don't wanna get charged for it.

For as much as reporting home for license checks these programs do, you think they would be a little understanding.

> I wish it was "If you use it this month, we'll charge you $15"

Yes, or even "if you use it more than 2 hours this month, we'll charge you $15".

It's mostly impossible to do anything seriously meaningful in under 2 hours except minor edits and file conversions etc.

Subscription models really need to grow up and be more dynamic.

Yes, it's a turn off and disappointing that monthly subscription is the only option for something as simple as saving the 3d model you just created to your computer.

Too many of these small time software and web app publishers just default to thinking people are going to subscribe and forget about it. For a 3d modelling program with many free competitors, it's wishful thinking.

The problem wouldn't be a problem if there was an option to buy outright. But they sold their investors the promise of monthly revenue.

100% agree. If I create 4 documents now I'm stuck paying $108/year forever to keep them on their server. No thanks.

Here is a video of what is possible inside a browser! https://www.youtube.com/watch?v=0dC1ZRdf9JM

Wow! (not affiliated with them)

Wow so it plays electronic music in background while you work. That's different.

...so, what kind of rendering tech does a tool like this rely on these days? WebGL? What's considered best practice on this front these days?

WebGL, which is now supported by all modern browsers, with WebGL2 (more flexible, fewer restrictions) around the corner. Most of the professional level tools as exhibited by the OP will be using a custom engine built on top of it.

For smaller projects, the excellent Three.js [0] is an easy way into creating 3D in the browser.

[0] https://threejs.org/

WebGL2 is not around the corner. Apple has indirectly made it clear they will never be shipping WebGL2.

WebGPU is "around the corner". Apple/Mozilla/Google/Microsoft are all working on it. Google and Mozilla will ship it. Apple and Microsoft remain to be seen.

Didn't Apple kick off the whole WebGPU standardization process? I can't see why they would do that if they weren't interested in shipping it. Since Microsoft is moving to Chromium, I imagine they'll just leverage that implementation rather than doing anything with EdgeHTML.

Agree on WebGL2, it seems unlikely that they'll want to commit to supporting it forever with WebGPU on the horizon. Plus Edge will probably not add support until switching to Chromium, so WebGL2 will probably never have a chance to be supported by all browsers until WebGPU is taking off anyway.

Apple “made it clear” originally that WebGL would never be supported in iOS, but eventually buckled.

So I wouldn’t write off WebGL2 quite yet. Though I agree that it’s been stuck ‘around the corner’ for several years now.


It's supported in safari, currently gated behind a setting though.

No it is not. That report is wrong.

Apple added a flag. They didn't add an implementation. There is zero code in WebKit to support WebGL2. All there is is a stub that returns a WebGL1 context when WebGL2 is asked for. Feel free to dig through the WebKit code. You'll see there is no WebGL2 code, all the functions are actually missing and no code has been added in > 3yrs. Try running the tests. (https://www.khronos.org/registry/webgl/sdk/tests/webgl-confo...), click off "all" then search for "conformance2", check the box and run. You'll see almost nothing passes.

Definitely webgl

This looks pretty cool. Tinkercad is another tool like this that works in a browser. I find Tinkercad is a little more intuitive but from what I can see this seems to be a little more powerful.

To get extended functionality, you need to sign-up for a Premium account. This tool need's to be more open in first place as 'fully functional' before it can attract people to sign up for premium account.

The 'collaboration' aspect isn't what will sell this product. It could have potential when they allow creator's and designers to fully explore it first.

So can you actually do useful work in this? Or is it just drag & drop of prefabs like the marketing would lead you to believe?

What about modifiers and tools, what can I actually make with this aside from drag&drop of prefabbed iphones?

Why would I choose to pay for some cloud BS over just downloading blender?

We've come a long way since VRML [0] running in java applets!

Re-live the experience: https://www.youtube.com/watch?v=eghlSdGvuC0

[0] https://en.wikipedia.org/wiki/VRML

> We've come a long way since VRML [0] running in java applets!

I'd argue "yes" and "no" on this at the same time...

