I didn't really expect to see very much moving from 2.9 to 2.91, but I just keep on scrolling and there just keep on being more features when I scroll down. And they're substantial features, too! The sculpting tools and the holdout option for grease pencil look like they're going to be really useful.
And the release page is performant and pretty, and it works without Javascript but it still has all these great little diagrams and demo videos for each feature. Like, you can look at this page and even if you knew nothing at all about 3D modeling you'd still want to go download this thing immediately.
hahaha, great comments. Blender is so brilliant. I met Ton Roosendaal last year when conferences were still a thing. He is awesome. Those guys/gals are amazing. They got a huge grant from Unreal, its great to see software like this just crushing it.
You are thinking in bay area figures. The blender foundation is based in the Netherlands and engineers make far less in wages over in Europe. And lastly by open source standards those grants are quite significant.
It's been waiting for review for about 9 months. :(
If this gets reviewed and merged (probably with some hopefully minor fixing), it'll put the foundations in place for Blender to start competing with CAD/CAM packages.
Which would be a really good thing (as a CAD/CAM and otherwise Blender user). :)
In industry, the model file (eg STEP) is turned into a gcode action plan for the machine cut out the part by CAM software. eg MasterCAM (https://www.mastercam.com)
---
With that branch (D6807), my understanding was it enables proper in-depth NURBS (solid geometry) support.
The commit comments in that branch indicate it also allows importing of Rhino (3DM) and subdivision surface (subd) files.
If that's the case, it's the needed foundation piece to allow for creation of importers and exporter for widely accepted CAD/CAM formats.
The second an even basic STEP exporter is functional in Blender, it will start pulling users across from the expensive proprietary CAD/CAM systems. As almost no-one actually wants to use them, it's just a "lack of choice" problem. ;)
I use both Rhino and Blender and I for one am glad that both packages exist. Part of the problem (from my POV) w/ Blender is that it tries to do a bit of everything and understandably ends up coming up short on those other things it tries to do. Sure, it's somewhat convenient to have it bundled with a video editor, but besides some really basic functionality what purpose/segment does it really serve? Before the release of 2.80 it even shipped with a built in game engine which very few people ever really used so why devote valuable engineer resources for such a small niche user base? Why not focus on improving its core features rather than try to compete w/ specialized tools such as Solidworks or Catia when their core user base is not interested in modeling planes or rockets which can actually fly?
Personally, I'd rather use Blender instead of having to fork out many thousands of $ for Solidworks, or Catia (where even available).
Especially when the other parts of Blender (animation, grease pencil) are already very complimentary to modelling and already support the right workflow(s).
You personally might prefer to use Rhino. But the vast majority of new users who come to the CNC places I frequent (eg PrintNC Discord) would easily choose a functional Blender over any non-OSS modelling package.
Does it really make sense to use Blender as a CAD/CAM package?
CAD is all about building mechanical parts with features that match needed specifications. For example you need a block with holes in it with particular spacing, particular diameter and particular depth. CAD packages like Fusion 360, SolidWorks, FreeCAD etc. are all dedicated to this task and make features like constraints and measurements front and center.
Blender on the other hand is, at least in my experience of it, all about bringing somebody's artistic vision into a 3D environment. It's usually not important to somebody's vision that parts meet dimensional constraints, it's more important that it looks "right". And so Blender is full of tools that allow humans to change things to arbitrary dimensions.
Does it really make sense to try and fit both of these use-cases into one tool? I can't help but think that Blender would come out worse for it.
And that's CAD, I really can't come up with any reason why CAM would make sense in Blender.
Just to point out, Blender already has basic NURBS capabilities.
They're just so many years outdated, that they're pretty much unusable which is probably why you're not aware of them being there. ;)
This patch is to update the NURBS capabilities in Blender, so the "artistic vision" objects people create can then be turned into real life objects via industry standard processes (eg CNC machining).
If you personally don't want to use the NURBS pieces, then feel free to ignore them. But please don't block other people from doing so. :)
> If you personally don't want to use the NURBS pieces, then feel free to ignore them. But please don't block other people from doing so. :)
I'm not saying NURBS shouldn't be fixed/updated, I'm just debating whether "so that Blender can be a CAD/CAM package" is a reasonable motivation for it.
