Hullo everybody !
Here is the Lighting Club #05 !! You probably already heard about our new topic cause this film is actually in theatre : Minuscule 2 !
To present a bit more Minuscule, it’s an ward-winning series of French animated shorts exploring all things insects and bugs, created by Hélène Giraud and Thomas Szabo. The first episode of the first season has been released in 2006. Then in 2013, Minuscule: Valley of the Lost Ants, a feature film mixing live action and cartoon animation is born. With more than 22.8 million USD box office, it was obvious to see a second opus : Minuscule – Mandibles From Far Away. The lighting work on it is really impressive and we can imagine how it was challenging to create this mix between cartoon and live action. We were lucky enough to ask a few questions to Jean-Michel Bihorel, Lighting and Lookdev supervisor on Minuscule 2 at The Yard !
Hello Jean-Michel, thanks for answering our questions, could you talk a bit about you please ?
I started CGI when I was 15 years old, 15 years ago, almost an half life time ! I work in this industry since 2008, I started in advertising company for a few years. In my mind, advertising company are great ‘’school’’ to learn a lot, you got both reduced timing and reduced crew. Thus it decides me to be CG Generalist. Then I worked a few years in print industry, which refine my attention to details. In the meantime I always worked on personal projects. My visual universe started to come alive but The Yard proposed to me the Minuscule 2 lighting and lookdev supervision. The promising experience was so big I couldn’t refuse and then I accepted. In parallel, as if it wasn’t enough, with some friends, we created a 3D School called Creative Seeds. This is now the second year and we are proud of our students.
How many people were in the lighting, the lookdev and the compositing crew ?
We were not too many people on this project. Concerning the lookdev we were around 6 people globally, and when we started the lighting we have grown around 9 people. Lookdev TD were lighting td as well and more or less this team made the compositing too. In maximum we were around 20 artists between lighting and compositing. Obviously without talking about modeling, fx, support etc…
Could you talk about Minuscule 2 and the challenges ?
Minuscule 2 is, as you know, the second opus. In the first movie and in the tv show, an artistic direction mixing live action and cgi takes place. The main goal of Minuscule 2 was to increase the quality and stay in the meantime faithful to this proper artistic direction. We couldn’t revolutionizing the style but we tried to refine what we could.
We had to create more than 700 shots in different sequences. In these shots, 30 minutes were full CG sequences and the challenge was to not feel the difference with the live action shots in the edit. If we except some shots with an obvious CGI camera movement, I think everything is working pretty good.
Another challenge was the size of the filmed elements. We were always really closed to the things and we had to generate a accurate texturing/lookdev to get enough details.
The sequence in the clouds has been especially tricky and took time to create. These shots take place inside a clouds corridor. It was tricky enough to find a nice lighting and to not cheat to much to stay a minimum physically accurate. We had to keep the interaction between the boat and clouds.
Minuscule has a strong artistic direction, cartoon plus live action, what does it mean for you on set and in lighting/post production ?
Juggling between cartoon and vfx is not simple. We had the live constraint, but sometimes we had to manage lighting to keep the cartoon “appeal”. The shooting was classic enough, with tracking, HDR and photogrammetry…
Concerning the lighting, sometimes we didn’t used our HDRI taken on set cause the artistic direction needed more control to keep the look of the characters.
Did you worked the lighting more as a live action or more like a animation feature film ?
The lighting as been treated more like a VFX feature film I could say. Even if we didn’t always used the HDRI taken on set, we obviously had to light the shots depending of the plate. Each frame was really different, we could not use same lightrigg and we had to tweak shot by shot to improve the integration.
What was the renderer software for this show and why ? Did you reused asset/set from the first Minuscule ?
For Minuscule 2 we worked with Guerrilla Render. With Guerilla Station we had a ‘’Katana like’’ solution but cheaper, easier and user friendly. We had the chance to share our studio with Guerilla developer, which was a privileged situation for our dev request.
Guerilla Render structure allows us to solve important pipeline points as scene builder or share lightrigg easily. It is a powerful software and it contains a procedural tool which we generate the groom on our characters and all the instance for grounds close up.
We reused an asset library from Minuscule 1 but we had to tweak them to obtain the level of details we wanted and to transfer the lookdev on Guerilla.
What was the main challenge in terms of lighting and compositing ?
It may seem trivial, but what we had to tweak the more depending of the director feedbacks were the ladybugs elytra. We used to integrate them a bit to much in a photo-realistic way and thus loose the cartoon vision the director had for those characters. So it was paradoxically about this point we had the more goings and comings, in contrary of full cgi sequences which were ‘’simpler’’.
A specific sequence you wanted to talk about ?
Except the clouds sequence which was hard with huge render time, the caterpillars nest was challenging as well. It was an exclusively silk environment and it was almost impossible to get a soft scatter result without enormous render time. Even using “brute force” method, the look was not enough soft for directors. We decided to create a two times lighting workflow. Firstly we generated an object : the wire position. Inside this object we filled with a volumetric shader really dense and we could work on our first lighting. With this lighting we could render a first pass with some AOVs. Then the FX artists generated wires on the geometry we provided to them. In the meantime they transferred the lighting data in the points curves, we used these informations for the final render like a multi camera mapping. We get a scattering based on a volume, but everything was rendered on proper curves.And finally a good amount of compositing for the caterpillars interaction cause this workflow does not allow a dynamic lighting
A lot of different set, different type of lighting, how did you proceed ?
Indeed ! As I said earlier, the lighting is completely different on each shot. We had to work shot by shot and we could not use Keylight-rigg. All the lighting td/artists should be proud to make their own keylights, even the juniors ! My work was to maintain the consistency between lighting artists even if we tried to give similar shots to same people.
Concerning the lookdev, any asset a bit tricky you want to talk about ? Why ?
The lookdev has not been a huge problem for us. There is however an interesting point to emphasize : as the size of object where the action take place, characters had to be with a strong SSS, not far away from translucent sometimes (the praying mantis for example). Thus we used volumetric shaders instead of classic SSS. This allows us to get a more natural light transmission and in the meantime to keep transparency aspect of the objects.
What could wish you for the future ? A Minuscule 3 incoming maybe ?
We hope to work on another feature film soon. No Minuscule 3 for us for now but business to follow !
A last question which is not concerning Minuscule at all but as we are lucky enough to interview you and as we were really impressed by this project, we have to ask you : Could you talk a bit about Intimate Cities – Old Paris Underwater and the lighting of this personal project you created ?
With “Intimate Cities” I wanted to play with the perception we have from our surrounding based on the lighting or other atmospheric effects. I wanted to convey a feeling of intimacy by revealing only small parts of the city, hiding the rest of it in the fog. I also lit my scenes using very near light sources instead of distant lights in order to reduce even more the feeling of scale of the environment.
As a result we always feel close from the subject and we almost forget that we are in open streets. It creates a unexpected bubble of intimacy inside an urban area.
At the beginning I was planning to build the models using VR oculus medium, but I ended up using more 3d scans than actual modeling or sculpting. This project is a mix of 3d scan, Vr modeling and procedural modeling. It was fun to experiment all these different workflows.
A huge thanks for your time Jean-Michel and welcome in the Lighting Club ! Anything to add ?
Thanks for this interview. It was a pleasure to share about our work on Minuscule 2. All the best with the Lighting Club !