After two Houdini releases, another short film called Freelance and many other things in between, SideFX is happy to introduce both the completion of the animated short film Turbulence, as well as an interview with the key people behind it -- Magnus Møller, Christopher Rutledge and Jesper Andkjær. While this project was fully made by Tumblehead Animation Studio and Christopher Rutledge, it was done in collaboration with SideFX, with support from the R&D team. Production of this short, spanning across two versions of Houdini, is not what you generally want to do, but the desire and motivation to utilize cutting-edge new tools in Houdini was equal in part to the dream of making an animated short film together with friends.
SIDEFX: So let's start from the beginning. Give us a small summary of who you each are please.
MAGNUS: I’m Magnus, I’m co-founder of Tumblehead, an animation studio based in Viborg, Denmark. We love to make fun and cartoony animation in 3D, often inspired by UPA and the Cartoon Modern era. We have always been drawn to the mix of highly creative workflows and the latest and greatest technology. It’s maybe not a big surprise that we are all very excited about Houdini, as it opens up so many new possibilities for creative workflows.
CHRIS: Hello! I am Chris! I’ve been animating since I was a kid and finally got into 3D, back in 2015. A year or two later, I got into Houdini and had a ton of fun learning it, discovering Magnus and Tumblehead’s work in the process. I always loved and mainly focused on character work and making short films, which felt like a less-common use case for Houdini, but when Magnus and I started becoming friends and nerding out about our love for SideFX, we thought it would be awesome to try to make a film together using Houdini.
JESPER: Hi, I'm Jesper and I have been part of Tumblehead as an animator for about 4 years now. When I was getting into animation and during my studies, my main focus was 2D cartoony character animation, and I was always a bit hesitant when it came to 3D animation, as it often feels more confined and limited. So the way Tumblehead combines 3D with the silliness and flexibility of 2D worked perfectly for me and has allowed me to transition into a CG generalist, exploring cool tools and software like Houdini.
SIDEFX: How did this idea even crystallize in the first place and who has the fear of flying?
CHRIS: When I was a lot younger, I had the feeling when turbulence hit, that it felt like I was going to die and it was scary, but also weirdly exciting. If the plane starts going down, there’s nothing you can really do about it, so you might as well make the most out of your final moments! I’d been wanting to explore this topic further and especially when flying around to different animation festivals, like Annecy where I first met Magnus, it got the gears turning again.
I think specifically the feeling that when you hit turbulence, you FEEL like you are going to die, but you look around and everyone else is acting completely normal, which in turn can make you feel either calm, or even more insane. I talked to Magnus about it and we thought that a film themed around turbulence would be a lot of fun and have a lot of potential for building fun effects and procedural setups, character acting, rigs, and animation.
I don’t know that either of us has a true fear of flying, but every time the plane hits turbulence, or you are taking off or landing, I always get a little anxiously excited.
SIDEFX: Who made the story, the design of the characters, the overall style?
CHRIS: The character designs are a bit of a mix of the classic Tumblehead style and some of my silly sketches in the animatic. We talked about the idea a lot and wrote some scripts / notes, and then our storyboard artist, Benjamin Agersted, did a pass of the film as boards. I took the boards and made an animation using them with some of my own crappier drawings mixed in. I added sound FX/voices, and then Jesper modeled most of the characters himself using Nomad Sculpt on his iPad, based off the animatic.
The goal in my mind was to lean into the amazing “Tumblehead style” – but push it slightly towards the more grossly detailed “Chris Rutledge” style. I think we hit a nice medium, and in the end it was really a big group effort combining the talent, styles, and ideas of all of us :)
SIDEFX: Why did you want to use Houdini, for all aspects of the film - particularly, rigging and animation? Houdini is not the first thing that jumps to mind when one thinks about doing character animation... What was the initial reaction when Jesper was told that he's going to use Houdini instead of Maya?
