Ultraleap Stereo IR 170 review: use your hands naturally in VR!
I have had the joy of trying the new UltraLeap Stereo IR 170 hands-tracking sensor for VR and evaluate its performances. I have appreciated it a lot, so I thought it could have been a cool idea to make a post for you to describe how it is and compare it with the previous Leap Motion controller. Are you in?
UltraLeap Stereo IR 170
Some weeks ago, I have reviewed the new Leap Motion Gemini (v5) runtime, and I have appreciated its robustness. I tried it with the Leap Motion controller, the one that most of us have bought years ago to experiment with the OG Vive or the Oculus Rift DK2/CV1… but actually, UltraLeap (the previous Leap Motion) has created a second-generation camera sensor. It is called “UltraLeap Stereo IR 170”, a name that is totally unsexy, so from now on, I will call it UltraLeap 2.
UltraLeap 2 is actually not an accessory for VR, but an “evaluation kit”: that is, it is a product thought for hardware makers, that can buy it, attach it to their headset and test it, and then if they like how it works, they can contact UltraLeap to decide how to integrate it into their hardware products. So if you are an AR/VR headset manufacturer, or you are someone that develops digital signage solutions that would like to implement hand tracking in kiosk machines, you can buy this sample, make some tests, and see if it is the right solution for you.
Of course, it is not only for them: if you are a maker or a research center, you can also buy it as an accessory for your headset, to experiment with high-quality hands tracking in VR. It is not a consumer product, though, and in fact, it has no FCC certification and you have to attach it to your headset with a plastic mount that you have to 3D print yourself (Ultraleap can help you with some preset designs, though) or using some duct tape. I’m using duct tape at the moment because I was too eager to try it that I didn’t want to wait for the 3D print 🙂
Let me dig into how it works. But before just let me remind you that in this article I’ll just talk to you about the hardware, because the UltraLeap 2 works with the exact same runtime and SDK as the Leap Motion Controller. If you are interested in the tracking software, you can read my detailed review article on the Gemini runtime.
UltraLeap Stereo IR 170 video review
I know that you love my review videos, so here you are my impressions on this new sensor by UltraLeap!
The textual impressions follow here below, instead!
UltraLeap Stereo IR 170 Specifications
Here you are the main specifications of the UltraLeap 2 evaluation kit:
- Dimension: 145mm L x 18.6mm W x 11.1mm H
- Interaction zone depth: Between 10cm to 75cm preferred, up to 1m maximum
- Tracking FOV: 170×170° typical field of view (160×160° minimum)
It is around +20cm of tracking depth and +30° horizontal FOV and +50° vertical FOV from the previous Leap Motion controller. As we can see, this will make for a big difference in the end. These features are possible thanks to the fact that the two cameras are better than the previous ones and have a wider baseline, being so able to capture the depth of objects with better precision (because there’s a bigger parallax between the two images that they see).
Unboxing
The unboxing of the sensor has been very simple since the box just contains the sensor and some USB cables… but I couldn’t resist making an unboxing video the same!
Design
The device features:
- A port on the top, where you can plug a custom connector to attach the device to the USB port of your PC;
- Two wide-baseline cameras, that are the two internal blue-ish circles that you can see in the photos;
- Two IR emitting diodes, that are the two external holes on the device. They illuminate the surroundings with IR light, so that the cameras can better see your hands.
UltraLeap 2 vs Leap Motion Controller: what doesn’t change
Let’s start the comparison between the two sensors evaluating what doesn’t change that much. First of all, the accuracy of the tracking: the detection of the pose of the fingers is more or less the same with both sensors. With the UltraLeap 2, I still had some issues of some mis-detections between the pink and ring fingers when touching the thumb for instance. When the palm of one hand gets on top of the other one, the tracking of the occluded one gets lost. Sometimes there were tracking glitches, sometimes I had one hand that was not detected immediately.
Also regarding the FOV, it is wider, but it is not able to track correctly your hand in all the nominal 170°x170° volume with the same accuracy that it has when your hands are in front of your eyes.