Blender is an amazing tool that keeps getting better. The killer feature for me is scripting. I’m a huge fan of running Blender in Python batch mode as part of an asset pipeline; being able to dynamically generate assets from a tool like this so easily is really powerful.
Yea, i'm new to Blender and am working on a 3D->2D game - in the planning stages. One of the early things i identified was that an extremely well optimized (for dev UX) asset pipeline needed to exist. So i started testing Blenders API.
It has been great so far - and even more surprising i'm so far writing all the asset code 100% in Rust via Pyo3, which is really nice.
I was just trying to portray that it's a 2D game, but all my assets will be in 3D. This is just because i have a decent bit of experience /passion in 3D modeling and animation, but couldn't draw a box to save my life hah.
So i'm going to mock 2D design with 3D models. Starting also with isometric, so it's naturally a bit more visually 3D anyway - but still sprite based.
I'm also going to toy and learn a lot about the boundaries of where tons of 2D sprites become bottle-necked and instead 3D ends up being cheaper. As it stands i know nothing, but some of the things i've planned have quite an extensive set of assets - so i suspect there will be some limitations there.
Scripting is amazing but it's available in most pro apps and most pro 3d modeling apps. Max has had MaxScript since it shipped in like 1995. Maya is built on scripts, it's entire UI is 1500 or so scripts around its lower-level engine. Every thing you do in the UI emits the script command that you could have typed to do the operation. Even Photoshop has been scriptable via JavaScript since CS1 (~2003). And of course all the Microsoft Office products have had VBA since the 90s.
My only point is that scripting is not an unusual feature of Blender. It's a "if it didn't have it it would be 25 years behind the times" feature.
Haven’t found much written on this at the moment, but it’s not hard to wire up: every blender command can be invoked from the Python environment and you can invoke your Python script in blender from the command line (see Python Options in https://docs.blender.org/manual/en/latest/advanced/command_l...)
A friend and I worked on a mobile 2D game in 2011-2013. Our level editor was just a heavily scripted Blender. It is awesome how much you can change the UI through scripts.
As a professional Blenderer, I'm most excited about Mesh to Volume. Making realistic clouds has never been easier! absolutely love the software and can't imagine where i'd be without it.
I've just started to get back into working with 3D graphics after a 20 year hiatus. It's amazing to see the quality and range of toolkits available in Blender.
I haven't done 3d modeling in about the same amount of time and would love to get back into it. Were there any particular resources you found helpful for getting up to speed?
Ditto. Andrew (Blender Guru) is a brilliant teacher with a good Australian sense of humour that kept the series fun. Amazing that it's free, though he understandably plugs his textures website so it doubles as a good advert (and I ended up paying for some resources from it, so I suppose it worked!).
It's good to see all the FUD about Blender slowly disappearing, it wasn't that long ago when half the comments on anything about Blender were something along the lines of "you get what you pay for".
This seems to be becoming a serious tool for professional use, right? Are people switching over from Maya, Rhino and others? Are any movie studios using Blender as their main tool now?
I imagine that Blender is becoming more and more of a tool for students who now don't have to pay anything to have a good 3D modeling tool.
Tangent animation in Canada did Next Gen on Netflix with Blender.
Most of the major film studios have huge pipelines in Maya and Houdini etc... Until recently, Blender wasn't even interested in partaking in a pipeline (Tons' own words at SIGGRAPH 2017)
Recently that has shifted and blender is looking to be able to be integrated into pipelines Which might help adoption. Such things as supporting Pixar USD and the VFX reference platform.
In addition to that, Blender did an overhaul of their UI and renderer that brought a lot of interest in.
I expect to see Blender in studios sooner rather than later, but I don't expect it to supplant the existing DCC packages.
There's still a lot of friction to working with other DCCs, still a lot of performance issues with larger scenes, and then there's issues with support, extensibility and licensing.
Corporations are averse to GPL and Blender is all GPL. They might get more adoption if they dual licensed their plugin API.
Also since it has no C api, it's not attractive from an extensibility perspective.
Finally, the major studios like to have support contracts to expedite fixes to blocking issues. Blender doesn't have such a thing. Yes , it's open source but it requires crewing up for something that a support contract would cover .
Thanks for laying that out. It's very interesting. Maybe it's on the Blender roadmap to become attractive to movie studios, but I haven't checked.
