This is what most beginners struggle with, thanks to glTF we have a much better format which works for most features. Still a lot tools don’t support it or like in my case, i would have to upgrade with about $1000 to the latest C4D where this feature got added.
I would like to know how good your workflow with your modelling tool and the choice of format works for you. A lot formats crop off features such as animations, morph buffers and material features. Currently only glTF exports all PBR maps.
The tool, the format, it’s version, the exporter and the loader can all shrink the amount of information handled, especially with feature rich formats like FBX or COLLADA there are many versions what can make them incompatible with the loader for THREE.
I personally experienced with FBX from Cinema 4D and Blender that there is only a minimal amount of information exported, materials especially don’t store back/frontface culling, transparency, opacity and many other features that have to be fixed in another step. FBX works best since it supports animations, i use a converter which applies all further details and stores it in a binary format, for any converter instructions i use the object and material names to store information. But another flaw of the exporter or C4D is, if the UV map is split up, it will turn the entire mesh into a non-indexed geometry, i have to fix in the convertion step as well.
I personally really like working with Blender. Especially now that 2.8 has been released, transitioning from other programs like Maya or C4D is very intuitive, since the vast majority of controls have been updated to match industry standards. Plus, it exports to GLTF very easily, and it costs $0! (Although they accept donations).
As you said, there are always quirks or features that are going to need manual fine-tuning, but I just chalk those off as a part of working with brand-new open-source software. GLTF is being developed by the Kronos Group, and ThreeJS is developed by… ThreeJS developers, so chances are their features will not always work seamlessly as if it all was done by a single organization. Shoot, not even Apple can release products that work with each other 100% of the time.
As the GLTF spec matures, and developers add more features, I think the format is going to become robust enough to handle all needed features. With that being said, GLTF already handles animations and PBR materials. I haven’t had any problems with the format in the last year I’ve been using it.
Now that’s a surprise, i will try it out immediately, since what always made me dislike blender a lot was mainly their alien cockpit. Just noticed the grid in blender looks exactly like mine, even the division
A main reasons i use a converter between is that it bridges any missing features and project specific features like generating impostors. Also that glTF is a generic and common format, using a custom binary gives me more specific control like for progressive chunks and adds a simple but helping level of protection, since a glTF can be directly opened in DCC tools.
It’s worth noting about Blender 2.8 that this version is so disruptive that all the old tutorials over the internet are no longer relevant… So here we have a new tool easier to use, but with less learning material. It should not take too long to have new tutorials about everything there is to know though.
opacity & emission inputs right in the pbr shader (no longer necessary to build weird node structures).
Material usability has been greatly simplified for us. For example, connecting ORM map, with rgb separation is pretty straightforward now.
Aside from that, skeletal, morph and SRT animations are all supported and export properly (as far as I’ve tested it ), and many many more features.
As mentioned in comments above, the interface and overall workflow has also been made clearer, with great official tutorial series on youtube/blender cloud covering all the basics you need.
I think it works great with Three.js, I highly recommend it
Edit: also its FREE! and open source, just like Three.js
I used Blender 2.8 to create the model aircraft flying in the clouds. I was able to switch from an obj model to a gltf model (using cycles) very quickly. When I used obj, I had to split the model into two models - a reflective model and a non-reflective model. And the insignia on the shiny model were unrealistically reflective. Once I switched to gltf, I was able to combine everything into a single model and use masks to make the insignia nonreflective. And I was able to use glb to combine the model and textures into a single smaller file.
It improved massively, from unusable to actually very good i will still stick to C4D, it has some other features, i’m used to it and i need to say the tag system of C4D is just the most intuitive solution, but i will use both now. That they actually have a toolbar now with the different edit modes is like a quantum step forward, like you said, before most was hidden behind obscure shortcuts.
Open source is nice and all, but ultimately it means non-profit, no payment and i see that more critically since it can also ruin a market with something that might not even offer reliable support (not related to this case). For huge projects such as Blender this is different, they receive millions by donations, sponsors etc. and there is no real easy affordable alternative, since this kind of industry level software usually ranges around 1-3k.
Blender is one of the greatest projects for sure, it’s only real weakness was the U/controls.
I don’t get on very well with any modelling software, though I use blender from time to time because it’s free, and the sculpting tool is almost, but not quite as good as zbrush. Unless you have a huge budget, it’s a no-brainer.
The UI isn’t very intuitive, though it now gives you the option to choose between 3dmax, maya and default blender controls, which is a great idea, and shows that they do actually listen to the community.
Also, I’ve been playing around with some of the VR options, like tilt brush, blocks and medium.
It’s early days with VR, and it’s not completely practical, but I think its the future of 3d creation, especially now I can upload a model to google poly, and a creation video to youtube, all from within the VR environment. I’m pretty sure i was building in minutes what would normally take hours, but I’m not that great at modelling. It’s now feasible to build in VR, and publish to webVR, that’s something to think about…
Yes, if you can, stick with what you are used to. I haven’t figured out all the controls/capabilities of Blender 2.8. So some of the smallest things can be very frustrating.
For those who are interested in modeling aircraft, Witold Jaworski has an excellent publication that describes how to make highly detailed/realistic aircraft models using Blender. And he is updating his publication to include Blender 2.8.