The Art of VFX

DEADPOOL 2: Stephane Nazé – VFX Supervisor – Framestore

At the beginning of the year, Stephane Nazé explained the work of Framestore on DOWNSIZING. He talks to us today about DEADPOOL 2.

How did you get involved on this show?
At the end of March 2017, Dan (Glass, Overall VFX Supervisor) called me to ask if I was involved on a movie or if I was free; that was the beginning of the adventure.

What was your feeling to be in the Deadpool universe?
I have been incredibly excited, to be part of something so different. The first DEADPOOL is one of my favourite superhero movies, mainly because the film changed the codes for this kind of movie.

How was this collaboration with director David Leitch?
David Leitch is incredibly nice, he has been very open-minded to ideas and suggestions, so the collaboration has been very smooth and constructive.

What was his expectations and approaches about the visual effects?
Initially the brief was to get the majority of things in-camera, with creatures in CG. We had to adapt ourselves quickly, due to edit changes and adjustments of the choreography for the third act; in the end, 80% of the shots for the final big fight were full CG.

You already worked with VFX Supervisor Dan Glass. How was this new show with him?
It makes life much easier when you know each other. DEADPOOL 2 was not my first collaboration with Dan – we collaborated on JUPITER ASCENDING, CLOUD ATLAS and a bit on SPEED RACER as well, therefore I know his level of expectations. On DEADPOOL 2 things were moving quite fast, being a very demanding show, so having worked with Dan before was a definite bonus.

How did you organize the work with your VFX Producer?
DEADPOOL 2 was my first collaboration with Sabrina Gagnon; she produced for BLADE RUNNER 2049 and BEAUTY AND THE BEAST. The production team and the different supervisors (CG Supe Ben Magana, Compositing Supe Romain Rico, Animation Supe Bernd Angerer) were in the same production office, which helped massively for communication, making it fluid and efficient.

How did you split the work amongst the Framestore offices?
DEADPOOL 2 has been a Montreal-only show.

What are the sequences made by Framestore?

Framestore was responsible for the shots featuring Colossus and Juggernaut. It included [spoilers]:

Plus, of course, many smaller sequences starring Colossus and Juggernaut.

How did you work with the art department for the design of Juggernaut?
We got concepts from the client art department, and decided to quickly move forward in CG to figure out all of the questions around proportions; also because the plan for the client was to build a real-size helmet as a reference for the shoot. Each of the concepts had something very interesting, but we didn’t land on one picture as the ‘ultimate’ reference. The design was still evolving, and different questions like, ‘how loose or tight should the yellow costume be? », or, ‘how dry, or fat, does he need to be?’ were still not solved.

We also did internal concepts to determine how damaged the costume would look, and the progression of the costume throughout the sequences.

Beside the comics, did you receive any specific indications and references for Juggernaut?
For the client, the main idea was to create a contrast between the military aspect of Colossus and the more uncontrollable brute force of Juggernaut.

Initially, the brief was that we wouldn’t see his skin, and that he wouldn’t speak, in order to help the asset build. He had to be a brute force, to act before he thinks, to be unstoppable. In the movie, however, you see his skin all over the place and he has dialogues. Framestore came quite early to the process with this idea of Juggernaut acting like a wrestler.

How did you enhance the Colossus asset made by Digital Domain on the first movie?
We received the model from Digital Domain as a base, which was great! The concept for DEADPOOL 2 was quite different. The client brief was to keep the original head, and to change everything else. We started by changing the proportion, to make him more athletic.

One of the biggest changes with the new design was to make the muscles very sharp, more ‘salient’. The other big change was to use the horizontal feature lines to split the geometry into different objects. It was quite challenging to do, but it has been incredibly useful to help to keep maximum rigidity in the skin. We were able to play with the distance between the objects to simulate the elasticity and also the rotation, to maintain maximum rigidity for the different pieces.

Can you tell us more about the rigging of Juggernaut and Colossus?
Colossus is an athletic character. A lot more skin and anatomy detail were added directly onto the model on this version. Colossus’ rigging challenges included finding a good balance between the metal rigidity of the skin and an organic, anatomical feeling. This was done using our proprietary muscle system, primarily. On top of that we procedurally created controls to each part of the body to rigidify the skin simulation. Offset controls were added per plate to create slow motion hit effects, or better outline of the arms.

Juggernaut was a more straightforward character for Framestore. We built a muscle and skin system. We spent some time on the arm and shoulder anatomy behaviour, matching some bodybuilder references. Juggernaut’s collar is also part of his design. It restrains him, and had to feel tight with the suit. In our simulation process we added an additional pass of collision between the collar and the anatomy before the skin simulation, giving us better results than just collision on the suit.