A question on pipelines: are those different from what's mentioned on Blender's Pipeline page? [1]
> Finally, the major studios like to have support contracts to expedite fixes to blocking issues.
It seems like blender has something along those lines provided by the Blender Institute [2]
>> Co-Development with Blender Institute
>> If your company or studio has specific development targets or needs, you can contact the Blender Institute to discuss options.
>> We will hire or involve the developers for topics that align well with the Blender roadmap and benefit the contractor well. Such agreements can have milestones and deliverables. Co-development will especially work well when aligned to other projects that run in Blender Institute. Only projects that involve multiple months of work will be taken in consideration.
For pipelines, studio pipelines extend to far more than just the file format as I/O. It's things like integration of asset management tooling, referencing setups and the ability to send data easily between applications and departments.
Blender has alembic support which is great for geometry, but its USD support is still nascent and limited to export. Additionally because it doesn't use Qt, it's hard to integrate tooling that can otherwise be shared between apps. (note there are third party projects to integrate Qt in Blender).
It's also only recently that Blender has started following the VFX reference platform, which previously meant that all your dependencies and Python code needed to be redone to support Blenders versions.
The support for things like OCIO etc is also relatively new.
With regards to the Blender foundation and support contracts, it's quite different than what a traditional support contract offers.
Usually a support contract means you have a dev that responds to your companies needs, where they'll support you with custom builds of the software if needed, including back porting patches to your production locked version.
The Blender foundation instead takes donations towards development of future versions of Blender and it's still very hard to convince them to support specific studio needs. You're most likely NOT going to get backports of fixes to a production locked version of Blender.
It's only recently (2.83) that they've started having LTS releases..
----
To caveat all of this, I'm not saying Blender shouldn't be used for productions. It's very capable. I'm just providing perspective on why it's not finding a strong foothold against incumbents even when it has some very strong on the surface features that Maya doesn't for example at an infinitely better price.
I have confidence that the Blender foundation are moving in the right direction after what seems to be a cultural and priorities change with 2.80 onwards.
Part of the 2.80 changes also brought a lot of outside funding from Epic, Ubisoft, Facebook and more that I believe will help steer Blender well in the future to unseating Maya etc.
I don't see Blender unseating Houdini anytime soon though.
Sure, but you take a massive performance dive marshalling things through Python. Additionally the Python API is only semi stable between releases.
If you want to add performent extensions to Blender, it requires maintaining an entire fork of the app.
Contrast this to other DCCs, where they almost all provide compiled plugin support (usually C++) and API stability is largely guaranteed between releases.
This makes Blender very unfriendly for production pipelines.
Add to that, Blenders Python extensions don't have a very friendly UI story via ghost, whereas other DCCs largely provide a Qt + PySide2 backing.
Fwiw I'm very much pro blender, having rolled it out to artists in production settings, but I am far from recommending it in any core production pipelines unless the studios are willing to maintain an entire fork of the app.
Some newer game companies are using Blender, I think. It's definitely making inroads -- I've seen it mentioned in job postings. But larger companies tend to have decades-old pipelines that tightly integrate Maya/3DS/etc, so they're not going to switch any time soon.
For big budget Hollywood work it's still primarily Maya/internally developed tools. Interestingly I've seen more people working in the industry adopting Blender for side projects.
Lately I came back to Blender after years to design some visuals for Summon The JSON programming decks. It is amazing how easy to use it became over the years. And it does not crash like it used to. I'm impressed by it. Never thought me, a programmer, will be able to generate real 3D stuff
Some would say this amazing diss by Captain Disillusion at BlenderCon had something to do with the major interface changes in Blender 2.8. The entire presentation is hilarious but for the interface related bit I've linked to 21m15s here: https://youtu.be/1qSTcxt2t74?t=1276
I really enjoyed that, thank you. Alas, I wasn't even thinking about the version he used to demonstrate. Based on some cursory searches, I must have been thinking about the 1.x version where everything was at the bottom. Feelin old.
Years ago I looked at, and sorta-learned, Blender for use as a tool for designing simple 3D printed objects. It wasn’t the best tool for that, and I switched to Fusion 360, which is awesome. But F360 isn’t actually free for hobbyists; you have to affirmatively renew your trial every year, which makes me nervous. Has Blender become better for this kind of thing?
