Week 3

This week I used Metahuman for the first time. At the start of class we just had a practice at setting up Metahuman and customising a premade metahuman. It was cool being able to change things like the hair, makeup and the facial structure. I also played around with different lighting to see how my metahuman would look with red lighting.

I also went with my group to our photogrammetry session. Since I am the actor for our assignment I am also the one that will be used to make the metahuman model. I had to wear a cap with dots on it and sit in front of 9 cameras connected into the computer. Photos were taken of me from different angles and then the photos were put into a software called Reality Capture where a 3D model was made of my head and a texture from the photos was made and applied.

The nine cameras I was facing

Some of the photos taken of me

3D model with textures applied

My 3D model head in Reality Capture

The texture sheet made from the photos

Once I had the files of my 3D head, I was able to import the OBJ and texture files into Unreal Engine after installing a Metahuman plugin. The process was easy enough following the PDF tutorial Greg made for us. I created a new material with the textures and applied it to the mesh of the head and changed the lighting so it was visible. Then I had to add a Metahuman identity, assign a body and add tracking markers. Then meshed it to a Metahuman.

I was then able to open the metahuman and add skin, hair and other features. After first importing I noticed the head was lumpy from the cap I wore but I was able to smooth it out with the sliders.

It's a bit strange seeing myself as a Metahuman. It's very uncanny. It looks like me but at the same time it looks slightly different too. The hair is not how my hair really looks but to me it was the only hairstyle that actually made it look like me. The eyebrow options also don't really match my real eyebrows because I have quite dark, full eyebrows but the Metahuman options are quite light and not as dense.

This week I also did some ROM practice in Analyzer, since I used it last year I had a good idea of what to do but still needed a few reminders so I watched Dr Kennedy's lecture. I did tracking of the eyes, brows and nose using videos of Dr Kennedy making different movements with his eyes and brows. Once the markers were successfully tracking, I exported the tracking data and training frames. Then I was able to import them onto another video footage of a small facial performance Dr Kennedy did. Once the data was imported it was quite simple, I assigned the markers on the neutral frame and pressed track, using the imported tracking data the markers were able to assign themselves to where they should be and tracked pretty well. There was a bit of jitter and some points moving around but considering how quick it was to apply to the footage I think it's quite good and the cleanup that would be needed after is minimal.