How did you create the various textures and shaders?
The stunt for Colossus used a big metal helmet. This was useful on set to figure out the height of the character, but also to mimic internally the shader. For each shot we got a good reference of how the metal was reacting in context. Of course, this piece of metal was approved by the client before proceeding with the shoot.

For Juggernaut, we used the yellow suit of the other prisoners in the convoy as a reference.

For Colossus and Juggernaut, we received from the costume department a piece of fabric for each of their parts: the pants, shoes, belt. This was very useful to analyse how the light was reacting, to adjust the spec and to get a very precise shader.

How does the shiny aspect of Colossus affects your work?
Because Colossus is reflective, we had to build everything around him. We had to roughly body track everything in motion to get a sense of movement in the reflection, and also to get a real sense of depth. Basically when Colossus was in the framing, almost everything around him was done in CG. In some shots we had to put in almost all of the digidoubles interacting with him, to render an accurate reflection.

Can you explain in detail about the face animation of Colossus?
Again, Colossus is a tricky character because he is a metal guy. Metal is stiff, and should not deform smoothly where everything is connected. For the face we decided to go with more localised blendShape – that way, animation could control how much fall-off they wanted to get the required feeling of rigidity. All of the blendShape were created based on a mocap session, to help the realism.

How did you share the asset with Method Studios?
We shared mainly the model and the textures; everything else – the rigging system, the muscle system, shaders – were proprietary softwares. Of course, we sent turntables with different lighting rigs and different HDRI files to help the match. We sent updates throughout the creative process, to help us move forward together.

The movie is really gory. Can you explain in detail about your work when Deadpool is cut in two by Juggernaut?
This sequence was possibly one of the most funny to review in dailies 🙂 And sometimes one of the most disgusting to review with the team. I won’t forget the day the FX guys came into dailies with references from the internet of guys having been shredded…
We built all of the entrails and organs, and we added a lot of blood – the client asked for it to be over-the-top, for it not to be too realistic.

Colossus and Juggernaut have a big fight at the end. How did you approach this fight?
It started with a very long process of previs; The Third Floor was involved to help with the edit and to choreograph the whole thing. Framestore had been involved as well, to help with the key shots and to find solutions for the key moment.

We could split the process into three main steps:

Did you receive specific indications for the fight choreographies?
We had very specific briefs for the fight but the process became quite organic, due to the edit changes and the modifications linked to it.

What was the main challenge with this fight?
The third act sequence is split in two actions: what’s happening inside and outside of the orphanage. It meant some intercuts, and one of the challenges was to not become lost or confused by the action.
However the main challenge was to feel the progression of light during the fight (the light going down), and the amount of damage on the environment. The damage is also on the characters – there is more and more dust on Colossus, and on Juggernaut more and more damage to his costume and his helmet, more bullet holes – therefore the continuity was a real challenge too.

How did you recreate this Orphanage environment?
We created the orphanage environment and all of its components – the main building, the pool, the gymnasium and the trees – in one month. We took a Lidar for the set, which was a good base on which to construct a solid geometry. For the textures, we got a solid photo shoot.

The fight causes a lot of destruction and ends in a pool. Can you explain in details about those various FX simulations?
Most of the broken windows and flames are 2D; the big final shot, with the heroes leaving the framing in a dusk mood with the building on fire, is comp only.

Because there is a lot of interaction with the characters, all of the damage, smoke and debris on the ground is CG.

For the action in the pool, we decided to paint out the water in the plate and to replace it with CG, to better match with the timing requested and to fit better with the choreography (initially the big splash in the pool was in-camera). The big difficulty was to simulate the water with a timing locked so late; this part of the third act was a real race against time.

Which sequence or shot was the most complicated to create and why?
The third act sequence; because the fight had been locked six weeks before the delivery, we had to build the environment in CG three months before the end. There were a lot of fairly heavy things to manage in a short amount of time.

Is there something specific that gives you some really short nights?
Some shots using full CG Deadpool changed at the very end, to use the head from the plate, to be closer to Ryan Reynold’s performance.

What is your favorite shot or sequence?
My favourite sequence is the third act sequence, the part where Colossus and Juggernaut are fighting each other, mostly because we had great creative control on this part and a big input.

What is your best memory on this show?
There are so many good ones, but without a doubt I would say the collaboration with the team. The Framestore team has been very involved and dedicated, and the vibe on the floor was great.

How long have you worked on this show?
May 2017-April 2018.

What’s the VFX shots count?
245 shots, including 16 shots for the Director`s cut.

What was the size of your team?
189 people in total.

What is your next project?
ARTEMIS FOWL.

A big thanks for your time.

// WANT TO KNOW MORE?

Framestore: Dedicated page about DEADPOOL 2 on Framestore website.





© Vincent Frei – The Art of VFX – 2018

Quitter la version mobile