Unfortunately no. Other options you might look at for GUI solid modeling are FreeCAD (the one I tried before giving up and learning Fusion) and Solvespace (you'll want to build 3.0 from source, the most recent release build is 2.3 from 2016).
Or for code based modeling, OpenSCAD is the big thing, and cadQuery is a newer python-based system that seems very promising. Version 2 of cadQuery is built on Open CASCADE, the same backend as FreeCAD, and I gather it makes some modeling operations like fillets and chamfers much more possible than what you can do in SCAD.
I've been meaning to dig into that but haven't made the time yet.
To answer your question, I think not significantly.
But I started 3D printing Feb this year and started 3D modelling only for that purpose. I don't see myself investing in anything else.
Right now it works ok for 3D printed models. Not great but if you are aware of it's faults you can work with it.
In future it'd take a single blip of that focused developer firehose to address a whole lot of concerns for us 3D printer types.
But even without a big business throwing money at them for that exact purpose, I am excited by boolean modifiers and the knife intersect tool getting significant updates in this release.
Right now Blender is so much more than a 3D printed object modelling tool. I like having the option to explore other fields in the 3D modelling space too.
> Right now Blender is so much more than a 3D printed object modelling tool. I like having the option to explore other fields in the 3D modelling space too.
True and funny as my primary Blender use is for video editing. I like having the option to explore other fields in the 3D modeling, animation, and VR space too.
If you want to design artwork, Blender is probably the right tool. If you want to design functional items in a similar manner to Fusion360, FreeCAD is probably the best FOSS solution, though it is not as good as Fusion360.
If you are looking for an alternative to Fusion 360, I can recommend OnShape. They also offer a free for makers style account where all your models are public and non-commercial. It is professional CAD and works great, parametric and all. It should be quite similar to Fusion. I use it for 3D printing.
Fusion 360 and Blender are fundamentally different tools. Stickign with Fusion is probably the best course of action if your goal is to 3D print a part.
Neat! I just started learning some Blender this year, and I know just enough to make what I need for the games I'm developing (abstract puzzle and strategy games), but I want to do a deeper dive down the rabbit hole at some point.
What would people recommend to learn Blender? I like tutorials which you can follow along and explain the why of what you're doing (a bit like the Rust book).
Private sector company would have to operate under the benevolent-dictator-for-life model in order to produce something with an underpinning philosophy on par with Blender's. That uncomfortable constraint alone makes it unlikely.
Stated another way, the product of a combined effort by tens of thousands of enthusiasts made over a span almost two decades would be costly to duplicate and would probably reflect a different philosophy in it's design. The company can throw money at a project to buy developer time but it isn't quality time. Look at autodesk, why improve a product when you can just buy the competitor?
And the release page is performant and pretty, and it works without Javascript but it still has all these great little diagrams and demo videos for each feature. Like, you can look at this page and even if you knew nothing at all about 3D modeling you'd still want to go download this thing immediately.
https://www.blender.org/press/unity-joins-the-blender-develo...
And a huge grant from Facebook:
https://www.blender.org/press/facebook-joins-the-blender-dev...
https://fund.blender.org/corporate-memberships/
I suspect it doesn't go back to version 3 though
https://www.chromium.org/developers/bisect-builds-py
https://developer.blender.org/D6807
It's been waiting for review for about 9 months. :(
If this gets reviewed and merged (probably with some hopefully minor fixing), it'll put the foundations in place for Blender to start competing with CAD/CAM packages.
Which would be a really good thing (as a CAD/CAM and otherwise Blender user). :)
Am I looking at the right thing? Could you maybe elaborate a bit on why this will help Blender become competitive in the CAD/CAM space?
We might need to ping the patch author "xueke pei (yuzukyo)" to make sure I have that correct... and make sure I'm pointing at the correct thing. :)
https://developer.blender.org/p/yuzukyo/
---
Blender doesn't really have any presence in the CAD/CAM space because (my understanding) is the geometry Blender uses isn't "solid geometry".
Because it's not solid geometry, Blender isn't able to export to any of the widely used file formats in CAD/CAM. eg STEP, IGES, 3DM, etc.
Nor even to other OSS packages (eg FreeCAD), as they too use solid geometry.