With VRML, it was designed so that - like HTML - anyone could easily create 3D objects and "worlds" using a simple markup syntax that was very similar to HTML (IIRC, both are subsets of SGML?).

It was easy to create something really quickly that could be interacted with.

These tools, while they simplify the creation part (somewhat), are more akin to today's "web site builder" systems vs raw HTML in an editor.

What you get out of them, though, isn't anything like HTML - if you are lucky, you'll get a standard file format or maybe JSON - but then what do you do with that? Furthermore, it isn't likely very human readable or editable.

WebGL - and libraries like three.js built on top - are all well and good, and allow for much easier access for 3D graphics work - but neither are anything like VRML.

There are some VRML-like tools out there - one that comes to mind is Google's A-Frame:


So some stuff like the old-school VRML is out there, and VRML had it's issues, which is part of why it didn't survive (the other part being that it was too early for the available PC hardware - unless you happened to own an SGI workstation at the time).

It's one of those pieces of "pioneering" technology that came too early for the tech - but gave us a glimpse of what might be possible in the future (much like the first "wave" of virtual reality hardware technology).

To add to the list of other similar projects already mentioned, Nunu Studio[0] is a solid MIT-licensed tool based on three.js.

[0] https://nunustudio.org/

Dimensions? Master model? Assemblies? Or is this only for pretty pictures?

It claims to be for modeling not CAD. Think Blender not Solidworks.

So ultimately yes, it's for pretty pictures.

I have used Blender for CAD like work before, not because its good at it or anything, but it was good enough for my purposes and it was free. Boolean operations in blender are very good for that sort of work, you don't need to commit to anything and can easily change dimensions of sub parts on the go, screw modifier has also been handy at some tasks (like making screws and screw like features).

blender can create 3d printable STL files, so if we're "thinking Blender," then that implies we can 3d model for printing, if i understand what you're saying

I think a better distinction here is mesh vs. solid geometry. This is a tool for creating mesh-based models and has a focus on rendering. STL is a mesh format that you can print with consumer 3D printers and it looks like Vectary supports that as an export format, but it's not a product built for engineering or mechanical design.

That said it looks really cool and I would definitely print something made with it. Trying that out now, actually.

The distinction you are looking for is parametric vs. geometric.

Parametric models such as Solidworks or Creo use complex equations to derive the 3D geometry. Geometric models use a point cloud built from polygons to develop surfaces. These point clouds are defined by the user and are not as "stable" when used in product development setting where you are trying to create multiple iterations/sizes of an object

it's more like Brep is the distinction, is the internal representation a list of operations on a Brep model?

But it's true that model history is preferred for mechanics, but I have seen some proto model history in 3DS max with the modifier stack.

Thank you for clarifying. I wasn't sure if the Vectary editor was or wasn't parametric, but it seems clear that it isn't for building CSG models.

It lists 3D printing as a use case. It also has a dedicated set of tutorials on working with exact dimensions.


Amazing choice of typography and design!

Oh man I hate receiving spam from these people. They don't provide an unsubscribe link, and when I ask them to stop sending me these e-mails, they say they won't and that I should manually log in to their website and delete my account. I don't have time to deal with this but I hope someone takes them to task for that.

If you mark it as SPAM, it should cause their mail reputation to take a hit, and your mail server should prevent their emails from getting to your inbox.

I think their marketing emails do have unsubscribe link, only the service emails don’t cause they are obliged to inform you about important changes you are subscribed to, like changes in terms of use... however unsubscribe should be provided, cause people don’t really see a difference.

Thanks for the warning. I don't need more of that crap filling my spam folder.

Microsoft did it for a long time if you signed up for their Azure service, but just recently finally provided and unsubscribe after so many people complained.

@ dcplogic

Yes, the Streisand effect is still news to some control freaks, who think they can make humanity march according to theire terms.

> I should manually log in to their website and delete my account. I don't have time to deal with this...

In the time it took to write that comment you could have deleted your account. At least they do have a "delete account" button, a lot of services do not.

BTW, you can sign up with a fake non-existent email to check it out, you don't need to confirm your account to start using the app.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact