LogInThe creative mind behind ‘The Lord Inquisitor’ Erasmus Brosdau talked about the way he managed to produce amazing CGI-animation with such a lean team.
Erasmus Brosdau is a unique Warhammer 40k fan. He doesn’t just paint figurines and read ‘Horus Heresy’. Instead, he and his small team created a spectacular CGI-animation film about one of the greatest fictional universes in the world. In our exclusive interview, Erasmus discussed the way he managed to build an epic 9-minute short ‘The Lord Inquisitor – Prologue‘. This awesome animation is rendered with CRYENGINE!
The Team
The team for making this movie got constantly up and downs in terms of size. So once I originally released the first major trailer in 2011, I got hundreds of applications from people wanting to help out. With this huge success I expanded the project a bit, knowing I have a bigger team – however it turned out that 90% of all people are just making false promises and nobody helped. I must admit, due to the not yet complete story at that time I couldn’t give lots of tasks, but those which I gave were never completed by team members and I ended up doing them on my own, which was funnily way faster. So in the end for the final production of the prologue I reduced the team to an absolute minimum for knowing they’re all reliable and also for a way easier communication workflow. So for the prologue it was still only me for like 90% of the entire art, including characters, environments, textures, lighting, materials, etc. A friend of mine who now works at ILM made the imperial guard characters and my good friend Soheyl, who recently shipped The Witcher 3, took care of everything animation related. I’m quite experienced in a variety of fields in CG, but I was very lucky that somebody else took care of the animation pipeline. For music I knew that Adam Harvey was the best choice, as his music for Warhammer stuff is amazing and he is absolutely reliable and fast as well. Other than that I had a good colleague from Crytek working on a few props and assets, which also saved me quite some time.
You’ve just released the Prologue. Magnificent work. Could you talk a little about the direction of this piece? It doesn’t really have that much animation – mostly the glorious shots of the city and the parade. Was that an artistic decision or the more practical one, cause you did not have a lot of time to do character animations?
That was very intentional from the very beginning. I’m a huge fan of Mamoru Oshii and really love movies where they take the time to introduce the environment and the world. Nowadays, this seems to be quite forgotten and instead focusing more on exploding things – which are of course also awesome, but I’m more interested in a great balance. I intentionally showed lots of scenes from various places of this city, so that people could understand how it must be to live there. A place like this doesn’t convince with only one single establishing shot, so I came up with the idea of a parade walking through the city allowing me to connect all areas easily. That multiplied my workload by factor 3 I guess, but in the end it payed off nicely. So as this is only the prologue, the main story would kick in afterwards and you can see a glimpse of it towards the end of the prologue. But this balcony scene at the end would be way less effective without knowing what kind of place this is. You watch the prologue and know that this city is gigantic and that was very important to me to sell the sense of scale as it’s omnipresent in the Warhammer universe.
How did you work on the environment production here? The whole architecture, the sheer size of the sets is overwhelming. How did you arrange the creation of these beautiful pieces? How did you model these impressive landscapes? Especially the architecture. Are all the models original? Did you maybe buy something? It’s just stunning.
I thought of a clever construction workflow, quite similar to kitbashing. In one shot I used a free skyscraper from an old sci-fi collection, but in the end had to completely re texture it. Everything else is completely uniquely modeled and textured by myself, together with a good friend of mine, Frank Meinl, who created some sci-fi skyscrapers a few years before which I could use as well. I took a lot of references towards gothic architecture and started to make a 3Ds MAX scene with various small assets. These I then assembled to some bigger assets, which I then again assembled into something bigger until I eventually had a whole cathedral. Everything UV wise is simple box mapping with tiling textures. If an asset is close to the camera I used decals to put over them to show local dirt leaks, etc. Also I made kind of various levels of quality, so for the far background I used very simple buildings I could never use for a closeup.
Terra_04
How do you work with the materials here? What tools are you using? How did you organize the process of all the material production to conserve the time and build everything faster?
As the whole prologue is rendered in real time using CryEngine Cinebox I set up all the materials in engine of course. That’s pretty straightforward, they all get their albedo /spec / gloss / normal map (no metal pbr) and occasionally I use parallax occlusion mapping for giving the impression of highly detailed geometry, but it’s just a height map so to say. For my kitbash scene in 3Ds Max I only worked with one Multi/sub material that then had over 40 IDs for all different materials. So while I was assembling something new out of my gothic kitbashing assets, I could throw exactly the same material on it in Cinebox later and all the materials like stone, gold, marble, etc would instantly find their position based on the material IDs. And as everything are tiling textures they can all share the same material and there is no need for custom UVs + unique textures.
prologue_02
The parade is the centerpiece of the whole intro. How did you animate so many elements here? Did you use Mixamo? Golaem? How did you manage to work with all those animations?
It’s all without any plugin usage, I just set it up in a way that was kind of easy to control. This only works however, as the parade is just marching completely straight and forward. So I modeled all these wizards and after Soheyl had made their walk animation cycle for a couple of seconds, I made the cloth simulation for all elements and exported the character + cloth sim + materials into Cinebox. Then I took all these characters and linked them to a simple sphere obejct, that was below the street. In the animation editor from Cinebox I then only animated the sphere forward in the appropiate speed, while all characters had their animation cycles constantly played in a loop. So when you hit play, it actually looks like all characters are marching forward in the correct speed and as if I had them all animated by hand or using some cool plugin. But basically they’re all following an invisble sphere.
The Prologue starts with the words that the entire thing is rendered in real-time. Why did you decide to go with real-time rendering engine(I presume it’s Cryengine)? What are the advantages of this tech? Why is real-time better?
Yes, it’s CryEngine Cinebox, one of the early versions. I originally started the project with using V-ray, but even I have a very broad understanding of the software, I was “only” able to turn down rendertimes to something around one hour per frame, which is just too long for a project like this and when you have no render farm. Real time rendering has progressed amazingly, so this pretty much saved the project and allowed me to render it. Also, it brings in the fun again, as when you light your scenes in 3Ds Max you can’t see anything, always have to hit the preview render button and wait a few seconds to see what has changed. In realtime, every time you change the sun, fog, camera, any kind of light, obejct, etc you see it updating instantly. Once you worked with this you can’t really go back, it’s just the most natural thing you could have as an artist / director. You change something, you see it updating. The downside of course that it can’t match the quality of V-ray, however I got really close in a few shots, but a lot of knowledge is required to make realtime graphics look really good, as no fancy path tracing is putting nice shadows and colors everywhere. Also there are more problems when it comes to transparency rendering like hair, etc, so I needed to figure out some workarounds for making the hair look good and such. But the ability to edit and manipulate your film in real-time is a huge deail and in the end made it possible for me to finish the project.
How does your character production process look like? How do you build your amazing characters? How do sculpt these characters? What are the peculiarities of building these amazing W40K characters with all their skulls, elaborate clothes, and armor? It seems like a lot of work?
It is actually a ton of work indeed, but I got so routined in character creation that I can create them kind of fast. I learned a lot of the production of characters compared to when I started. Marcus for example, I modeled him 4 times as I have become so much better with my art skills over the years. The final version we see here in the movie however only took me two weeks to make from scratch (without the head) So the pipeline is really traditional, Max for basemesh, Zbrush for sculpting and retopo, then Max again for UV layout and then off to Substance Painter for texturing and baking. The heads took a lot longer time as every detail is incredible crucial. It took me a lot of time to learn, especially as I was also doing all the facial animation myself and needed to observe every wrinkle and such things and how they change with each expression. At that point I was working on the characters for Star Citizen and reviewing the animation quality with bones and blendshapes, so that knowledge also helped a bit to speed things up. I started with the head for Marcus and took care the base head model is perfect, all eyelashes, teeth, topology, etc was correctly in place and that I was happy with his face design as I didn’t use any scans or so, it’s all hand made. Once I locked this down, I made 50 blendshapes for his head, sometimes with inbetweens (for example to fake sticky lips) and made sure they all blend correctly and natural. It’s always the little things that counts, for example when you open your jaw, your ears slightly move as well. It’s probably something you don’t think of first, but when you see it in animation it just looks more realistic. After Marcus head was complete with all the blendshape – which was done in Maya btw. As their blendshape pipeline is a 100 times more efficient and flexible than Max’ one – I used that head for sculpting every other character. So I punched Marcus so often in the face with the brush until he looked like a different character so to say. Once that was compelted, I baked the new vertex position to all the blendshapes, which updated accordingly and were ready to use. Then only creating new textures for the new head and boom you have a finished new working head. Even the heads that were done before Marcus’ final head were done this way. I took Marcus head and conformed it to the underlying old head mesh. Than I baked the textures from the old head onto this new head topolgy and it looked pretty much 99% the same as the old head, but had now all working blendshapes + better face geometry and details.
At that point I was working on the characters for Star Citizen and reviewing the animation quality with bones and blendshapes, so that knowledge also helped a bit to speed things up. I started with the head for Marcus and took care the base head model is perfect, all eyelashes, teeth, topology, etc was correctly in place and that I was happy with his face design as I didn’t use any scans or so, it’s all hand made. Once I locked this down, I made 50 blendshapes for his head, sometimes with inbetweens (for example to fake sticky lips) and made sure they all blend correctly and natural. It’s always the little things that counts, for example when you open your jaw, your ears slightly move as well. It’s probably something you don’t think of first, but when you see it in animation it just looks more realistic. After Marcus head was complete with all the blendshape – which was done in Maya btw. As their blendshape pipeline is a 100 times more efficient and flexible than Max’ one – I used that head for sculpting every other character. So I punched Marcus so often in the face with the brush until he looked like a different character so to say. Once that was compelted, I baked the new vertex position to all the blendshapes, which updated accordingly and were ready to use. Then only creating new textures for the new head and boom you have a finished new working head. Even the heads that were done before Marcus’ final head were done this way. I took Marcus head and conformed it to the underlying old head mesh. Than I baked the textures from the old head onto this new head topolgy and it looked pretty much 99% the same as the old head, but had now all working blendshapes + better face geometry and details.
Once I locked this down, I made 50 blendshapes for his head, sometimes with inbetweens (for example to fake sticky lips) and made sure they all blend correctly and natural. It’s always the little things that counts, for example when you open your jaw, your ears slightly move as well. It’s probably something you don’t think of first, but when you see it in animation it just looks more realistic. After Marcus head was complete with all the blendshape – which was done in Maya btw. As their blendshape pipeline is a 100 times more efficient and flexible than Max’ one – I used that head for sculpting every other character. So I punched Marcus so often in the face with the brush until he looked like a different character so to say. Once that was compelted, I baked the new vertex position to all the blendshapes, which updated accordingly and were ready to use. Then only creating new textures for the new head and boom you have a finished new working head. Even the heads that were done before Marcus’ final head were done this way. I took Marcus head and conformed it to the underlying old head mesh. Than I baked the textures from the old head onto this new head topolgy and it looked pretty much 99% the same as the old head, but had now all working blendshapes + better face geometry and details.
There is still a lot to do, as this is supposed to be a full short movie in the end. I can always need the help of very skilled 3D artists, that are also reliable. As I said I have very bad experience with working with people on non-commercial stuff all over the world, but if somebody really interested in the project wants to help out, that would be more than appreciated.