While it is towards the end of the year and a week out of my work schedule adds a tiny bit of stress onto my mind, I couldn’t pass up the opportunity to spend a week at the Animex Festival learning from industry. The college had offered to take us for the Games side of the event only, however Kelly and I signed up for the VFX side too and have been given official leave.
- The FX of Lego Batman – Matt Estela, VR supervisor, Animal Logic
- Animating Ethel and Ernest – Peter Dodd, Animation director, Lupus Films
- Creating the Characters of Fantastic Beasts and Where to Find Them – William Gabriele, Rigging TD, Framestore
- The Animation of Fantastic Beasts and Where to Find Them – Colin McEvoy, Animation supervisor, Double Negative
- The VFX of Rogue One: a Star Wars Story – Bradley Floyd, 2D Sequence supervisor, and John Seru, Generalist lead, Industrial Light and Magic
After a short break there would be a social meet and greet on the night, allowing us to talk to all of these talented people and hopefully learn a thing or two in the process.
The FX of Lego Batman
Now recapping an entire lecture would be insane, so between Kelly and I we took some shorthand notes. I’ll try to highlight some of the major points and leave the rest out.
The Lego Movie was a total success, most people didn’t expect it to be but by god we were all proven wrong. For Lego Movie and Lego Batman, Lego provided their brick database used in the Lego Digital Designer, giving the teams access to 3D models of the entire catalogue from the beginning of development. This allowed the assets to go straight into the pipeline. It also goes to some lengths at explaining how the entire development only took two years on a movie this size (more on this in a moment).
Originally a Lego Ninjago movie, along with the Lego Movie sequel were planned to happen before a Batman movie. However after the initial pitch and first draft concepts being green-lit, it was moved ahead of both other productions giving the team a mere two years to complete the project rather than four. To achieve this the script writers, story writers and art department were running in parallel to speed up pre-production.
The Lego movie was mostly kept to smaller desktop sets, however the team didn’t feel they could accurately portray Gotham City in such a way and wanted to aim for a larger scope with an atmosphere that (quote from Matt) “the movie should be as colourful as the Joker, but as dark as the Dark Knight”.
The next tid-bit was what completely made the Lego Batman movie for me, they tried as much as possible to reference as much of the 78 years of Batman history as possible, drawing design references from comics, the animated shows, the 60s show etc. For anyone in the know this made the movie a laugh a minute if you knew your stuff.
Now for the technical part. The in house renderer for the movie was called Glimpse which originally started as a plug-in for Renderman and slowly grew into its own renderer. Glimpse would trace the rays and feed them back to Renderman. Towards the end of the movie some scenes were rendered entirely with the now independent Glimpse but otherwise the bulk of the movie was still Renderman with plug-ins. The software is still being developed in house for future Lego movies. To wrap this up a single Lego stud had 1000 vertices, which…blows my mind to be honest. Gotham City was in the trillions of vertices and is absolutely staggering that they brought down rendering to a mere three minutes a frame. Hats off to the technical team at AnimalLogic!
Also a big thank you to Matt Estela on the night during the social event for some cracking industry advice, some confidence pep talk and brilliantly funny career stories! If I manage to learn any Houdini over the summer, its thanks to this guy.
Ethel and Ernest
Going into this lecture I wasn’t aware of the movie being discussed but as soon as the first slide came up I recognised the art style and knew it had to be related to the Snowman in some form or another. Turns out yes, its an adaption from another Raymond Brigg’s book about his parents. While the movie only took a year to make, it has seen its share of development hell and took eight years to gain momentum before production began. My hats off to all the dedicated people that believed in the project and kept pushing to make it a reality. The movie was created at Lupus Films (responsible for the shorts Snowman and the Snowdog and We’re Going on a Bear Hunt) but this was their first feature film.
Raymond’s art style on paper is a rough one, known for very quick pencil sketches and then photocopying these before reinforcing his line work and then over painting these with goache. It makes for a very hand crafted and personal feel, something which the development team wanted to recreate for the movie. Initial tests were done to animate old school using pencil and layout sheets but this was too time consuming and costly. Therefore for this talk I’d like to focus on some of the methods and cheats that were used to aid development.
Lupus made great use of the tool TVPaint with plug-in LazyBrush, which Peter Dodd had made use of during working on the recent TSB ad campaigns.
This tool allowed large areas to be quickly coloured using strokes across an area. The magical plug in reduced the colouring per frame from 30 minutes to 2-3 minutes. To achieve this hand painted colour feeling, photographs and scans of textures, paint swatches and other physical media were added to LazyBrush.
Going back up a level to TVPaint itself, many attempts were taken at making pencil strokes look as real and natural as possible. Line work was then duplicated, blurred and inversed for a white outline, colour/texture added to this layer and then the darker hard lines added back on top. To aid the animators with the drawing process, Animbrushes were created containing 30-40 turn arounds of character heads which allowed quick stamping of varying angles. Oh and finally many backgrounds were modelled in 3D to use as perspective reference, such a cheat is great to see being used in industry as its something I’ve done myself in the past.