---
Solid models are used in industry (CNC, plasma, waterjet, etc), somewhat similar to how STL files are used in 3D printing.
With 3D printing, the model file (eg STL) is turned into a gcode action plan for printing by a slicer. eg Cura (https://ultimaker.com/software/ultimaker-cura)
In industry, the model file (eg STEP) is turned into a gcode action plan for the machine cut out the part by CAM software. eg MasterCAM (https://www.mastercam.com)
---
With that branch (D6807), my understanding was it enables proper in-depth NURBS (solid geometry) support.
The commit comments in that branch indicate it also allows importing of Rhino (3DM) and subdivision surface (subd) files.
If that's the case, it's the needed foundation piece to allow for creation of importers and exporter for widely accepted CAD/CAM formats.
The second an even basic STEP exporter is functional in Blender, it will start pulling users across from the expensive proprietary CAD/CAM systems. As almost no-one actually wants to use them, it's just a "lack of choice" problem. ;)
Especially when the other parts of Blender (animation, grease pencil) are already very complimentary to modelling and already support the right workflow(s).
You personally might prefer to use Rhino. But the vast majority of new users who come to the CNC places I frequent (eg PrintNC Discord) would easily choose a functional Blender over any non-OSS modelling package.
That being said, I don't have an account on developer.blender.org.
Is someone with an account able to contact them somehow?
CAD is all about building mechanical parts with features that match needed specifications. For example you need a block with holes in it with particular spacing, particular diameter and particular depth. CAD packages like Fusion 360, SolidWorks, FreeCAD etc. are all dedicated to this task and make features like constraints and measurements front and center.
Blender on the other hand is, at least in my experience of it, all about bringing somebody's artistic vision into a 3D environment. It's usually not important to somebody's vision that parts meet dimensional constraints, it's more important that it looks "right". And so Blender is full of tools that allow humans to change things to arbitrary dimensions.
Does it really make sense to try and fit both of these use-cases into one tool? I can't help but think that Blender would come out worse for it.
And that's CAD, I really can't come up with any reason why CAM would make sense in Blender.
Is there something I'm missing?
They're just so many years outdated, that they're pretty much unusable which is probably why you're not aware of them being there. ;)
This patch is to update the NURBS capabilities in Blender, so the "artistic vision" objects people create can then be turned into real life objects via industry standard processes (eg CNC machining).
If you personally don't want to use the NURBS pieces, then feel free to ignore them. But please don't block other people from doing so. :)
I'm not saying NURBS shouldn't be fixed/updated, I'm just debating whether "so that Blender can be a CAD/CAM package" is a reasonable motivation for it.
It has been great so far - and even more surprising i'm so far writing all the asset code 100% in Rust via Pyo3, which is really nice.
So i'm going to mock 2D design with 3D models. Starting also with isometric, so it's naturally a bit more visually 3D anyway - but still sprite based.
I'm also going to toy and learn a lot about the boundaries of where tons of 2D sprites become bottle-necked and instead 3D ends up being cheaper. As it stands i know nothing, but some of the things i've planned have quite an extensive set of assets - so i suspect there will be some limitations there.
My only point is that scripting is not an unusual feature of Blender. It's a "if it didn't have it it would be 25 years behind the times" feature.
Absolute joy to use.
Disclaimer: I have been happy customer since blender 2.7 days and I expect their old courses to be relaunched with new blender support.
Blender 2.8 onwards did a lot to rectify those criticisms, which in turn allowed people to view it and dismiss the FUD
I imagine that Blender is becoming more and more of a tool for students who now don't have to pay anything to have a good 3D modeling tool.
Most of the major film studios have huge pipelines in Maya and Houdini etc... Until recently, Blender wasn't even interested in partaking in a pipeline (Tons' own words at SIGGRAPH 2017)
Recently that has shifted and blender is looking to be able to be integrated into pipelines Which might help adoption. Such things as supporting Pixar USD and the VFX reference platform.
In addition to that, Blender did an overhaul of their UI and renderer that brought a lot of interest in.
I expect to see Blender in studios sooner rather than later, but I don't expect it to supplant the existing DCC packages.
There's still a lot of friction to working with other DCCs, still a lot of performance issues with larger scenes, and then there's issues with support, extensibility and licensing.
