Photogrammetry and face topology

During the last leg of the 3D computer graphics course, we students got to look at photogrammetry as a different way of going about creating your 3D models. Photogrammetry is a process whereby you use photos of real objects taken from several different angles and later import them into a software like “123 catch” that it puts all the pieces together and creates you a 3D model of your photographed object. This method of creating 3D models is by no means any less time-consuming, but it is an interesting way to get all the details of realism that our brain takes for granted into 3D models and textures. To get a better understanding of this fascinating subject I suggest you read the article listed below that talks about the results and how photogrammetry was implemented in the game “The Vanishing of Ethan Carter”.

http://www.theastronauts.com/2014/03/visual-revolution-vanishing-ethan-carter/

In class, the students got a chance to photograph their own faces using the method of photogrammetry and later import them to 123 catch that puts all pictures together into a 3D model. There are a few things to keep in mind when taking your photos to make the whole process go by a lot smoother. An advice is to mark at least 4 areas of your face with a marker that you can later use as reference points when compiling your 3D model. Good reference points that may already be on your face are for example birthmarks.

Reference points aren’t the only things to keep in mind though. You’ll also have to make sure that the photos are of good quality and that the object being photographed remains as static as possible during the whole photographing process. Remaining static becomes a problem when photographing faces because the muscles in our face tends to move even though we mean not to. The photogrammetry people in the article linked above solved this by having a set-up of cameras surrounding all around a person’s face that gets good shots from different angles and having them all snap the photos at the same time.

Since I was not present at that time to take photos of my own face, I will show an example of how the photos might look like taken from the different angles. The photos listed below, which I gave myself the liberty of drawing green reference points to show how it might look like, were actually used when creating faces for Spartans in the game Halo 4. You can see some of the set-up of cameras in the background of some photos.

To further develop our photo scanned face models we got to later create the topology on top of them using the software 3DCoat. Working on an real face is great for better understanding the topology of a face that we a later can implement into our organic models that we are doing as our last assignment of the course. Since I didn’t have my own face to put topology on, I used a previous student’s face scan given by my teacher. Below is a screenshot of the topology I made during class using 3DCoat.

The main topology of a face is based around the mouth and eyes with cohesive loops that support the deformation and movements those parts of the face are capable of doing. One thing to note is that if the part you are creating is more rigid, less topology it needs. And if it is supposed carry for a lot of movement, then more it needs. So a basic understanding of anatomy is crucial in creating believable characters.  Below are a few photos to showcase on how the full topology of a face might look like.

If you want to know more about face topology, I suggest you read the article about this subject that can be found here; http://www.phungdinhdung.com/Studies_paper/Realistic_face_modeling

As a conclusion I would like to share my thoughts on the whole photogrammetry for 3D modeling thing. If you have seen the article about photogrammetry, you would have seen the results and the screenshots of the game they made using this method. The results are completely astonishing. I could definitely see myself using this method to create some of my 3D models to really analyze them further and later apply that to my other models.  Just as they say in the article; Assets are no longer simplistic approximations of reality – they are reality. Photogrammetry is unquestionably something I can see the gaming industry using more of in the future, which is one really exciting thing to look forward to.