Siggraph Asia 2024

Welcome to our coverage of SIGGRAPH Asia 2024, where the brightest minds in computer graphics gather to share their latest discoveries and innovations. The must-attend event brings together experts in animation and technology to discuss ideas, latest trends and innovations. In this article, we’ll explore the key takeaways from the conference, featuring interviews with six key players who discuss topics from computer vision to AI to the state of CG in Japan and worldwide. We’ll also delve into the beating heart of the Tokyo industry scene, where the conference’s after-hours events and networking opportunities come alive.

Trends and Innovations in Computer Graphics

Prof Takeo Igarashi Conference Chair	Dylan Sisson Featured Exhibitor	Ken Anjyo Featured Sessions Chair
Prof Ariel Shamir Technical Papers Chair	Prof Ruizhen Hu Technical Communications Chair	Prof Yuki Ban Emerging Technologies Chair

What are your thoughts on this year’s conference, and what are you most excited about?

Takeo Igarashi: As everybody knows, generative AI changes everything, and is rapidly moving. At SIGGRAPH Asia 24, I see lots of interesting new results. That’s an important thing for me, in terms of content.

Ken Anjyo: From my view, the charming point of this conference is collaboration. There is a new stream of Japanese content such as Shogun or Ultraman Rising. Ultraman’s concept, for example, is from Japan, but Ultraman Rising is quite different from the original. It’s a new interpretation. And Godzilla, on the other hand, was made by only one Japanese company, Shirogumi. I invited the CG Director to give a featured talk. Shirogumi is a very well-established company that started in models and miniatures-making.

Dylan Sisson: My first SIGGRAPH was 1997. It’s not as big as it used to be, but it’s interesting to come to SIGGRAPH because you read the temperature of where the entire industry is going as a whole. For this year, a lot of what we do is focused on what we’re developing for RenderMan, talking with vendors, partners and collaborators, and figuring out where the right direction is to go for the next five years. It’s hard to figure out what anything’s going to be like in five years right now, but if we can figure it out, well, this seems like the right path to get there.

“You see much more diversity. And that’s what always draws me to SIGGRAPH”

Ariel Shamir: I can start with saying that SIGGRAPH has always excited me since I began at Reichman University. What I like about the environment is the interdisciplinary nature. They have, of course, science, which is where I fit in. But they also have a lot of art. And this mixture of people is what draws me to these types of conferences.  If you ask me what is special about this year, what is really nice is the way that AI is entering the field. And there’s an adjunct field called vision, where it basically took over. So everyone does AI now in vision. You can also see more classic works where you don’t involve any AI, which is also nice, again, because it’s more diverse, both outside and inside the technical papers.

Ruizhen Hu: I think it’s definitely the trend of combining computer graphics with robotics. There’s several sessions like those traditional professors in computer graphics. They mainly work on animation, now they switch to robotics and try to apply those computer technologies to help to boost the intelligence of the robot. Because I think in China, sometimes researchers at SIGGRAPH cannot get big funds because they think it’s all game, video. It’s not very important for society. Now when we start to use technology to help with the improvement of machine intelligence to really help those products in industry and also improve humans’ lifestyle in a more intelligent way. So currently I think it’s a good trend, both in China and also in this international research community.

Yuki Ban: This year, one of the standout characteristics is the remarkable diversity of topics being covered. While VR has consistently been a dominant theme at e-TECH each year, this time the scope has expanded significantly beyond just VR to include robotics, AR, AI, haptics, development using HMDs, and a wide range of applications, as well as foundational developments and technologies. This breadth of coverage is a defining feature of this year’s event. Among these, what stands out as particularly new and emerging is the inclusion of experiences incorporating Generative AI, which has recently become a very hot topic. This seems to be a key trend at the moment. Additionally, when it comes to VR, technologies related to avatars—already a prominent focus in previous years—are again represented by a diverse array of innovative offerings, which I feel is another hallmark of this year’s event.

Challenges and Advice from Industry Experts

In your journeys from researcher to professor to prominent chairs of SIGGRAPH, what memorable challenges have you faced? What advice would you give to young people getting into the field?

