Case Study of Logan, by Image Engine

Part 1- Introduction

My name is Cheri Fojtik and I'm the current texture lead at Image Engine Design in Vancouver, BC.
I have been a texture artist in the visual effects industry for nine years.

I started as a Junior Texture Artist at Rhythm and Hues, where I worked on films such as Life of Pi and Snow White and the Huntsman.
I then went on to work at ReelFx and Industrial Light and Magic, working on The Book of Life and Tomorrowland, respectively. I’ve been in my current role at Image Engine for over three years; projects I’ve worked on include Fantastic Beasts and Where to Find Them, Independence Day: Resurgence, Game of Thrones, The X-Files and of course Logan.

I was primarily responsible for the texture process for Logan's and X24's digital doubles. I was also involved in leading the texture team as we worked on all the assets needed for the creation of the visual effects for this film.

Logan Texturing Breakdown from Cheri Fojtik, Image Engine (video credit from Foundry)

Logan, also known as Wolverine, is the Marvel character portrayed most recently by Hugh Jackman. This movie depicted the end of this saga, which also included a clone of Wolverine, known as X24.

Part 2- Logan Production

A lot of teamwork went into the making of these two characters. We started developing these characters pretty early and the primary work was developed over the course of a year across all departments. Various tweaks and updates were then made as needed based on the development of the shots.

We were fortunate to receive ICT scans for both Logan and X24. This package included accurate color, displacement and specular maps, and several reference photos with camera information. This data was invaluable for getting the smallest detail perfect by lining up real world reference to the CG model.

Comparison between materials from ICT and Image Engine

Once the model was ready, we began the process of transferring the scan data via Mudbox. This process took several iterations as the model progressed to have a more accurate lineup.

The more precise was the projection, the more realism was achieved.

We added our own pass of custom painted micromaps for the head.
Mari was the primary tool for texture painting and projecting.
Zbrush and Mudbox were used for sculpting and Gaffer, our proprietary lookdev software, for lookdev.
We rendered using Arnold.

We used TexturingXYZ maps for the body and touch-ups for the face and head:

For the body and extremities, we used:

For the face touch ups we primarily used:

We needed good face maps for touch ups that gave us the flexibility to mold them how we needed. The male 40's face package and the female 60's face package included wrinkles and pore detail similar to that of the scan data received from ICT.

The high quality of the the TexturingXYZ maps made it easy to grade them to match the scan data, and easily blend them in with the scan displacement to fill in the gaps and missing detail as needed.

The detail we can get from these maps is very crisp and clean.
There are subtleties that would be painstaking to paint manually and would still not achieve the same level of realism.
Additionally, in the always changing workflows and time constraints of visual effects, it is great to have maps that are as high quality as the TexturingXYZ packages that can help you achieve better results more quickly.

[widgetic-before-and-after id="5c41ae5decb2a1d62e8b4567" autoscale=on width="2048" height="1002" autoscale="on" resize=fill-width]

Tips:
Combining my passes into a multi-channel exr saved my sanity and a lot of time, i.e, the three displacement passes, (main, fine, micro) were copied into different channels of the exr.
Then I was able to projection paint with one map, using the copy channel function in Mari to extract the data I needed for that particular pass.
I've done traditional digidouble painting in the past where I had to manually line up detail across passes.
It takes quite a bit of time and it never looks nearly as accurate or realistic in the end.

Part 3- Aiming beyond realism

With Logan, we were presented with a challenge of two hero digi-doubles that were the same actor but not the same character.
And with that included multiple outfits and several damage states.
That being said, organization was a huge factor in keeping this all straight, especially since the entire texture team here at Image Engine worked on various aspects of the characters at one point or another.

With the skin details and projections it was particularly challenging to keep up with what needed to be shared across states versus which character. For example: X24 and Logan each had their own ICT scan data. Logan was a bit more haggard with scars and X24, had cleaner, younger skin. I had to set a base to make each unique, but then as I touched up other things, like the ears, eyes and rest of the face, I could share that data across not just multiple passes, but also each character.
This eliminated the need of doing the same work twice! It was a lot of management but it was worth it in the end.

The biggest challenge in terms of painting involved adding in all the missing details so the full takeovers would match.

We studied the CG version vs the live reference version with such scrutiny that I hand painted in missing details of lumps in the forehead, deepening wrinkles where they didn't quite match, and adding a missing scar that the scan data didn't pick up on because it was too close to an eyebrow.

There were several iterations of these type of changes to get them just right.

[widgetic-before-and-after id="5c41b136ecb2a10a2f8b4567" autoscale=on width="2048" height="1002" autoscale="on" resize=fill-width]

Total shot count was 299. Total CG head count 51 shots, but some had multiple heads. Total 63 head replacements; 10 full digidouble shots.

Q: What was your typical day at Image-Engine while working on Logan?

A: The usual day was quite varied depending on where we were in the process. Typically I would come in, check renders, look for notes from supervisors first thing in the morning and chat with my lookdev artist. We would have meetings with the visual effects supervisor, Martyn Culpitt, for feedback.
We would wait until any model updates were done, then I would talk to my lookdev artist about how we should tackle notes or some other details we wanted to address. I would then work on texture updates, usually in several iterations until it was just right. I would then publish my textures for the lookdev artist to incorporate, render turntables and then repeat!

Q: Anything you would like to add?

A: A big shout out to the entire team here at Image Engine that worked on these assets. The work is what it is because every person at every stage showed they cared by going above and beyond. This includes all the members of the team who didn't work directly on the film; it takes a lot of support from everyone in the company to create something incredible.
It takes a huge number of talented artists to create amazing visual effects and texturing is only a small part of it. I'm extremely proud to be a part of this entire process at Image Engine.

From everyone at Texturingxyz, we would like to thanks Cheri and everyone involved in this article.