What is Polycam?
Polycam is the most popular reality capture app on iOS and Android. People use Polycam for everything from home renovations (getting instant 3D floor plans) to 3D printing (scanning objects to replicate them). 9.6M 3D models were processed with Polycam in 2024.
For lidar enthusiasts, Polycam is one of the easiest ways to tap into the powerful lidar sensor on iPhone Pro models (12 Pro and later). Lidar is not included for Android users, but photogrammetry, which is a fancy way of saying “take a bunch of photos and let an algorithm stitch them into a 3D model,” provides similar functionality. Lidar, by contrast, actively sends out laser pulses and measures how long they take to bounce back, making it more accurate for scanning depth and geometry.
A Major Upgrade for Polycam
Polycam is getting a major upgrade – Vision 25. Vision 25 brings a wave of improvements, including better scene editing, AI-assisted modeling, and advanced measurement tools. Here’s what stands out:
- Space Mode: If you want to scan an entire room or building, Space Mode makes it ridiculously easy. It creates not just a textured 3D model, but also a 2D floor plan and a vectorized model—all in one go. This is huge for architects, real estate pros, and anyone looking to quickly document interiors.
- Scene Editor: One of the biggest past headaches—merging multiple scans—is now way smoother. You can stitch together different 3D captures into a single, seamless scene, making multi-room or even multi-floor scans far easier.
- AI Model Generator: Upload a single image, and Polycam’s AI will generate a complete 3D model in seconds. This is still in its early stages, but it has the potential to be a game-changer for quick object captures.
- AI Spatial Report: Now, instead of just getting a 3D model, you can generate a detailed report with measurements, dimensions, and material estimates—making Polycam even more useful for professionals in construction and real estate.
- Gaussian Splatting in Object Mode: This one’s a bit technical, but it’s a new way of representing 3D data that improves realism and efficiency in object captures.
Polycam Scan Using iPhone 15 Pro. This spinning plant model was captured using an iPhone 15 Pro. Overall scan time was less than 10 seconds. The Polycam app is easy to use for someone with no prior experience with 3D technology.
The Most Innovative Aspects of Vision 25
While a lot of these features improve workflow and usability, the AI Model Generator and Gaussian splatting stand out as the most cutting-edge.
AI-Generated 3D Models from a Single Image
Creating a full 3D model from just one image? That’s wild. AI-based 3D reconstruction has existed before, but Vision 25 makes it faster and more accessible than ever. This isn’t photogrammetry or LiDAR—it’s a neural network trained to infer depth and shape from a single 2D photo. I will be curious to see how this functionality performs. The field of computer vision has been attempting to make 3D models from single images for over a decade, yet nothing has crossed the threshold of utility. Marked improvements have been made recently with deep learning algorithms like Convolutional Neural Networks (CNNs) as well as techniques like Gaussian splats, so we may be due for some impressive developments on this front. I was playing frisbee with a friend who has only one eye the other day, which got me thinking—if he can still judge depth well enough to catch a flying disc, then maybe a single image (and, in his case, a somewhat underpowered neural network) can be enough for depth perception. Of course, in reality, this relies on a time series of single images—but still, it’s a pretty solid argument for how much information can be extracted from just one viewpoint.
Gaussian Splatting: The Next Evolution in 3D Representation?
Polycam is introducing Gaussian splatting, a hot topic in the 3D world. Gaussian splatting is a new approach to 3D rendering that replaces rigid meshes with soft, overlapping Gaussian distributions. Gaussian distributions—remember? The normal distribution, STATS101, it wasn’t just for grading on a curve. Each splat is a 3D Gaussian with a position, size, color, and opacity, allowing for smooth, realistic reconstructions without sharp edges or excessive polygons. The Gaussian part comes from the fact that each splat follows a probability distribution, meaning objects blend naturally rather than looking rigid or pixelated.
The Future of Polycam and Reality Capture
Looking ahead, Polycam is on the cutting edge of turning smartphones into full-fledged 3D capture devices. It’s not just about lidar or photogrammetry anymore—it’s about using AI, advanced data processing, and ever-improving phone sensors to capture reality faster and more accurately. In the coming years, we can expect:
- More AI-powered reconstruction tools, making 3D scanning as easy as snapping a photo.
- Better smartphone sensors, improving lidar resolution and depth perception.
- Cloud-based 3D processing, allowing users to offload complex computations to Polycam’s servers.
- Seamless AR and VR integration, bridging the gap between 3D capture and interactive experiences.
Reality capture is no longer just for high-end hardware or professionals—it’s becoming something anyone can do with just a phone. And with updates like Vision 25, Polycam is making sure it stays at the forefront of that revolution.
So whether you’re scanning your next DIY project, digitizing your dog, or creating the most detailed 3D model of your sandwich ever, Polycam is proving that reality capture isn’t just the future—it’s already here.
For more information, visit https://poly.cam/vision-25.