Ruizhen Hu: I think the biggest challenge I faced maybe, was when I decided combine computer graphics with robotics. I’m more into geometric modeling, but I knew nothing about robotics at the time.  My one undergraduate and I spent a lot of time together searching for problems that fit our expertise. Eventually we found it: transport and packing. A very complicated optimization problem.  So I knew how to utilize my expertise to help. Researchers in robotics, they used heuristics a lot,  we started looking at those papers in 2018, and we published our first paper in SIGGRAPH Asia on the topic in 2020.  That was a very important experience for me. And also a big challenge – sometimes you’re just not very confident about what you do because it’s an unknown area and you haven’t any guidance. You need to explore. The advice I give is it’s always good to step out of your comfort zone, maintain your persistence, and really try to find something that can utilize your expertise. Find something you are really good at and try to solve it in an elegant way.

Takeo Igarashi: Before becoming a professor, it was easy just doing whatever I wanted. I was lucky. But now after having a professor, my role changes, right? So now I’m more like a manager. And managing is different. Yeah, maybe not the right answer for you, but that’s my experience.

Yuki Ban: What I often tell my students is to value the small discoveries and insights they encounter in their daily lives. For example, with the concept of perceiving strong stimuli [see below], I began by pondering the afterimages or residual effects of bright light—how those lingering impressions affect our perception. I encourage students to constantly think about how the various stimuli they encounter in everyday life can be incorporated into their ideas. As for my research, I focus on using everyday discoveries to make small but meaningful improvements in the world. I strongly believe in the importance of cherishing and utilizing these daily moments of awareness.

---

KEN ANJYO

Your research spans visual computing and the application of differential geometry in graphics. Can you explain this in simple terms?

KA: It is the merging of 3D models naturally with the hand-drawn. Everything can essentially be drawn by hand. But sometimes we add 3D, like big monsters are very hard to describe by hand. Shading or rendering a “cartoon look”. We often need much more sophisticated tools to control light and shade. Almost like hand-drawing. So we develop new techniques.

I was reading that you enjoy teaching mathematics to young learners including your own grandchildren. How do you approach making complex concepts engaging to young minds?

KA: I show the effect, the end result of the ideas. Not by explaining the exact meaning of differential geometry. I try to express that, we can deal with all kinds of things with mathematics, with equations you will learn in the near future, that kind of way.

As the Featured Sessions Chair for SIGGRAPH Asia 24, what are you most excited to showcase?

KA: SIGGRAPH is always high speed, involving new techniques into our field. And this year we have many AI or deep-learning based ideas as you’ll see in the Featured Session. They show how to integrate generative AI into production work, such as Director Aramaki’s session. I’m also looking forward to NVIDIA’s presentation on digital human techniques. In my own 1995 SIGGRAPH paper about character walking, we extract the essence of walk briskness to control the degree by parameter. At the time, it was kind of heuristic to make it. But now we use AI.

I was reading that you were a drummer as well. Do you still play?

No, no, no, no. My last performance was maybe in 2018. I chaired that conference, but at a midnight show the president of Polygon Pictures – he is basically a rock singer – invited me to join his team, and then I played that one. I hope I can find the next chance to play.

What was your dream project?

KA: In Pixar’s old film Luxo Junior, they needed two or three years to make five minutes of animation. At that time there was no distinction between artist and engineer, everyone worked together to reach the goal. I’m old enough to remember such things… Anyways, before joining OLM Digital I worked with a Japanese broadcasting company making titles for TV. It was around that time I visited Studio Ghibli, they were making Princess Mononoke and they asked me to automate the background animation. 3D camera moves were hard to create by hand, very hard, almost impossible. So we built a unique interface (GUI) to address this problem. It was presented at Siggraph 97. We were given a still image and we were instructed to give it a rough 3D structure and to split apart the foreground and background areas. At that time we used Photoshop to fill the missing sections of the background, but now, it is done by AI. So new things improved our original idea.  And so, back to the original question. I’m not an engineer. I’m an artist. But, of course, heavily from the engineering side. It’s always fun to make such things.