Corporations are averse to GPL and Blender is all GPL. They might get more adoption if they dual licensed their plugin API.
Also since it has no C api, it's not attractive from an extensibility perspective.
Finally, the major studios like to have support contracts to expedite fixes to blocking issues. Blender doesn't have such a thing. Yes , it's open source but it requires crewing up for something that a support contract would cover .
A question on pipelines: are those different from what's mentioned on Blender's Pipeline page? [1]
> Finally, the major studios like to have support contracts to expedite fixes to blocking issues.
It seems like blender has something along those lines provided by the Blender Institute [2]
>> Co-Development with Blender Institute
>> If your company or studio has specific development targets or needs, you can contact the Blender Institute to discuss options.
>> We will hire or involve the developers for topics that align well with the Blender roadmap and benefit the contractor well. Such agreements can have milestones and deliverables. Co-development will especially work well when aligned to other projects that run in Blender Institute. Only projects that involve multiple months of work will be taken in consideration.
[1]: https://www.blender.org/features/pipeline/
[2]: https://www.blender.org/foundation/
Blender has alembic support which is great for geometry, but its USD support is still nascent and limited to export. Additionally because it doesn't use Qt, it's hard to integrate tooling that can otherwise be shared between apps. (note there are third party projects to integrate Qt in Blender). It's also only recently that Blender has started following the VFX reference platform, which previously meant that all your dependencies and Python code needed to be redone to support Blenders versions.
The support for things like OCIO etc is also relatively new.
With regards to the Blender foundation and support contracts, it's quite different than what a traditional support contract offers.
Usually a support contract means you have a dev that responds to your companies needs, where they'll support you with custom builds of the software if needed, including back porting patches to your production locked version.
The Blender foundation instead takes donations towards development of future versions of Blender and it's still very hard to convince them to support specific studio needs. You're most likely NOT going to get backports of fixes to a production locked version of Blender.
It's only recently (2.83) that they've started having LTS releases..
----
To caveat all of this, I'm not saying Blender shouldn't be used for productions. It's very capable. I'm just providing perspective on why it's not finding a strong foothold against incumbents even when it has some very strong on the surface features that Maya doesn't for example at an infinitely better price.
I have confidence that the Blender foundation are moving in the right direction after what seems to be a cultural and priorities change with 2.80 onwards.
Part of the 2.80 changes also brought a lot of outside funding from Epic, Ubisoft, Facebook and more that I believe will help steer Blender well in the future to unseating Maya etc.
I don't see Blender unseating Houdini anytime soon though.
Blender is very extensible via its Python API
If you want to add performent extensions to Blender, it requires maintaining an entire fork of the app.
Contrast this to other DCCs, where they almost all provide compiled plugin support (usually C++) and API stability is largely guaranteed between releases.
This makes Blender very unfriendly for production pipelines.
Add to that, Blenders Python extensions don't have a very friendly UI story via ghost, whereas other DCCs largely provide a Qt + PySide2 backing.
Fwiw I'm very much pro blender, having rolled it out to artists in production settings, but I am far from recommending it in any core production pipelines unless the studios are willing to maintain an entire fork of the app.
https://www.blender.org/user-stories/japanese-anime-studio-k...
Or for code based modeling, OpenSCAD is the big thing, and cadQuery is a newer python-based system that seems very promising. Version 2 of cadQuery is built on Open CASCADE, the same backend as FreeCAD, and I gather it makes some modeling operations like fillets and chamfers much more possible than what you can do in SCAD.
I've been meaning to dig into that but haven't made the time yet.
But I started 3D printing Feb this year and started 3D modelling only for that purpose. I don't see myself investing in anything else.
Right now it works ok for 3D printed models. Not great but if you are aware of it's faults you can work with it.
In future it'd take a single blip of that focused developer firehose to address a whole lot of concerns for us 3D printer types.
But even without a big business throwing money at them for that exact purpose, I am excited by boolean modifiers and the knife intersect tool getting significant updates in this release.
Right now Blender is so much more than a 3D printed object modelling tool. I like having the option to explore other fields in the 3D modelling space too.
True and funny as my primary Blender use is for video editing. I like having the option to explore other fields in the 3D modeling, animation, and VR space too.
I can recommend solvespace - it's quite basic in terms of features, but it's really well designed software, so it's nice for designing simple stuff.