Creating the Characters of Fantastic Beasts and Where to Find Them
This talk by the very talented William Gabriele from Framestore was easily one of the more complex of the day. He walked us through the rigs used by some of the goblins used in the movie, specifically Gnarlak and the goblin singer in the same club scene.
Framestore worked closely with JK Rowling to create these new characters and the film overall had 500 artists working on 36 full characters across 400 shots. The amount of work involved is staggering. Fifteen million CPU hours were required to render these shots and we were told this would have taken 1700 years on a single CPU, this boiled down to 274TB of rendered data. You see stats like these and realise movies are on a whole other level to games.
As for the rigs used, I never even considered that movie level rigs would have musculature, tendons and accurate skeletons rigged too. I give my utmost respect to technical artists able to work at this extreme level of anatomy. I wish I had examples to show but like all of the talks, photographs and filming were not allowed due to sensitive behind the scenes material. For Gnarlak they took a standard male human body rig and adapted the proportions to give a starting point.
As for the animation, most of it was motion capture from the actors giving the performance which was then put through a validation process, locked to the proportions of the character and then some manual clean up by hand.
I wish I had more to say on this one, or at least say I’ve picked up some great hint of technical knowledge but honestly it was WAY above my head in terms of complexity. Amazing talk though and very eye opening to the kind of quality required for movies these days.
The Animation of Fantastic Beasts and Where to Find Them
Another great talk revolving around Fantastic Beasts, but this time from Animation Supervisor Colin McEvoy from Double Negative, discussing bringing Frank the Thunderbird to the screen.
All kinds of different animals were looked at and integrated into Frank’s final design such as various birds of prey, horses and oddly jellyfish. Due to his size the team were worried his wing span was too narrow and wouldn’t look believable, however once animation tests were done his two extra smaller sets of wings gave enough visual language to suggest he would still be flight capable.
Being graceful was brought up a lot in their briefs and this fed into Frank’s animation a lot, the curve and arc of his tail were a particular area where the team needed to focus on this idea and implemented many slow fluid arcs to achieve this.
I’m going to derail from the bulk of my notes here and focus on two points, having recently animated a few cycles and had some major headaches with my own animations Colin mentioned something that changed everything. Animation layers. By using layers you can create a cycle, move to a new layer and key onto a new timeline without affecting the layer below (similar to how one would use layers in Photoshop). He would begin by doing his blocking phase, move to a new layer and then key on 4s using straight ahead techniques, new layer and work on 2s and crazily another layer and work on 1s. That is an amazing amount of work but also a crazy level of control down to the tiniest level. This is something I’ll have to test in animations next year, this could really help and many thanks to Colin for spending some time on the night explaining this to us in more detail, we really appreciated it!
Colin also mentioned he occasionally animates just the silhouette to focus on the shapes and the negative space, refining not only the animation but the composition. The field chart tool comes in handy for this, independent to the camera it lets you track your character movement without any background, great for tracking arcs and making sure your character is framed perfectly.
The VFX of Rogue One: a Star Wars Story
Being a huge Star Wars fan, was I excited for this? Does the Pope wear a funny hat?! The only thing stopping me from jumping up and down in my seat was the tiny lecture theatre seats! Bradley Floyd (2D Sequence supervisor) and John Seru (Generalist lead) from ILM London were here to discuss the process of concepting and creating the city of Jedha for the movie.
Harking back to the visuals of the old movies it was lovely to hear they went straight back to Ralph McQuarrie’s concepts for the original for inspiration along with location scouting and visual inspiration from the Middle East, Egypt and Africa.
Here is the single greatest thing I learnt from this talk and it is an absolute source of motivation from now on. The model for Jedha as seen above was put together through a mix of Maya, Nuke and ZBrush, although you might think it took a team of 20-30 artists maybe? Two artists, just two artists modelled the entirety of Jedha and I am blown away by it. How was this possible you ask, by using simple building models which were unique on each side, cleverly rotating each building and arranging them into small districts surrounded by the occasional bigger hero building. In total Jedha only had around 30 unique buildings. Layered on top of this were smoke/steam effects, they built up a small library of these and placed them around pipes/grates for added effect.
For the explosion of Jedha (sorry anyone that hasn’t seen the movie yet) the team researched nuclear weapon test videos specifically looking at clouds, shockwaves, how the ground was affected, heat and exposure. A slightly grim task but the results speak for themselves. To achieve a lot of this the terrain that had been created in Maya was procedurally shattered and had various FX simulations added to it (tectonic cracking, debris, dust etc). Amazingly most of what you see in the final movie was rendered in Arnold, I’ve obviously underestimated Arnold as a renderer and should spend more time figuring out how it works.
I had a few quick words with both John and Bradley on the night, just about general paths into the industry and the kind of things to aim for in a showreel. Both assured me that degree education absolutely isn’t necessary as long as you have talent, keep your show reel short and focus only on your absolute best work. Don’t build up to the best, open with it and wow early. As for movies their suggestion was simple enough, model and texture well enough so that it would blend into reality and not be out of place in a video composite. Thanks for the tips and your time guys, along with generally nerding out over Star Wars!