Ken Anjyo is a renowned figure in the fields of computer graphics and animation, with a career spanning decades in academia and industry.  He has been active for SIGGRAPH programs for years as an author, organizer and committee member. He chaired SIGGRAPH Asia 2018 and co-founded the ACM Digital Production Symposium (DigiPro) in 2012

---

DYLAN SISSON

RenderMan, cornerstone of visual effects for decades. How has it influenced storytelling capabilities of filmmakers?

DS: Early on it was really interesting to see the relationship between what was possible and how far the filmmakers could push. And if you look at Toy Story, there’s a reason why we made a movie about plastic toys, because everything looked like plastic at the time. Jurassic Park, it was easier to make a movie about scaly creatures than furry creatures. And those films worked within the limitations of the technology at the time in very interesting ways, but maximized the new opportunities that we had with the new micropolygon renderer architecture, you know, the REYES algorithm. We were able to render massive amounts of geometry we couldn’t do before with motion blur that was anti-aliased and composited into live action. And that at the time, like the first time somebody rendered a dinosaur, or the first time somebody rendered a plastic toy, was a historic moment. Now we’ve gotten to a place where renderers can render almost anything that you want physically, they can simulate light transport

And the challenge really is providing direct ability for the creatives to do new things. And you see at Pixar a trend towards, well, let’s make characters not out of polygonal meshes, but let’s make them out of volumes and things that we wouldn’t have been able to potentially render 10 years ago. Yeah, so all that kind of stuff is a new territory for us. It’s maybe historic and groundbreaking in some ways, but our focus is still on that frontier of like, what’s next? What’s next as far as how can we tell a story? So if you look at Pete Sohn in Elemental, his story about characters that are fire and liquid, that’s something that we couldn’t have really rendered like 10 years ago or even five years ago.

Tell me about the REYES algorithm.

DS: It’s named after Point Reyes, which is in California, and that’s where Loren Carpenter came up with the concept of micropolygons and bucket rendering.

As he was standing in the water, right?

Yeah. And for that moment he was the only person on earth who knew this! To make digital VFX feasible, the original architects of RenderMan invented all sorts of new technologies, including the alpha channel.

With the rise of real time rendering engines like Unreal, what role do you believe RenderMan will play in the future of visual effects?

DS: I think RenderMan always focused on kind of that race towards the top and the most complex images with the most creative flexibility. And in order to achieve real time right now, we still have to do a lot of pre-compute. So there’s a lot of caching and there’s a lot of compromises with the geometry and it’s getting better and better and better. But at the same time, it’s not full on path tracing. So there’s some limitations that you have when you’re doing real time rendering that you don’t when you’re doing the offline rendering. So to get that final 10% or 5%, you need a full blown path tracer to do that. And it turns out people don’t necessarily need to watch movies in real time. It’s great if the artists and the people that are making it can work in interactive time. So we’re really focused on providing fast feedback for artists, but real time rendering is not our primary goal.

I think what’s interesting in the very near future is giving control over light, shadow, volumes and lenses to the animator.

DS: And layout. Sometimes there’s certain shots that are composed in a reflection. That type of shot is much easier to work with if layout, animation, and lighting can all see the final ray traced reflection.  In Incredibles, there’s a scene where Mr. Incredible is in bed and he has his phone. And that’s the primary illumination of the scene. For the animation to work correctly, the animator had to have the light illuminating him in the right way to make it look good. So having that kind of control to see the final pixels throughout the process is something that we’re working on. So that’s one of our primary goals. We want to bring it all together so the lighters can work with the animators on shots that maybe we couldn’t have done that shot before because it would have been too hard to set up.

Anything that puts more control into the hands of the animators and the artists.

DS: Yeah, that’s maybe more of a focus for us now than actually adding new features to the render. Because we’ve gotten so many features now, we can do a lot of different stuff and make a lot of different images. So if somebody wants to render something that looks X, Y, and Z, we can do that. It’s just about delivering the best experience. So the creatives have the biggest opportunity to retain the creative flow when they’re working.

