The project was to make a short VR drama film on memory loss where the viewer dives into the memory of a young woman trying to remember the last moments she shared with her partner.

To illustrate memories slowly appearing and forming around the viewer, I decided to use point-cloud visuals which are vast amounts of small points.
The challenge was to use this technique with live-action performance and real environments instead of building everything in CGI. 
Six different softwares were used through our research to achieve this. Here is the final result as well as a summary breakdown of our process.

Step 1 - Researching the look with a 360 photo
I used photogrammetry to create the point-cloud. This technique allows you to create a 3D model of a real object or environment by taking loads of pictures of it. I tested it with this train station.

1 of the 168 photos taken of the train station.

1st pass of the photogrammetry. We were not looking for a photo-real look so this was perfect for us already.

We then extracted those millions of points and started shaping them to our desired look.

Flat version of 360 capture of the scene.

No CGI was used, this is solely rendered from hundreds of photos of a real environment.

Step 2 - Testing light and human skin in point-cloud with a 360 photo
We built a set to control the light and took hundreds of pictures of it. The result is a vibrant but rather strange scene, which creates the perfect look for this.
Step 3 - Testing video point-cloud
Because of limitations in terms of rendering power, we had to go for a black and white look and dramatically reduce the amount of points we were using. We always had to keep in mind that we had to be able to deliver the film in a downloadable VR experience for audiences and therefore keeping it light. However, it allowed us to play even more on the mystery and confused look of what a deconstructed memory could look like.
Step 3 - Final Look​​​​​​​
Back to Top