MAGNUS: I’ve always dreamt of being able to rig in Houdini. When KineFX first launched, it was a really exciting time. You could make some really advanced set-ups fairly quickly, only using nodes and a bit of vex. This was a big change of pace from what I was used to. However, at the time, it wasn’t quite set up for proper shot workflow with multiple characters etc. So when APEX came along to expand the capabilities of KineFX, we immediately got hooked, and of course the only reasonable thought was to make a short film using APEX and KineFX.
JESPER: Having only animated with Maya, and with almost no prior experience with nodes, I was slightly worried at first about having to learn an entirely new way of working. It can be a steep learning curve at first, but at the end of the day, animation is animation; with the same technical rules, so once you get used to looking at the nodes and the layout and the shortcuts, the rest is not so alien. And with the way the shot is set up in Houdini makes it much quicker to add effects or edit geometries as you go, which makes for a more creative workflow.
CHRIS: I’ll just add that – part of the fun of doing character stuff in Houdini is just that people don’t tend to think of it for that. Having access to the rigs and doing all kinds of secondary effects/simulations, all within the same program is really awesome and has helped me find my own techniques and styles. There’s tons of potential to think outside the box with Houdini. I love seeing the stuff that people like Saša Budimir or Rich Lord come up with, in creating crazy procedural character rigs and stuff, and I hope we can inspire more people to do just that.
SIDEFX: What is your opinion of animating in Houdini, today?
JESPER: With the animation tools and interface, it feels pretty natural to animate in Houdini. The integrated selection sets and inbetweening tools feel intuitive and well-known. There are a bunch of features I haven’t had time to thoroughly explore yet, like the collision sim on rigs or editable motionpaths, that I think are really going to speed up animation. So I’m excited to keep learning new ways to improve my workflow, and look forward to the tools becoming even more integrated and accessible to animators without a lot of prior Houdini-experience.
SIDEFX: Tumblehead Animation Studio has been using Houdini for some time, since “Tales from the Multiverse” (2018). Then, the leap to using Solaris in “The Swineherd” (2021). And then using KarmaXPU for the first time when making “Freelance” (2023). Tell us about this progression - what was the first task that Houdini got used for and why?
MAGNUS: Previously we were using Maya and Redshift. We found that we were missing features to make it easier to structure the flow of shots, assets and general pipeline. We had done a lot of custom Python tools in Maya, but it always felt like you were doing something you were not supposed to do. When we tried Houdini, we immediately felt at home since it was clearly built with a node-based approach in mind and had a very solid Python API. Then Solaris and Karma came along and we have never looked back!
SIDEFX: In most productions, you stay with one version of software for ... sanity, stability and being able to support any custom tools made for a show. Why didn't you play safe and stick with one version of Houdini for this shortfilm?
MAGNUS:
Our goal with the short was to utilize the new features of Houdini 20. However, many of the features we wanted to use were pushed to Houdini 20.5, and then also came COPs, there was no way we could resist upgrading. It’s of course not ideal to change versions of a software during production, and especially going from a release build to an alpha build. We paid with blood, sweat and tears…(maybe not blood though), but in the end it was a 100% worth it. We learned so much from this process, and I’m very happy we took the risk.

COP Network used to create sweat
SIDEFX: What build did you start making Turbulence on?
MAGNUS:
It was in the beginning of the Houdini 20 Alpha. APEX looked very different than it does now. There were no APEX scripts either. The only interface was a very limited, how-to-put-it-nicely “programmer art”-type node view that felt quite alien to all the other Houdini UI’s. It was a fun time, and although it was a bit hard to get started, it was really interesting to get a glimpse into how it looks like for the RnD team when they are developing new frameworks. Like looking into the Matrix. I’m really thankful to the APEX team who took the time to explain everything to us.
SIDEFX: In terms of assets, please give us a breakdown of the work you did.
MAGNUS: In total, there were 10 characters, 6 props and 2 sets - the interior and exterior of the plane, with surrounding environment.
SIDEFX: Which areas of support did you get from SideFX, while working on this short film?