That’s what’s exciting when it comes to technology driving the art forward. Having to no longer have to worry simulating hair, water…

DS: When the animator can see the simulation of hair, with rain, or snow, and all these things together, all of a sudden you can compose shots and do things that you couldn’t do before. Being able to bring all the stakeholders in the room and have everyone contribute and have the director there and say, oh, well, can you try this thing or that thing, all of a sudden you can get someplace better, faster than you couldn’t have gotten to before.

Do you have a personal memory of a time where the teapot appeared unexpectedly in a project or an Easter egg?

DS: Yeah, I guess for me, when I was starting out, Martin Newell’s “Utah Teapot” was in every software package, so I could just bring in the teapot and light and shade with it So it was one of the models that I first would practice out shading and creating a plastic shader, a metal shader, I would do it on the teapot just because it has concave and convex surfaces and it was just perfect. But there was one day when I found out that there was actually a geometric primitive, like an RI point or an RI sphere, in RenderMan, we also had an RI teapot and you could just call an RI teapot and it would just bring up Martin Newell’s teapot, the same RIB. And I think that was after I’d worked with RenderMan for maybe five years, I learned about this hidden primitive inside of RenderMan and I thought it was really, really cool. That’s cool. Yeah.
That it was on the same level as Plato’s Plutonic forms.

Dylan has been a Key figure at Pixar Animation Studios since 1999, significantly contributing to the evolution of RenderMan. As an active artist, he explores emerging media in VR, AI, 3D printing, and stylized rendering. His experience in 3D spans over two decades, with a background in traditional art and illustration. Additionally, Dylan is the creator and designer of Pixar’s RenderMan Walking Teapot.

---

ARIEL SHAMIR

It must be great to get access to what everyone is currently working on.

AS: The challenge now is that SIGGRAPH is becoming increasingly popular, much like other conferences in the field. This year, we saw a record number of submissions and accepted papers, which has led to the fast forward session becoming quite lengthy—perhaps a bit too long for some. However, it remains an enjoyable and engaging way to get an overview of the conference’s content in a single extended session. It allows attendees to pinpoint talks or papers that catch their interest and decide what they’d like to explore further.

Your talk, Prompt Aligned Personalization of Text-to-Image Models, is about creating graphics easier and faster. Explain it to someone who’s just getting into art and video creation.

AS: My paper is part of a larger collection of research focused on developing tools to make generative AI more accessible to everyone. By now, most people are familiar with models like DALL-E, which generate images from text prompts. This is now a major trend in both graphics and computer vision. While these models are groundbreaking, they are far from perfect, and many papers are tackling their various limitations.  Our research addresses two key challenges. The first is that while it’s relatively straightforward to generate simple images, like a dog on a beach, things get trickier with more complex prompts. For instance, if you specify a dog wearing a red hat on the beach with a cat running around with a ball, the models often struggle to follow all the instructions accurately. The second challenge involves a growing trend called personalization, where users want to generate images of specific, familiar subjects, such as their pet, themselves, or their friends. Combining these two — handling intricate prompts while also personalizing content — is a difficult problem. Our paper focuses on a solution to create highly detailed, complex images that are also personalized to the user’s specific needs.

You’ve been to Japan many times, in 2022, you were a visiting professor at the University of Tokyo. Has your experience in Japan shaped your vision of computer science?

AS: I have to say that I deeply appreciate Japan and have a strong admiration for Japanese culture, particularly Japanese aesthetics. While I can’t say for certain whether Japan and its culture have directly influenced my work in computer science, many of my projects revolve around aesthetics, art, and abstraction. Looking back, I’ve been drawn to Japanese art and aesthetics since childhood, so it has likely shaped my creative perspective in some way.  Visiting the University of Tokyo was an incredible opportunity for me to explore Japan, visit museums, and fully immerse myself in its culture. During my visit, I was hosted by Professor Takeo Igarashi, who, coincidentally, is also the general chair of this conference. We share many common interests in our field and had wanted to collaborate for years. His lab is remarkable, with many brilliant students, and the experience was both inspiring and rewarding. This visit also laid the foundation for an ongoing collaboration between our labs and students, which has been very exciting and productive.

