Self Portrait - Digi Double

Here are the different  shots of my digital double. This project involves many steps.  I explain and show you all the process below!

TURN AND BREAKDOWNS

The Process

Firstly, we shot our faces from multiple angles with a soft light and neutral expression for the photogrammetry. I converted the images into a 3D scan of my head with Agisoft Photoscan. Then, I used R3DS Wrap to do the retopology, which helped me to create a good base for the next steps.

After that, I spent a lot of time in Maya, switching between modeling and sculpting tools to completely rework my face. I paid attention to details such as the eyes and mouth, using anatomy references and the pictures of the shooting in order to have a face that looks like me. I also used Mudbox to sculpt some details that I couldn't achieve through modeling, such as wrinkles and spots, and generated a first displacement map. I returned to Mudbox later to refine the displacement once I had the albedo. In Mari, I worked on the micro displacement, bump, and roughness details with the Texturing XYZ maps, which use RGB channel informations.

The roughness part was particularly challenging because I had to generate two kinds of roughness: the rough specular with micro variations of the skin, such as pores and wrinkles, and the primary specular with stronger variations like on the lips and cheeks.

( Direct specular / Direct Rough Specular / Direct Primary Specular )

To create the albedo map, we took photos of our skins with a polarizing filter to capture color details like freckles and beauty marks. I then used Photoshop to remove shadows and blend the different images to generate the albedo map, which I projected on my model in Mari. Finally, in Mudbox, I added and accentuated some details. To generate the SSS Mean free Path Color I just increased a lot the saturation of the albedo. And I painted a mask of SSS to create variations, such as more SSS on the ears and mouth and less on the cheeks.

For shading, I used Pixar's Renderman and started with a neutral 3 points lighting setup to establish good parameters. I then spent a lot of time testing different values for each parameter and going back and forth between the different softwares. It was tricky but very important to find a balance between every settings. I also studied a lot of references and documentation to achieve this level of detail.

I spent a lot of time on grooming, paying attention to details and observing real-life examples. I used Xgen interactive in Maya to create hair, eyebrows, eyelashes, and peach fuzz and I shade it in pixar's renderman mainly in procedural.

Moreover, I used the MetaHuman workflow by Unreal Ungine to add a light expression and to make my face look more alive. I also stylised my portrait with makeup and accesories.

Finally, this digidouble was also a good exercise of lighting and compositing which allowed me to create also more creative shots.

Overall, this project was a very fun and challenging exercise which helped me to develop my technical and artistic skills!