Jingle Jangle

The VFX magic of Jingle Jangle, with Framestore Montreal

3DVF: Buddy 3000 is another memorable character from the movie. This small robot has big eyes and a glowing heart. These two elements are very important to make him alive, especially since he has no dialogue. What were the expectations of the director David E. Talbert and the VFX Supervisor Carlos Monzon for this character ? What were the main challenges ?

For sure Buddy has history behind him ! Loads of concepts and tests were made. The director and our VFX supp really put a point of importance on his eyes and feelings we could read through them.

At the beginning, Buddy’s eyes were empty, it was mainly a tube going back into the depth of his head. This was really fun in terms of animation, to get those BIG huge black pupils, he felt a lot like a cute cat all the time, but he was also feeling very high sometimes, you know… The main challenge was to find a way of adding emotions to those big black pupils. After looking at other robots, one day I took a random binoculars asset we have at Framestore in our library and put it inside, so we could feel that there’s a mechanism inside, and which could then be our new pupils, exactly like a camera works when you zoom or unzoom, you see all the lenses moving around, this got us into a discussion with the modelling and look development department, and when this was proposed to the director (after our VFX supervisor Carlos Monzon approved the whole remake over), he loved it, and for sure it fixed a lot of issues in terms of feelings, we could even now add direction to where Buddy was looking, but it created some challenges for us in lighting and rendering !

As you mentioned, eyes and heart were this asset’s challenges.

First of all, we have a lot more close ups of Buddy than Don Juan, so that means render a 2cm thick glass lens with dirt and reflection / refraction that take a big percentage of the screen, and with some inside reflections going on for those new lenses (we had a separate lightrig from the one to light the shot, just to treat and control what was reflecting onto those new inside lenses). Another challenge with his eyes was to render physically accurate caustics, and that’s complicated to render. But damn, so beautiful ! It’s the glass aspect of the eyes that got technical, making sure to render its eyes clean, without noise, visible on more than 70% of the screen frame, but not eating too much rendering ressources (CPUs), was the challenge. We ended up splitting a lot of elements, like we had a default looking eyes rendered with the glass, just the eyes (we had an AOV to control only the reflection of the glass) , no body, then our caustic pass was being trace through the glass but we were not rendering the actual reflection of the glass, the inside lenses setup was also being rendered on another splitted pass/layer. In some shots we even had another pass to add some specular hits of specific light reflection, again with a separate lightrig, to match the set reference even closer but making it look better, like we would not reflect a simple square area light, we would had a texture to it to make it look like a window, or add some break ups so it just look nice, but not square or circle perfect like a light. If we compare, Don Juan was only one beauty pass, for Buddy, just for his eye look we were rendering 3 or 4 passes.  

For the heart, we have less close ups on his heart than on his face, so that was a relief for us, but we still had to figure out how to deal with the ones we had (which are awesome dope looking shots to be honest). Main issue is that we are again dealing with refraction ! But even worse now, we have tons of little gears, each one rotating so then, creating motion blur, with a self illuminated asset for the filament of the light that’s lighting the whole heart, which is itself, built inside a glass tube for the light bulb. A huge part of how awesome it’s looking is due to the build of that heart (rigging, modelling) and the look development team. For most of the shots, as the heart is not too big in frame, we rendered it with the body, including an AOV for the glass reflection and it was working perfectly. Also in his default setup, we had a volume box constraint to his heart so we could get a nice glow rendered. But when you start to zoom in and get a close up, we had to split it out as well from the body so we could treat it separately, why ? Because there was no way for us to get a clean render without noise in the motion blur of those fast rotating gears without exploding our render farm, As Buddy’s body was quite fast to render, it’s only metal right ? By splitting it we could use the minimum sampling needed for the body and make sure that all the sampling we were putting onto the heart would be redistributed ONLY for the heart, as the body was resolved in a separate pass/ layer. – lets not forget that we are not alone on the farm and several shows and in need of those resources as well. The shot inside the box where Buddy is heating up crazy when escaping the tunnel was a big challenge for us and we had to split everything to save rendering ressources. So we ended up with four passes for the eyes, one pass for the body, two for the heart – plus all the others elements to render (FX sparks and smoke – interaction on the box ). And i’m talking beauty passes only, you can add a utility one for each.

3DVF: The tunnel sequence is quite complex and involves a fully-CG environment, fully digital kids, and a big explosion. Can you tell us how you worked on this thrilling escape?