All of this is to say: the UltraLeap 2 is a step forward, but it is not a perfect hand tracking solution. There are still glitches here and there, a big part of which are actually a software issue, and not a hardware one.
UltraLeap 2 vs Leap Motion Controller: what has improved
There are three things that make me say that the UltraLeap 2 is an amazing camera if compared with the previous device:
- Hands initialization. Most of the time, when you put your hands in front of you, they are already tracked (and tracked well). With the v1 sensor, there are times when the hands require some seconds to be detected, and in most cases, you first see your real hand (if you are using passthrough) and then you see the hand tracked in XR. With the UltraLeap Stereo IR 170, most of the time, thanks to its wide FOV, it initializes the tracking while the hand is still outside of your vision, and when you see it, it is already tracked. This feels like magic;
- Tracking fluidity. I have the impression that the tracking with this device is more fluid. Do you know the sensation that you feel when you play with the Valve Index at 120-144Hz? You feel the reality more fluid, more realistic, much more immersive than with the other headsets. Here I have the same sensation: it is hard to convey in a video, but it was like my virtual hands were faster and more reactive in VR, and so I felt them more “mine”;
- Big tracking FOV. You can really feel the difference of the field of view of this device compared to the previous one. If you are using a VR headset like the Quest+Link, as long as you see the hands in your visual field of view, they are tracked by Ultraleap. And when they are in the periphery of your vision, you can move them, even fast, and the tracking keeps working! Yes, in the periphery the detection of the pose of the fingers is less precise than when they are in front of you (and this is a problem in some apps that rely on it, like the Paint sample), but the tracking of the hands and the fingers still works! The big result of all of this is that you can use your hands naturally in VR to interact with objects, and they are always tracked, even if you move your hands and your head. The magic is that it simply works. While the Blocks demo (where you can move and stack some cubes) was a bit frustrating with the Leap Motion Controller, it was a true pleasure with the UltraLeap Stereo IR 170. I could interact with objects in a truly natural way.
UltraLeap 2 vs Quest hands tracking
Someone asked me to compare UltraLeap hands tracking with the one offered by the Oculus Quest. Well, UltraLeap wins by far: the tracking is more fluid, more resistant, and it allows stable two hands interactions, something that with the Quest is just impossible. It is incredibly better.
In defense of Facebook, though, I have to say that they are working on a less performant Android mobile device and they have started experimenting with hands tracking much after Leap Motion (that started around 2012). So let’s give the Quest the time it needs to improve 🙂
Price and availability
The UltraLeap Stereo IR 170 Evaluation Kit is available from UltraLeap’s partner vendors for around $250. If you are interested in the purchase, you can head to this dedicated page on UltraLeap website.
Final considerations
I have loved my time with the UltraLeap 2: I think that UltraLeap Stereo IR 170 + Gemini tracking is the first complete solution that showed me truly working real-time hands tracking. For the first time I have tried a hands tracking solution that can be usable for consumers for short demos: it is stable, it works on an acceptable FOV, it allows two hands interactions. It doesn’t lose the tracking easily and as soon as the related experience has been designed in a proper way (so that to cope for the temporary faults and doesn’t require to move the hands too fast, for instance), it allows for any kind of people to perform natural interactions within a virtual world. It is not just a gadget that only technical adopters can use.
It is not perfect, and it is not ready for an every-day consumer use, it should work 99.9% of the time on a full 180°x180° for that, but it is a big step in that direction.
If you are a hardware integrator, a maker or a UX researcher, I highly suggest you to check it out.
I hope this review has been useful for you. If you have any kind of questions, feel free to ask them to me in the comments, here below or on my social media channels. Cheers!
(PS And what about giving me a little donation on Patreon to support my work as a blogger?)
Disclaimer: this blog contains advertisement and affiliate links to sustain itself. If you click on an affiliate link, I'll be very happy because I'll earn a small commission on your purchase. You can find my boring full disclosure here.