Unreal Engine 4 A.R.T plugin for Blender

Hi. Recently, Epic Games announced Animation and Rigging Toolkit plugin for Maya. I have contacted them to ask if a Blender version of their plugin is possible. They said they don’t have a timeframe, but with community access to their C++ source code, this might happen sooner than they think it will.

      So...is someone interested to develop an A.R.T. version for Blender?

  Here is the Maya version:



Having someone to write an add-on for Blender for anything, not to mention advanced animation tools/exporter, has been a royal pain. Even finding paid programmer has been a real issue. I have my reservations about A.R.T. add-on for Blender ever materializing.

What stops you from just making an animated model and exporting it as FBX?

What is the particular purpose for A.R.T. module? (I saw vids, it looks like parametric character design, but I doubt old school method of making characters is not going to work for UE4)

I don’t think the UDK4 C++ code has much at all to do with A.R.T. in Blender. From what I can tell, the A.R.T. is basically Rigify on steroids, protein supplements, and an exhausting daily work-out routine. It’ll export animations to UDK4 in pretty much the same fashion as other rigs in Maya, but it shines in creating the rig, tweaking it’s pre + post binding parameters inside Maya, and simplifying the ability to load, save, & mirror poses for faster animation.

It’s indeed pretty awesome… but it’s awesomeness is not related to UDK4 as such.

So…could a similar plugin be done for Blender?

The question is - why?

I think I would personally prefer something like 3DS MAX C.A.T., but it’s a lot of work to implement and not feasible without dedicated artistically inclined talented programmer.

Short answer though - everything is possible given resources and time.

I have searched the internet and couldn’t find a place where to download that Maya plugin. Probably, it is bundled with the editor. Usually, Maya plugins are written either in MEL or in Python. (not C++) If that plugin is written in Python, we might take a look. Probably some people would find their inspiration in the code for the port.

It’s not in the source code zip for the main engine. Could be in the DMG (which I don’t have on this machine) or zip files containing all the added libraries and stuff. I’ll look later.

FWIW, when Cessen (Nathan Vegdahl) was working on Rigify, I was able to pay for it being finished quicker (and to get minor features added). I’m not sure if he’s still interested in doing large changes to Rigify but, if anyone was to be the developer to ask, it would be him.

Be prepared to require a decent amount of money though - development is not cheap & an A.R.T. equivalent in Blender is not a small task by any stretch of the imagination. Trust me, I’m one of those guys that develops code for a living, am now actually quite familiar with the Rigify code, and it’s not a couple of days (or even just a couple of weeks) to get what the A.R.T. video demonstrates!

1 Like

Why do a thing similar to ART (besides Epic allowing to clone it) and not CAT ?

In this case, a funding campaign could be the way to go. What do you think? Could a funding campaign bring enough money to pay this work? In the case of C.A.T…I don’t know. How is C.A.T. compared to A.R.T.?

Yes, I think a funding campaign could bring in enough money. Especially if attention is paid to the export to game engine features (FBX being critical from the looks of things). A crowd-funding campaign that focused only on making tools for creating, animating, wiring up to mocap, and exporting rigs would be an awesome idea. There is a focused target, one can construct the project so that even if not every aim is accomplished, the work to date makes Blender better than it was before, and it has immediate use in game asset creation AND the open film stuff the Blender Foundation prefers to focus on.

However, you would first have to find a developer (or more likely, developers) willing to code it up & with the experience to do a good job. As I mentioned, I reckon Cessen is the man to go to being the author of Rigify… but I have no idea of his availability or interest.

1 Like

ART is too specific to UE4 and I don’t think it’s necessary.

CAT stands for Character Animation Toolkit and would be useful to any animator regardless whether animation is made for a game engine or cinematic.

Here are the videos showing character rigging and animation using CAT for UDK:

Just to highlight CAT’s features:

  • Parametric motion cycles (footsteps and in-spot)
  • Non-destructive motion layers (and no, Blender’s NLA is nothing like that)
  • Parametric rig generation (N-legged, N-armed, winged, bipedal, etc.)
  • Full body IK/FK with snapping
  • MoCap tools
  • Save/Load presets
  • Interoperability with standard bone, HumanIK, etc. (for Blender it would be using and mixing standard bones and generated rig)
  • Muscles system

It’s artist oriented, meaning you just build your rig with a few clicks, adjust it to you character, and animate. No mess on the screen and in scene with bones and custom shapes. Quick and intuitive process.

Just to clear up some misconceptions, The maya animation and rigging code for ue4 is COMPLETELY python.

1 Like

I have oppened, along with this thread, another thread on Unreal Engine forums and someone in Epic Games staff just replied where the ART code is downloaded. Either ART or CAT, I think that this kind of animation tools would be a great improvement for Blender.