You hold 17 patents in areas like driving performance, classification, and 3D modeling. What inspires you to branch out into all of these different areas?

AS: I am naturally very curious, and this curiosity is one of the main things that drew me to SIGGRAPH and SIGGRAPH Asia conferences. It’s also a big part of why I became a professor. Being a professor is largely about asking questions and seeking answers, which aligns perfectly with an inquisitive nature. Most professors, I believe, share this curiosity — whether it’s focused on a specific topic or spans broader areas. I fall into the latter category, with an interest in a wide range of subjects. Obviously, I’m not an expert in all of them, and I focus primarily on areas related to computer science and computer graphics.  Beyond academia, I’ve been involved in industry as a consultant for many years, working with both large companies like Disney and Google, as well as startups, particularly in Israel, which is often called the “startup nation.” When I choose to consult for a company, first of all, it has to interest me, and second, I have to feel that I give some value to the company — otherwise, there’s no point. This collaborative work often leads to new technologies, which companies frequently patent. As a result, you’ll find my name associated with various patents, many of which emerged from these consulting projects.

Story Albums combines advanced computational techniques with storytelling. Tell us a bit about it. 

AS: Story Albums was a project I worked on during my time at Disney, rooted in my longstanding interest in storytelling. At Disney, there’s a strong belief that “story is king” — everything else serves the story. I completely resonate with this philosophy. One of my research directions has always been to explore ways to tell personal stories using people’s own materials, like their photos and experiences. The idea behind Story Albums was to take an ordinary experience — perhaps a trip to the beach or Disneyland — where you’ve captured a multitude of photos, and create a fictional story. Often, people take many pictures but rarely revisit them. Even back then, there were reports suggesting that most photos taken are never viewed again. This served as our key motivation: to re-imagine these images and make them more engaging. The concept was to craft a fictional story based on your experiences, transforming your collection of photos into a personalized narrative. For instance, after a trip to Disneyland, you could create a story where your child is on an adventurous quest to find Pluto, meeting characters along the way. For adults, it’s clear the story is fictional, but for children, it creates a magical and memorable tale. This personalized story could then be turned into a movie, a book, or another keepsake.  At the time, the technology we used was relatively basic compared to what’s available today. It was before the rise of large language models and generative AI. We relied on graph-based algorithms and heuristics to build these narratives. Today, however, there are likely websites where you can simply upload a picture of your child and generate an entire storybook in seconds. Despite this, the core idea of using storytelling to give life and meaning to personal memories remains as compelling as ever.

Professor Ariel “Arik” Shamir is the former Dean of Efi Arazi school of Computer Science at Reichman University in Israel. He has more than 150 publications in journals and holds 17 patents. He has a broad commercial experience working with, and consulting high-tech companies and institutes in the US and Israel. His main research field lies at the intersection of graphics and vision with machine learning.

---

RUIZHEN HU

Let’s talk about your new robot planning method, PC Planner. 

RH: Yeah, so the PC Planner, the key idea is to solve an equation. For a path planner, we have the start point, we have the end point. So the goal is to find the shortest path with the shortest time from the start to the end. The key idea is if we have a time field which records the shortest time from the start to the point, if we already have this kind of field, then we can just follow the gradient direction of such field, which can guarantee the shortest path. Which means we need to define what we call the “time field”. But how to get it? This is the key of the Eikonal equation. The Eikonal equation builds a connection with this kind of time field with the speed of the robot. So it’s kind of because conceptually, this is the goal point, this is the start point.

If you have a high speed, you can go to this point faster. So the transport time is lower, right? This is common sense. But the Eikonal equation goes deeper. It says if the speed is faster, then the change of this speed will have less effect to the shortest time when traveling to the goal point. So once we define a speed field, and the speed is defined for every configuration, then we can solve this equation to compute such a time field. With such a time field, we can just compute a vector field based on the time field and follow the gradient direction until we reach the ultimate goal. Then the Eikonal equation says that if we follow this direction, the path is guaranteed to be the shortest path.

What is the Eikonal equation, and what is the “curse of dimensionality”?

