
I Ditched 2D Photos for 3D Gaussian Splatting: Is This the Future of Memories?
- Technology
- 17 Aug, 2026
Let's be honest for a second. We all take thousands of photos on our smartphones every year, but how many of them actually capture the feeling of being in that moment? A 2D photograph is just a flat representation—a frozen slice of reality.
For me, that realization hit hard during a recent family gathering. I was taking photos, but I knew I wasn't really capturing the depth of the room or the spatial dynamics of the moment. So, I decided to try something completely different. I stopped taking traditional flat photos and started capturing memories using 3D Gaussian Splatting directly on my smartphone.
Here is what it is actually like to step away from traditional photography and start saving your life in full 3D in 2026.
What is 3D Gaussian Splatting and Why Does It Matter?
If you haven't been keeping up with the bleeding edge of computer vision, 3D Gaussian Splatting (often just called "Splatting") is a technique for rendering highly detailed, photorealistic 3D scenes. Unlike traditional photogrammetry, which creates a clunky 3D mesh and pastes textures onto it, Gaussian Splatting represents a scene using millions of microscopic, semi-transparent blobs (the "Gaussians").
The result? It captures specular highlights, reflections, and fine details like hair or foliage with mind-bending accuracy. And the best part? In 2026, the processing hardware on modern smartphones has gotten so powerful that you can compute these splats almost in real-time, right on your device.
The Reality of Capturing 3D Scans on a Phone
To capture a scene, I just slowly pan my phone around an object or a room, similar to taking a panoramic photo, but moving in a short arc. The phone's spatial sensors and camera array do the heavy lifting, grabbing visual data from slightly different angles.
The Good: The sheer fidelity is astonishing. When I captured my living room, the resulting 3D splat didn't just look like a photo; it felt like a tangible space. I can look back at the file on my spatial display (or simply swipe around it on my phone screen), and the lighting and reflections shift realistically as I change my viewing angle. It is the closest thing to a "pensieve" from Harry Potter that I've ever experienced.
The Bad: It requires a shift in mindset. You can't just quickly snap a photo of a moving target. If my dog runs through the frame while I'm capturing a splat, the result is a blurry, ghostly smear in the 3D space. Gaussian Splatting currently excels at static scenes—landscapes, rooms, still objects, or very patient people.
Storage and Sharing: The Elephant in the Room
One of the biggest hurdles I encountered was data management. A high-quality Gaussian Splat file can be massive. While compression algorithms have improved significantly, capturing a dozen 3D memories a day will chew through your phone's storage at an alarming rate.
Furthermore, sharing these memories isn't as simple as texting a JPEG. While major social platforms have started supporting interactive 3D embeds, sending a splat to a family member often means they have to view it through a specific web viewer, or they only get a pre-rendered 2D video fly-through of the 3D space.
Is It Time to Ditch 2D Photos?
So, am I completely done with regular photography? No. The speed and convenience of a point-and-shoot 2D camera are still essential for capturing fleeting moments.
However, for monumental life events—a new house, a beautifully decorated wedding venue, or simply the exact state of my child's messy bedroom—3D Gaussian Splatting is my new default. It doesn't just capture a picture; it captures the geometry and lighting of a memory. The ability to spatially revisit a place exactly as it was is an incredibly powerful emotional experience, and the fact that we can now do this with the devices in our pockets is nothing short of revolutionary.



























































































































































































































