Image Credit: @ 2025 James McCann
What if an infinity mirror didn’t behave like an infinity mirror? James McCann reimagined this classic optical illusion in his SIGGRAPH 2025 Poster. By combining custom LED arrays, parallax barriers, and the subtle textures of 3D-printed layers, he created a display that tricks your eyes and removes the usual mirrored repetition. The result is a striking “speed tunnel” effect that feels infinite but in a way you won’t expect until you step up and experience it in person. Read on to learn more about the process and evolution of this research.
SIGGRAPH: What first inspired you to revisit the traditional infinity mirror effect in this project?
James McCann (JM): Computer graphics has always been — in part — about fitting infinite worlds into little boxes. But even with all the cool stereo display technologies out there, it’s hard to do this with proper depth cues, for multiple users, with moving viewpoints, at high resolution (and the list continues).
Infinity mirrors are a display technology that can do all of these things, but with very strict limitations on the contents that can be displayed. I’ve been playing with the format for a number of years, starting with an “Open” sign I built for a rock and mineral shop in 2021.
The idea of breaking the mirror symmetry was something I started playing with in 2023 using custom rings of LEDs on two-sided circuit boards. This could make a good “speed tunnel” effect, but the mechanism was a bit too obvious to be really surprising. So I started trying to come up with better ways to do it and arrived at the prototype I showed last year at SIGGRAPH.
@ 2021 James McCann
@ 2023 James McCann
SIGGRAPH: Your work uses view-dependent appearance to remove the sense of reflections. How would you explain that concept to someone encountering it for the first time?
JM: If you’ve ever used a zig-zag-folded piece of paper to make artwork that changes appearance based on your view direction (an “agamograph,” an internet search tells me), then you’ve made something similar to the “pixels” in this display. The only difference in the “pixels” here is that they have LEDs setting the color of the front- and back-view images, and that there is a diffuser in front of them that blurs the details so you can’t tell that they have depth.
In graphics, we call any display that uses geometry and occlusion to send different images in different directions a “parallax barrier display” (and I’d say that both agamographs and my pixels qualify as such displays). Of course, that’s not the only way to steer light. If you use lots of tiny lenses to accomplish the task, we call it a lenticular display. And if you use wave optics effects to steer the light, we call it a holographic display.
One of the fun things about this project was brainstorming about how to solve the problem of sending different images in different directions with all of these different technologies. I settled on my particular parallax barrier approach because it was the easiest to implement with the technology I have in my home shop, but it is probably not the best solution. I am interested to see what other folks can come up with if they are inspired to take on this problem as well.
SIGGRAPH: When people experience the effect in person, what kinds of reactions would you anticipate?
JM: People’s reaction to the prototype is understated. Most folks haven’t thought through the limitations of typical infinity mirrors, so when the prototype breaks some of those limits (e.g., doing a “speed tunnel” effect), they don’t immediately notice. But if you get them to think about it for a moment, generally a smile breaks out and they realize it’s doing something pretty cool.
SIGGRAPH: Did working on this project change how you think about depth, repetition, or visual perception?
JM: For me, this was largely about developing a minimal prototype that I could assemble in my home workshop. The majority of my thought here was actually on design for small batch fabrication and the various little details that went into making something that was durable enough to bring to SIGGRAPH.
Perhaps my most interesting change in thinking was realizing that I could use the layer lines in fused-filament 3D prints for their optical effect (in this case, spreading the light in the diffuser). This saved me having to buy, test, and cut diffuser sheets. This realization was — I believe — inspired by an offhand comment that a Carnegie Mellon HCII professor made to me more than a decade ago about how the point spread function of lights shining through fused-filament prints might be an interesting design space. Sometimes, these interesting little observations just rattle around in your brain until you need them years later.
Also, I discovered a tiny tip on this project, and I am happy to pass on to other folks making portable demos: USB-C power delivery trigger boards can give you a nice way to get a stable 12v (or several other common voltages) input that allows either untethered operation (via power bank) or easy supply from any AC outlet (via USB-C charging plug). A lot more convenient than integrating a battery or trying to design your own power supply.
SIGGRAPH: What advice would you give someone preparing their first submission to the SIGGRAPH 2026 Posters program?
JM: To folks thinking of submitting something: Go for it! The program is great for first-timers who want a reason to chat with a lot of people, as well as for old-timers who want to show off some new ideas or side projects. Posters works as a venue for wild ideas that can stretch the bounds of SIGGRAPH, and for small details that you think people should know about but aren’t paper-scale. If you are on the fence about the program, walk through next time you are at SIGGRAPH and observe the interesting variety of work on show. Your ideas will fit in.
Feeling inspired after this conversation with James? Now is the time to submit your work to the SIGGRAPH 2026 Posters program. Submissions are due by Tuesday, 21 April.
James (Jim) McCann is an Associate Professor in the Carnegie Mellon Robotics Institute. He thinks that creativity – making things and ideas! – is the best thing that humans do. So his research involves producing creative tools (software, hardware, and even theories) to help people do more creating. He is particularly interested in systems and interfaces that operate in real-time and build user intuition; lately, he has been applying these ideas to textiles fabrication and machine knitting as the co-leader of the Carnegie Mellon Textiles Lab.
Jim earned his Bachelor’s degrees in Computer Science and Mathematics from University of Michigan in 2005 and his PhD from Carnegie Mellon in 2010 (advised by Nancy Pollard). He worked in industrial research at Adobe and Disney and made video games as TCHOW llc, before joining the faculty at Carnegie Mellon in 2017. He continues to make video games as TCHOW llc and music as part of Jimike.
