logo80lv
Articlesclick_arrow
Professional Services
Research
Talentsclick_arrow
Events
Workshops
Aboutclick_arrow
Order outsourcing
Advertiseplayer
profile_loginLogIn

An Easy-to-Use 3D Gaussian Splatting Solution From Polycam

Learn how the Polycam team has made 3D Gaussian Splatting easily accessible to both desktop and mobile users.

Throughout the past several months, a new ML-powered rendering method called 3D Gaussian Splatting has been gaining widespread attention on the internet.

For those unaware, Gaussian Splatting is a real-time 3D reconstruction and rendering method for images taken from multiple viewpoints. As part of this approach, the 3D space is defined by using millions of particles (Gaussians), each equipped with attributes such as position, orientation, scale, opacity, and color, while machine learning is employed to calculate the parameters for each Gaussian.

Compared to other rendering methods, like, for example, photogrammetry, a technique for generating 3D polygons from images from multiple viewpoints, or Neural Radiance Field (NeRF), which uses AI to accurately render open scenes and surfaces with reflections or translucent surfaces, 3D Gaussian Splatting requires much less computing power to render a scene in real-time.

This distinction is due to the fact that, although machine learning is necessary for training the Gaussian parameters, the rendering process itself is lightweight and can be executed in real-time without the need for intensive processing.

As is always the case when it comes to novel technology, starting from scratch as a complete beginner and extracting the full benefits of a brand-new rendering approach can pose challenges. Luckily, the team of developers behind Polycam, a company specializing in LiDAR scanning and photogrammetry, has addressed this challenge by making the Gaussian Splatting technique accessible on both the desktop via its website and on mobile through its app.

With Polycam's solution, users have the ability to capture a minimum of 20 images, up to 1000 images, for direct processing within the app. Furthermore, Polycam's website enables the upload of both images and video files. Here's a visual display from Polycam's website that shows the creation process of a Gaussian Splat:

To get the best result out of this technique, your image set should follow photogrammetry best practices, so it is particularly important to ensure that your images or videos are free from any motion blur or shallow depth of field effects. Similar to the principles of photogrammetry, clear, well-lit images will yield the highest-quality reconstruction.

With Polycam's solution, you can seamlessly harness the power of Gaussian Splats to capture intricate environmental details, a task that might be challenging with traditional photogrammetry, especially when dealing with elements like reflections or water. Polycam also offers users the capability to repurpose the raw images from their initial 3D captures, reprocessing them as Gaussian Splats.

Gaussian Splats can be exported from Polycam as PLY files and seamlessly integrated into various 3D software platforms, with ready-made plug-ins for Blender, Unreal Engine, and Unity already in place.

Furthermore, Polycam offers a Splat viewer on their website, allowing users to easily drag and drop their self-trained Gaussian Splat files, whether created within or outside of the Polycam environment.

Below is a series of examples from Polycam Gaussian Splats users, who leveraged the solution to capture landscapes, buildings, objects, people, and more:

You can learn more and get started with Polycam's Gaussian Splatting solution by clicking this link. Also, don't forget to join our 80 Level Talent platform and our Telegram channel, follow us on InstagramTwitter, and LinkedIn, where we share breakdowns, the latest news, awesome artworks, and more.

Join discussion

Comments 0

    You might also like

    We need your consent

    We use cookies on this website to make your browsing experience better. By using the site you agree to our use of cookies.Learn more