1 Like

I’m definitely interested, but unfortunately I just don’t have time anymore. Family + working full time at Nintendo. I do still try to maintain Rigify, though, so that it doesn’t suffer from bitrot (I have a todo item on that, actually…).

If I suddenly find myself with more time, though, I’ll certainly let people here know. But I also have other things I’d like to do with that time (like e.g. Humane Rigging 2, making a personal animated short, etc.). So we’ll see.

Finally, I don’t know if modifying Rigify is necessarily the best approach, as it has some pretty fundamental design decisions that don’t necessarily make it ideal for game work. And aside from just adding a whole lot more rig types and maybe improving the UI a bit, I’ve pretty much taken Rigify as far as I care to. I’d almost rather take stock of the lessons I’ve learned so far, and start on a new Rigify 2 or some such thing, with a better more flexible design that better accounts for use-cases like games.

But then I also consider that hopefully in the not too distant future Blender’s rigging features will get better (depsgraph refactor, proxy system overhaul, node-based modifiers…) and I’d almost rather wait until at least one or two of those things happen, so that Rigify 2 can actually be more than just incrementally better.

Finally, regarding A.R.T.:
It definitely looks like a really cool tool, and is clearly more powerful than Rigify in many ways. But I think part of that power comes from it being a bit more specific in its application. In particular, it looks to be specific to 1-head/1-torso/2-arm/2-leg setups. I could be wrong about that, but nothing they demonstrated in the videos I watched indicated that you could actually build fundamentally different skeleton layouts like you can with Rigify. That doesn’t make Rigify better, mind you, but it’s just aimed at a slightly different purpose.

Ohhh, Humane Rigging 2 sounds awesome! (especially if it would have some solutions for foot steps/non-slipping feet, instant IK/FK snap setup, N-legged creatures rigging, animation techniques, facial rigging using bones only, etc.)

Any rough approximate date for that tutorial project? :wink:

I have a suggestion that could speed up the development a bit: Every week, on #blendercoders irc channel, there is a meeting. Some of us could participate and express the importance of a new, better rigging and animation system, that could ease the rig creation and animation for game characters and more. In this way, we could attract more Blender devs in this development dirrection. You could also make use of what we discussed here.
I appreciate that if the time permits, you would improve Blender rigging and I wish you luck with your projects. Humane Rigging 2 would certainly become one of the best tutorials packs regarding Blender.

If this thing progress well, please say. I would be up to pay a fixed ammount of money per month (say 10 bucks), to support this development.

1 Like

Let me tell you couple of things about CAT. It’s awful system, trust me.

  1. CATMotion is very hard to setup on custom rigs, it’s struggling and overbending joints.
  2. Motion layers using is VERY rare. Basically, they are used when anim-lead have found mistakes in your animation, and when you need to add other than CAT-default constraints to your rig, they MUST be in some not-root layer.
  3. It’s armature roll hell. Bones are rolling randomly when you adjusting rig by moving them.
  4. IK/FK snap still works bad, it’s not snapping palm orientation. System of 3 bones (or 4 joints) with 2 pole vectors is very hard to set up, you need 2 additional anim layers.
  5. Don’t know about MoCap.
  6. Presets, poses and pose mirroring is good, though, and widget controls creation is easy.
  7. Don’t know about HumanIK.
  8. Muscles system contains 2 things: muscle strands and custom meta-mesh plane, which is only one way to make mesh collisions with muscles. So, you need 1 plane for each muscle, and you need to skin-wrap your char mesh to those meta-meshes after pure skinning. Performance is poor.

Extra things: no root control for entire rig, very bad scene referencing, premanent crashing while skinning (on Undo command, for example), overall bad Max skinning (no volume brush for layered objects like char in clothes, no normal weights projection, weight mirroring does not symmetrize middle-sew vertices on mesh).

CAT is not as good as Autodesk wants us to think about it. In fact, I still using CS Biped to animate characters.

As a postscriptum I’ll say “Animating in 3ds max is very bad idea”.

Always great to hear what you’re up to Nathan.

As for this ART thing, assuming that it was actually created for Blender, would it allow us to use the created rigs for other game engines, or would it be specifically for UE? If it’s the latter, I for one, don’t see myself contributing to it’s development financially. I’ve contributed to various projects in the past, and I’ll continue to do so, but I can’t put out the money for something that’s only going to be used for the UE since I won’t be using that engine any time soon.

I’m pretty ignorant about this, obviously, but if it’s something that could be used for Unity, CryEngine, and BGE, etc., then I would seriously considered supporting it. Just something to think about.