RH: The traditional method is a sample-based. So they sample a lot of paths, and then they pick the optimal one. If we do the path planning in only two dimensions, x, y, it’s simple. If we do it for the robot arm, there is a lot of configuration space, for example, 20 dimensions, then it’s 20 times 20 times 20, so it boils up. This is what we call the curse of dimensionality. What we do now is not sample paths and then pick the best one. We compute the gradient direction from the time field, then we follow the direction, we get the next point. In this way, we don’t need to explore too much, which makes the planning much more efficient. That goes back to how to get the time field. What we do in this paper is find a way to solve this equation. Once we define a speed field, we can compute the time field accordingly. In this way we do not need to compute the shape anymore because it’s already encoded in the speed field.

If the environment itself is not static, how do you account for that?

RH: Yeah, that’s a very good question. So currently, we assume the environment is static.

Your work in LLM and the enhancement of household rearrangement is fascinating. How do you picture this evolving?

RH: So the first paper, PC Planner, is focused more on low-level motion planning. You already know this is a starting point, this is a goal point. You want to find the shortest path to go there in an efficient way.  So in the other paper, our key idea is to use high-level planning. High-level decision. So the goal is we want to rearrange the room, then what actions you need to take, which object you need to move from where to where. This is high-level decision. So what I envision is eventually, it should be all integrated together, after the high-level decision, we need to relate to make sure it is executable. And also, currently the high-level planning paper, we only focus on a very specific task, rearrange the room, right? So we also want to be more open. You can tell me all kinds of tasks, and we can learn to divide those tasks into several steps and make sure we can really execute all those steps and really help to accomplish different kind of goals.

Using LLMs to determine why an object should or should not be where it is really sparks the imagination regarding autonomous robots roaming around the home.

RH Yeah, yeah. This is also one of our key ideas. For the rearrangement task, different people like different kind of rearrangement. So we kind of build a scene graph for the current scene, and we analyze the context, the object rearrangement in the current scene, and we try to analyze the personality or personal preference for where those objects should go. And then based on this, we determine which object is displaced and where is the right place to move this object to. 

How might a robot navigate Tokyo, known for bustling train stations and massive crowds?

RH: I think there are many directions we can go, like whether we make use of transportation, right? Because robots can be very intelligent, they can go into the station, take the bus, go to anywhere else. So if we take this into consideration, it would be very interesting because it also involves the scheduling, right? So if you want to accomplish a task, it’s no longer a static environment. You do path planning, which is also dynamic. And so the scheduling here then will be very important, like which time to take which line. This can also help us to make decisions when we want to go here and there, right?  So when we consider the problem setting, we both assume the environment is kind of static because if dynamic changes, everything changes. We need to recompute, reanalyze this. This makes things more complicated. So if we focus on path planning or high-level planning, as the environment gets dynamic, the problem becomes very challenging. And I know there’s a recent trend to deal with this kind of problem setting, which is unknown environment with dynamic change. So how to deal with this? How to do this local planning based on the vision change? How to escape those objects? This is what I think is quite important especially in a place like Tokyo. Very crowded and a lot of moving people. If you want to do some autonomous navigation, you have to be able to capture those dynamics and to give quick feedback and adjust your strategy.

Ruizhen Hu is a distinguished professor at Shenzhen University, China. Her research interests are in shape analysis, geometry processing, and robotics. Her recent focus is on applying machine learning to advance the understanding and generative modeling of visual data.

---

TAKEO IGARASHI

How do you think the creative world is going to change due to generative AI?

TI: I’m sure it changes. A very different way of making content, and a very different way of consuming content. How people perceive results is very different. There are many possible directions to go, but for me, the most interesting is the entire landscape changes. That’s the most interesting part. Previously, experts spent hours to generate something, but now it’s very easy to do for everybody. We in our professional enterprise, how we perceive content, how we manage, how we enjoy… I think the value system is changing. That’s most interesting for me.

Do you think that being here at SIGGRAPH gives you a bird’s eye view of the current state of the art?

TI: Definitely, in terms of the computer graphics area. But AI is now much more diverse. Computer visual people do lots of interesting things, and AI conferences are doing something interesting in the industry. In a sense, this conference gives a very good bird’s eye view, but also I’m aware that it’s not everything. 