MAGNUS: We got so much help from the RnD and support team, especially the rigging and animation team, and the Solaris/Karma team. Huge thanks to them for the patience and for answering our questions!
We learned a lot, but I also think in the end it was valuable for the developers to get a lot of feedback testing early. Not just small isolated tests, but a very broad test to see how everything would fit into a full pipeline. Many of the ideas that spawned from our sessions with the devs ended up as features in 20.5, and some scheduled for a future release. Many of them would maybe have been added anyways, but I like to believe that we contributed at least a little bit in shaping the roadmap ahead, especially for rigging and animation, and how that would work together with Solaris and USD.
SIDEFX: Did you *actually* composite in COPs?
MAGNUS: Yes, 100%! All the shots went through COPS. It worked really well for what we needed. We also made a pipeline that would set up your shot automatically, based on render layers and AOV’s, based on Houdini’s new recipe system. It worked really well, and automated a lot of tasks for us. It was also really fun to play around with the NPR techniques, like adding outlines based on depth, 2D distortion based on UV AOV, etc. You can see all of these FX on the shots, where Jimmy (the main character) is freaking out. That’s all made with COPs. One major drawback with the current version of COPs that I want to mention, is that it is not optimized for *.exr files yet. I know this is a big priority for the next release, so just be aware of that limitation before you try to do something similar.
Character
RIGGING & ANIMATION
SIDEFX: One of the most interesting aspects of this short film is the fact that it was rigged and animated in Houdini, and by a relatively small team, and it was done using this newly developed framework, with APEX graphs. How would you summarize your experience?
MAGNUS: As mentioned above, we started rigging already in the Houdini 20 alpha. And at that time, many of the HDA’s and nodes that are available in APEX now, where not yet developed. So therefore, we developed our own custom python based autorigger that would create the APEX rig network. Had we started the project today, we would have used the Auto-rigger components and APEX script.
I know our API inspired some of the features in APEX script (which is very much like Python) and so that made everything worth it. Our plan for the future is to rewrite our APEX python wrapper with APEX Script instead.
With that said, when we had our custom set-up up and running, rigging was a breeze. There were some limitations of course, but setting up a rig and also modifying it was very very easy. Something that would have taken an hour in Maya to fix, would take me minutes or even seconds in Houdini.
I’m very excited about where rigging and animation in Houdini is heading. I think we will see a wide adaption by the industry within a relatively short time. It’s that good.
SIDEFX: If you had to do this animation once over, knowing what you know now, what critical things could/would have helped save some of your troubles?
MAGNUS: We quite quickly figured out that we needed to develop a custom pipeline for the film. We needed a setup that allowed us to split assets and shots into departments, so we could work on different aspects of a character or a shot simultaneously. What was meant as a simple and limited tool grew quickly into a full Houdini-based pipeline framework. Our pipeline developer Søren Nørbek did an outstanding job putting this together! It took about 6 months, and resulted in about 17 000 lines of very neat and efficient Python code.
One of our design pillars was that we wanted to make USD the root of the pipeline. Therefore, all workfiles have /stage (Solaris) as the starting point. The whole point of this project was to use Houdini for all aspects of production, and so it made sense to us to also simplify the use of different contexts in Houdini as well. We wanted to take a stab at the Houdini workflow of the future, where /obj context is entirely replaced with Solaris. I know that might spark some emotions in the Houdini fan base, but I think that’s where it makes sense that we are heading (to). Mixing OBJ and Solaris makes things unnecessarily complicated in my humble opinion. There are still features we are missing for this new workflow to be 100% practical for production, like animation interoperability with Solaris, but I know the devs are actively working to improve that.
