You probably have been fooled before. A picture that looked like a photo was not real at all, it was actually a computer generated image. Whether in a movie, a game or a commercial, Photorealism is everywhere. To achieve such a resemblance to life, a couple of points need to be considered. One of the most important ones deals with the level of detail of the 3D objects and textures used - with augmentable and other Remesh tools. A reasonable amount can be achieved through different methods. Whichever one you use, they all provide you with a 3D object that is made up out of a lot of data. This is not only harder to work with, it also results in huge file sizes which take up a lot of space. The resolution is to optimize both the 3D object and the textures associated with it.
Remesh: Why not simply reduce to fewer Polygons?
Any reduction is inevitably associated with losses. In order to achieve a tiny GLB file without compromising on quality, the source information should always be examined. In this case we assume a photorealistic scan, which was calculated by a scan provider from tens of thousands of images. With cutting edge technologies, the "loss" here can be compensated by cleverly implemented workflows. Remesh is not a simple reduction, it is a shape-based reduction that adapts to the curvature of the original scan mesh, reducing in places where not so many polygons are needed. This can be achieved through a combination of different baking techniques. We will report on this in more detail in our next blog post on materialization. The excellent news here is that we have integrated this technique into our platform.
Remesh: Produce GLB - The 3D JPEG with augmentable!
“augmentable“, as the name suggests, is increasable or addable. It gives you the possibility to work faster than ever. The ideal platform for production and quality management.
Work processes can be saved, redistributed, and artists and organizers can work closely together. All this without the need of using additional Chanels or platforms. The perfect all-rounder for an undisturbed creative workflow. You can scan data with very high detail and reduce it’s file size to a couple of Megabytes. The use of the integrated remesh function avoids the loss of quality described above! For this purpose augmentable goes for the smallest, yet most uncompressed standard - for textures this is .PNG and for meshes .GLB.
The result is an "augmentable asset", which means tiny file size with excellent quality.
The surface of a 3D object is made up of multiple meshed polygons, by simplifying this augmentable finds an understanding of the shape, rather than simply calculating down the number of points.
First we start with 300-800MB Large 3D Object + 250MB texture file (4.5 million 3D points + 8K texture) becomes (40.000 points á 5MB + 2x 30 MB texture)
Unfolding" of the object into a 2D representation, in order to then "overlay" the colors (textures)
"overlay" it and 2D space optimization
Transfer of the original colors and details to new textures for the new object.
All these Steps augmentable are done within 5 minutes, so you don't need other programs or skills to Remesh a 3D Asset.
The structure, shape, color, clarity and quality of the surface of the 3D object remains. Upon customer request, augmentable can be used to achieve 3 different file sizes, customized to your needs. This way, a single program is sufficient for processing 3D assets instead of creating elaborate asserts with many different programs.