Your work with TEDDY, the sketch-based model…

TI: Yes, that’s right. I’m ashamed of it.

Well, many say that it revolutionized how people interact with 3D design. How do you see TEDDY evolving in the future?

TI: In a sense, that kind of problem is… When I worked on TEDDY, 3D content generation was very, very difficult. You had a very complicated user interface. But now I turn prompts to 3D, or 2D images to 3D. So in a sense, simple, easy content creation, either 3D or 2D, is kind of done. As a researcher, we have to always see beyond that, right? That’s my observation. I’m not directly thinking of future of TEDDY right now. We are looking more into different directions.

How do you picture the future of human-computer interaction? 

TI: So, in my point of view, I do two different directions. One direction is disappearing user interface or interaction, empowering people. Like, augmented human is a big thing in Japan. So just augmented human as if it’s a part of the body or brain. So user just do more, like a superhuman, and then computer becomes invisible. That’s one direction. And another direction is computer is still outside of us. It’s a tool, or agent, or additional person. So these two approaches, you know, each guy says this is the best way. But I say these two different ways, yeah, I think it’s interesting to see which one. I think both coexist in the future.

If you could design a new interface that lets people interact with the world in a different way, what would that look like?

TI: One possible answer people might say is a brain interface. Directly go to something, but I’m kind of skeptical. I think the fundamental change doesn’t happen in a kind of superficial thing. Text is good for text. Sketch is good for text. Direct manipulation is good for something. Maybe brain, maybe something. So it’s more important to think more how this existing tool changes over time. I don’t have a right answer.

If you could bring one piece of technology from your research or otherwise to life in the real world for everyone to use tomorrow, what would that be?

TI: So regardless of what’s actually doing in my group, you’re talking about just imagination? My research interest has always been content creation. So something I want to develop is to empower people to create content. And today, as I said, already text prompt or sketch to 3D or image to 3D is already there. But you have to come up with a prompt or you have to come up with a sketch. So we’re moving research interest more into the earlier stage to help people actually conceptualize what they want. So helping people with the early stage in creation is something I want to do.

Takeo Igarashi is Professor of Computer Science Department at The University of Tokyo. He has made significant contributions to human-computer interaction and design, earning accolades such as the SIGGRAPH Significant New Researcher Award and the CHI Academy Award​.

---

YUKI BAN

You’re known for your research in illusions. Tell us about the inception of your work.

YB: My specialty is cross-modal display. The combination between five senses, for example, vision and haptics, and vision and auditory, combined with new perception. I usually use the example of Japanese kakigori (shaved ice), as one example of cross-modal. Different color Kakigori, the taste may all the same, but the experience is different. For example, the strawberry, the melon, that difference depends on the smell and visual color changes our perception of the taste. It’s a cross-modal effect.  It’s a famous illusion of the five senses. Using a cross-modal effect, we can display various sensations with very simple hardware. But this time, I used a different illusion. This time, the illusion is displaying different images to the left eye and the right eye. Did you try that?  

Yes.

YB: It’s a famous illusion in the psychological field. But that illusion is not used in the engineering science for the displays. So we try to use the traditional illusion for the display technique.  

Your display uses three different simulated environments. A dragon scene, fireworks, and a horror game. What do you think would be another exciting use of this technology? 

“I have a deep sense of gratitude toward SIGGRAPH Asia, that’s why I accepted the role of Emerging Technologies Chair. I’m honored to be serving in this capacity.”

YB:This technology is primarily being utilized in the entertainment field, there are developments aimed at creating new applications such as magic effects or elements for horror – experiences that evoke a sense of the extraordinary. In the entertainment domain the technology itself is quite simple. However, identifying the most effective combinations is something that still needs to be clarified. Currently, guidelines are being developed to address this, and once they are in place, the technology can be readily applied to various areas like VR, AR, or glasses-free 3D panels. I believe it has the potential to make a significant impact in the entertainment industry.

Tell us about your research for next year.