CHRIS: In terms of tools, additionally we used a handful of ones I have hackily created for my own use when creating character-based work as a one man band. Specifically my “jiggler” and my “rig noise” – which we have demonstrated in some of the talks on turbulence, but I have found the jiggler to be such a great way to be able to procedurally add and adjust stuff like secondary motion and overlap, just by modifying point positions of bones (again, very hacky but, works wonders). “Rig noise” just adds some basic rotational noise to the skeleton, which helps make characters “stay alive” – giving some cheap idling wherever needed. Could also be used to add shakes and other stuff. Being able to layer in this kind of thing with KineFX procedurally, I think is where Houdini can prove to be really unique for character animation in the future. Not everything has to be keyframed, and a combination of a “destructive” layer of work, like keyframing, along with procedural edits afterwards, allows artists to work a lot more flexibly and quickly.
MAGNUS: You can get a peek into our current pipeline here, where we document stuff for ourselves, but also if people are curious outside about how we roll.
SIDEFX: What were some surprising pleasant aspects of working in Houdini for character animation?
JESPER: The automatic jiggle / settle / overlap / overshoot saves a lot of time, and depending on the style, you can go from keys to done, in no time. The transient constraint system is great. It makes constraints simple and easy to edit. And the way geometry can be edited and altered after the animation node, gives new flexibility for the animator to do things that are beyond the limits of the rig.
SIDEFX: As we are still in beta for APEX, what aspects would you prioritize and identify as critical for animators, that we still currently lack?
JESPER: I think the animation interface (viewport HUD) can be improved so that floating windows are not covering up the screen too much. Personally, I would prefer to have the UI as panels, so the viewport is less cluttered. I’m sure the developers are working on it, but there have been a few bugs with those floating HUD windows getting out of sync with the Scene Animate State, and sometimes they freeze and you can’t remove them, etc.
An absolute must-have, is a way to have real-time playback in the viewport while the stage is visible/ghosted. With the stage turned off, you don't have the sets or the camera, so you have to choose between shot accuracy and playback speed. (Note, this issue is not present if you use an OBJ context-based workflow, but for Turbulence, we made the whole pipeline around Solaris).
Another important thing (that the devs are working on improving as we speak) is to improve the audio playback when scrubbing the timeline. At the time of animating on Turbulence, this did not work properly, because the audio would always be a bit out of sync.
MAGNUS: We are very much looking forward to features like a built-in Pose Library and Picker. This will be a huge addition to the already growing animation toolset. On the rigging side, I personally look forward to seeing more spline/ribbon-like setups for APEX. I would also love to get some sort of way to fetch data from previous frames without having to invoke the whole rig graph. That would allow us to create CHOPs-like simulations directly in Scene Animate.
CHRIS: I’m coming from more of a place where I’ve been a one-man-band doing most of my character animation myself using KineFX, as opposed to more traditional (and less procedural) animation systems. I’ve found some really fun hybrid keyframe + procedural workflows where basically I can just do some basic keyframing and then let stuff like jiggle / secondary action get layered on procedurally. I think with APEX, this is the main thing I’d like to see leveraged. I’d like to have these tools be built in, and then with animation layers it should be possible to do a rough first pass, and then a procedural jiggle pass that is tweakable, and then a final polish pass. I think there’s a lot of potential for Houdini here to become a really unique tool for character animation that is also friendly to traditional character animators.
SIDEFX: We have worked together for at least two years and have received all the feedback firsthand from you. Now, getting it all laid out like this might seem like the opposite thing any company should do, and then, to publish it as such. But looking at it from another way, we are continuously developing a tool that people will (hopefully) enjoy using to execute their idea and vision. Specific to character animation, it was not until recently that this area was getting full attention from development. So while it is obvious to say it, the issues you experienced are of course not desirable but, knowing what needs to be fixed or made a-new, is exactly what helps to improve Houdini for not only you guys but for everyone, no matter the task.