The challenge here was to work in an almost closed environment, which was covered by a quite reflective dark ceramic, water puddles at the bottom and add inside of that environment a huge chasing fireball, that would illuminate and reflect everywhere. Add to that the awesome digi double lookdev that includes scattering shaders and fuzzy hair and the very few windows opening we had, so the tunnel is actually light by only a few direct rays, most of it is indirect – This is a great recipe for rendering noise and headaches. I think one huge challenge was handled by the rigging and layout department as before everything, they had to lock the tunnel shape and camera movement before, which was not an easy task. Once that was done (we started lighting during that process for time management issues so it got painful for the artists – when you have to replace your light three days in a row because the tunnel shape changed everyday, that was frustrating sometimes), we could put our lights where the actual modelled windows were (to match the portion of tunnel they had built for real on set). Thats a sequence with beautiful lighting but a simple one ! An elaborated combination of lights on the right and left side of the tunnel where the windows were did the trick, then Carlos Monzon directed us a lot regarding the hue and exposure of our lights, he had a very clear vision of that sequence and this helped us so much nailing it so fast. Which was a very good thing as we could then focus on denoising it. The whole team got on board and we blasted the farm with sampling test. How to optimise our use of the huge fireball ? In the shots we’re still using some bits of the plate and we are just doing an extension of it in CG, what can we spare to render ? We used a lot the render mask technic, which basically render a region of your frame you specify with an animated (or not) texture with alpha, so for the shots we were doing an extension of the real set only (all the shots at the beginning of the seq before the explosion) we were not rendering the area where we kept the tunnel from the plate and we were providing compositing with some customs reflections passes they used to integrate better with the plate and make it look sexier all together. For the fireball, FX did a great job on the simulation and providing us with something easy to work with, it was not too heavy to handle – The challenge here was to make sure we could denoise it’s reflection on the tunnel – In order to manage that we had to split it out from the default tunnel render. So we had a green tunnel render on top of which we added a fire reflection only pass, this was also to give compositing more control for the final look. The motion blur got us in trouble as well as some camera movements are quite extreme or just going very fast ! A dark environment that reflects something bright, detailed dark fuzzy hair that also catches light and moves fast ; the motion blur didn’t like that and was very noisy. In some shots we could sample it better to improve the look, but in some others we resolved that by rendering without motion blur and providing compositing with motion vectors so they could do the motion blur themselves – this couldn’t work in shots with a complex camera movement, it had to be quite linear. One shot that got rendered like that is the one in the tunnel we are behind an aeration vent, very close to the floor, kids coming from the left to the right being chased by the fireball.

One last challenge was the need in some shots to keep the kids from the plate, with a quite simple lighting on set, and integrate them in a fully CG environment in which the lighting could feel detached from the kids, we had to adapt our lighting to those plate kids for those shots but still adding some richness to it to make it look better and believable. This sequence is the result of a lot of discussions between lighting and compositing, a lot of compromises and workarounds from each side and a real sense of “we’re working on this together” .

3DVF: Last, but not least, the last shot of the movie features the toy factory and the town below: a beautiful shot with lots of lights in the streets and buildings. How did you handle this ?

Actually in lighting this was ok ! As i’ve mentioned previously, we had an environment team working on the show, and they handled all those village shots. In this last shot lighting actually rendered the falling snow and Buddy. The Toy Factory lightRig was designed by a senior lighting artist in the team and he was in charge of it all along the production, but the environment team handled the rendering of it and the village. Steven Donnet was the lead in charge of the town.

3DVF: During the making of this show, you had to deal with an unprecedented event: the Covid-19 pandemic. What actions were taken, both to ensure the safety of the Framestore crew, and to keep working on the show?

Very quickly the system team had to find a solution as people stopped coming to work quite fast – Myself I decided on a Thursday that I won’t risk it anymore and Framestore actually asked everyone to stay home the next Monday – It took them a week to deliver our screens and mouse/keyboards to everyone in the company + sending out a secure router. They did astonishing work, working restlessly day and night to work on the secure network – those guys truly impressed me and I’ll be forever grateful. And they are still working on it so we can have better experience and performances with the network everyday. I can now connect to my physical machine at work, from my living room! Also the clients were understanding and I know they are all very happy with how Framestore handled it, they didn’t feel the change in productivity and deliveries, so that’s great !

3DVF: As a lead lighting TD, communication is a key part of your work: was it hard to adapt to these new constraints?

Do you realise how much you are RIGHT about communication? IT’s THE KEY. And what could be frustrating sometimes, is that you can’t just get up, go to the person you need to speak, and be done with the subject in hopefully, a matter of minutes, and smiles. And you create a nice dynamic as well with the entire team, not just your lighting one. I wouldn’t say it’s hard, but it changed. Maybe I’m still discovering as well. Now, for every little detail, that would just need a simple visit to someone turns into a meeting, a visioconference. Some are necessary, some others, it’s way easier to talk about it than writing in a chat window, but it’s still much longer and energy consuming than a quick chat in person. (due to the fact that everything is slower in visio as you can’t talk at the same time as someone, there’s always a lag, not everyone has a great connection so sometimes you also just can’t get what the person is saying – the synergy is not the same). Also for review I had to make sure I could trust the colors i was seeing as it’s been years i’m not using my personal laptop on a daily basis. IT also pushes you to have better explanations as you can’t just get up, and finger points on the screen you know, especially when a junior asks you a question, it’s not pointing anymore – most of the time I’d just share my screen through a secure channel and spoint with my mouse. But again, time consuming ! I was lucky enough to have my entire team complete early march, so we already had a team life before saying goodbye. I’m really curious about starting my next show and organising that !

3DVF: Overall, what are you the most proud of on this show?

I’m very proud of the team spirit we build, this was one of my best career experience, and the work with our production and CG supervisor Britton Plewes and VFX supervisor Carlos Monzon was also very smooth and respectful and everyone was involved, probably because we all started the show in the building together, so we were not strange face on a screen to each other, but I truly like how natural it felt to work all together. This is definitely my golden point. In the movie itself, I’m amazed by how good Buddy is looking and how cute he is. I also think the tunnel sequence is crazy and even if i’ll always find flaws to it, I think it’s looking really nice, and this was my first explosion to handle, so that a personal win when it’s been handled in a tunnel ! And i’m also proud to be part of a modern Christmas movie, I love those, but they can quickly feel repetitive or old school, this one for sure is a new kind, and a very refreshing one ! More please !

3DVF: Thanks a lot Lorene and Framestore for this dive into Jingle Jangle!  To our readers: don’t forget the movie is available on Netflix.

For more information about Jingle Jangle and Framestore

A Lire également

Laissez un commentaire