YB: We are conducting various studies using illusions, and one of our new approaches involves the “thermal grill” technique. This method uses illusions to make individuals perceive extremely strong stimuli. For example, one study we presented last year demonstrated how to make users perceive bright light in VR. In the real world, when you look at an intense light and then look away, you experience an aftereffect or lingering image. By replicating this phenomenon in VR, we create the illusion that users have been exposed to a bright light, altering their perception of brightness within the virtual environment. Our research focuses on replicating bodily responses to intense stimuli in the real world and feeding those back into VR. This triggers illusions that make people feel as though they are experiencing strong stimuli. We have been working on several related projects and aim to present one of them next year.

Earlier, I mentioned the goal of making the world better through incremental improvements. As part of addressing this, we developed a cushion that mimics the abdominal movements associated with deep breathing. When users hold the device, over time, they find their breathing synchronizes with the rhythm of the cushion. Deep, slow breathing is essential for relaxation, but maintaining this rhythm can be challenging without prior training, such as in yoga or meditation. This cushion enables users to unconsciously adapt their breathing pace simply by holding it. This approach leverages a kind of bodily illusion. We collaborated with a company to bring the device to market in 2018, and as a result, it became my first research endeavor to reach full production. The product, now called Fufuly, officially launched last September.

YB: Personally, I’ve found it particularly useful in the context of the growing number of online meetings and interviews. For instance, I keep the cushion nearby, much like having a calming indoor plant in view. Holding it before an online interview helps me settle my nerves and feel more at ease, which has been incredibly beneficial.

I can picture this in a larger scale, like a breathing Totoro bed. 

YB: I’d love to create something like a Totoro-inspired bed someday.

Yuki Ban is a project lecturer with the Department of Frontier Sciences at The University of Tokyo. His current research interests include cross-modal interfaces and biological measurement.

---

SIGGRAPH After Hours: A Glimpse Behind the Scenes

While the sessions and keynotes offer a glimpse into the future of computer graphics – like the advancements in generative AI that Takeo Igarashi highlighted, or the innovative applications of robotics discussed by Ruizhen Hu – the real magic often happens after-hours. This year, the bustling evening events brought together a diverse crowd, mirroring the interdisciplinary spirit often mentioned by Ariel Shamir, fostering a unique melting pot of ideas and camaraderie. And perhaps, as Dylan Sisson suggested, these chance encounters could even spark the ‘next big thing’ in the ever-evolving world of computer graphics.

At the first reception party, the lines for food stretched long – especially near the coveted meat station – but the atmosphere buzzed with excitement. Speakeasy was packed to the brim with attendees. Among them, Sergey Morozov of MARZA met an interesting Canadian developer originally from Ukraine, who leads projects for the BC Attorney General’s Ministry. His impressive photography and creative ventures added a fascinating dimension to the evening.

For Kasagi Natsumi, also from MARZA, it was her first time navigating the party circuit, where she focused on networking. At Speakeasy, she connected with CGWire’s Frank Rousseau, whose Kitsu project management tool, popular among studios like Blender, is making waves in the industry. Kasagi also explored the Blender booth, meeting Gaku Tada, a talented Japanese developer whose Deep Paint plugin and visual artistry have earned him credits at Digital Domain, Weta Digital, and ILM.

Sergey and Kasagi had a serendipitous encounter with Masha Ellsworth, Lead TD from Pixar, another Ukrainian expat, whose seminar on animation careers provided valuable insights. She shared that hiring in studios like Pixar often hinges on project timing rather than individual merit, offering hope to those waiting on replies to applications. The TD and her daughter are fans of Japanese animation, particularly Chi’s Sweet Home, and took a liking to MARZA’s Samurai Frog Golf.

While the sessions and keynotes represent the bleeding edge of computer graphics, the hidden treasures lie in the connections forged, new friends made and the stories exchanged after-hours.

SiGGRAPH 2025 takes place in Vancouver, and SIGGRAPH Asia 2025 in Hong Kong.

---

interviews and photos by Brent Forrest, Chris Huang, Sergey Morozov and Natsumi Kasagi
The authors and LesterBanks.com have no commercial affiliations with any of the companies, products, or projects mentioned in this article.