CHRIS: I really appreciate SideFX’s development philosophies and open-ness about the projects they are working on, and love being a part of the alpha/beta testing programs. You can really tell SideFX cares a lot about making this software as good as it can be, rather than just releasing flashy features once a year and not touching them again to appeal to shareholders and have flashy marketing. Especially the fact that SideFX is able to have these longer-term projects, like KineFX, APEX, Copernicus, Solaris, Vellum, Karma, etc. Some of them have had rough rollouts and haven’t been easy for the community to make use of initially, but over time, SideFX keeps working on them and putting in the love needed to make them really awesome. I have faith that they will keep committed to a big new exciting feature-set when they announce it, and it will continue to improve year after year.
MAGNUS: What Chris said. I couldn’t agree more. This is the reason we are so passionate about Houdini and SideFX in the first place. Please never change!
SIDEFX: It's been a long journey, and we did mainly discussed rigging and animation. Was there anything else that you would like to say to the community?
CHRIS: I’d love to see more people use Houdini and push it in different ways and it’s been wonderful to watch SideFX continue to expand Houdini beyond being mostly an FX tool. There’s a lot more work to do on the development side, but it’s also super important for the users to keep pushing it and using it for new and unique things, because it’s such a remarkably flexible tool with so much unexplored potential still. Also, it’s always worth getting involved in the community, giving feedback, and getting in touch with the developers however you can, to make your voice heard. Feedback really does make a big difference!
SIDEFX: Asides from what you said just now, any other areas or perhaps specific tool that you wished Houdini could be better at?
CHRIS: I’d really like to see Houdini become better for “realtime” stuff. Being able to puppet and record animation while playing the timeline for example. I’d like to see the software start up faster, a small thing that would make a big difference. I’d love to see a realtime renderer / cinematic viewport (which I believe you are working up to with Vulkan), and generally more and more artist friendly tools. I’ve heard mentions of shot sequencing tools, which would really help someone like me who is using Houdini to make more than just shots, but entire films.
MAGNUS:
I have a lot of the same wishes as Chris. I would love to see a broader Houdini adoption from the animation community. For that to happen, I believe Houdini needs to become more approachable for artists. That includes simpler and more elegant/streamlined UI and UX, more artist friendly tools, and better viewport (artists loves a good looking viewport!).
SIDEFX: How many people were involved in the making of this?
MAGNUS: Not counting all the great people from SideFX that helped us out, we were in total 19 people involved, from initial script, to final render.

SIDEFX: And how did this work, with Chris being based in LA and you guys in Viborg, different timezones, working on commercial gigs and this being a passion-project?
CHRIS: One cool thing is that since we synced up on the awesome pipeline tools Søren built for Tumblehead and got really familiar with each others’ workflows, we were able to jump onto commercial projects together much more easily. If we hadn’t built that pipeline and learned the things we learned from Turbulence, it probably would have been a lot harder to turn around this ad for Nike that we made in less than a month.
One of the great things about our setup for this project also was that – because it was such a quick turnaround, and I was in LA and they were in Denmark but we had access to all the same files and were used to the pipeline, they were able to work while we were asleep, and then when I was in meetings getting notes and stuff, I was able to make adjustments and tweak things, kick off renders, etc, while they were asleep. So we were basically working around the clock.
Generally for Turbulence, the time difference was a pain, but it was fun that for a month between festivals I was attending in Europe (Pictoplasma and Annecy), I went up to Viborg and got to work with the Tumblehead team in person.
SIDEFX: Tell us about the hardware used in the making of this animation?
MAGNUS: Both us and Chris use mainly 4090 RTX cards for rendering. For workstations, we also use everything from 3070 and up. Else than that, we run Ryzen CPUs and usually 64GB of ram.
SIDEFX: Everything was rendered in KarmaXPU only?
MAGNUS: Yep! Every shot was rendered with Karma XPU and comped with COPs. It was a joy to render with Karma, though we are hoping for some of the modern temporal denoising tech to be implemented so we can get renders out faster.
SIDEFX: Did we miss any easter eggs???
MAGNUS: It’s hard to miss, but our favourite logo turned into a great textile pattern on the seats.
SIDEFX: Thank you guys and, congratulations!