Majestic Worlds By Niklas Werthner
I'm excited to share with you my latest three DMP/environment projects. You can see the animation for each project in my Demo Reel down below. Personally, I love an in-depth look in other peoples work, that's why I tried to make my breakdown as detailed as possible. Hope you like it!
my name is Niklas Werthner and I'm a 24 year old graduate from Pixl Visn Media Arts Academy based in Cologne. For this year's Rookie Award I'll present to you my latest work. You will find a more detailed breakdown and all the animations in my Demo Reel below.
Thank you very much and good luck to all of you!
Pantheon - The Captured City
The inspiration for this project was a concept art I've stumbled across on ArtStation a few months back (But I can't remember who the artist is and was unable to find it again...Please reach out to me if it's yours). I made a block-out scene based on the concept, but was not quite happy with the lighting I got from this setup. So I was experimenting with different camera angles and lighting. Once I was happy I went on from there building my scene.
As my method in this project depends on real images for the textures, I was looking a lot for reference. Finding the right image with good quality was definitely a challenge (Most of the time I had to buy them). As I was going for an old temple look in a pretty dirty and sandy environment, my main source of images were buildings in the Middle East and Asian countries. For the favelas type buildings on the mountains, Low-Res images of cabins and houses in the favelas were enough. At the end I had around 30 seamless textures made in Photoshop which I rendered in Clarisse and could use in Photoshop for overpainting my base render.
-Below you can find a rough but more detailed description of the techniques I used, my workflow and what challenges I had to face during the process of creating these projects. You could also skip this part and watch my Demo Reel below, which also contains a more detailed breakdown.
(I hope you don't mind that this description ends more like a small tutorial)
- If I have a concept, I start blocking out my scene based on it. As this was a personal project, I had no guideline in what I can or can't do. But one thing that is very important for me in the beginning is to think about what is in the FG, MG and BG (Still something I need to improve on). One thing I definitely struggle with is to find the right amount of scale in a scene. As I like to create large scene environments, this is one of the most important things to do right. After blocking the scene, I try to finalize my camera for the MP (There will be slight changes along the way). I created the mountains in World Creator and the thin, long ones in Zbrush based on the concept. After that, I needed a lot of assets. As this scene was gonna have a few dozens of different buildings and props, I knew I haven't had enough time to model all these, so I was relying heavily on different Kitbash3d packages (Vanessa Mai allowed me to use her sculpture from modelling class, so big thanks to her). At this point it's really more like playing with Lego bricks and definitely the most fun I have during a project. I tried to use the assets as useful, often and diverse as possible. For example building bridges out of walls using different assets and combined with the brilliant scatter system in Clarisse you can quickly build a large scene like this with hundreds of millions of polygons.
Let's move on and say we've finished the city. Now I import all my textures and apply each of them on a separate layer on all of my Geo (This takes quite some time doing it for all textures). After that I need some ID masks for Photoshop. The idea behind is to render each texture on all the geo and use the masks in Photoshop to apply them exactly where you need them. I have to group the geometry in the right combination, so it makes sense for me in PS. Grouping and creating shading Layers is probably the least fun part in the whole project, but is crucial for the next step.
After rendering each layer separately, we can go to Photoshop, where we import all our textures, ID masks, AO, Z, Volumes, lights (distant, environment light, + in my case another layer with the 1500 lights for the night scene). Place the environment light on the bottom and the distant light on top (blend mode color dodge or linear dodge) and we are ready to start overpainting. That part is pretty straight forward. I rendered all the ID masks as R,G or B so I can just select them in my channels and can use it as a mask for any texture. After that, it's really just spending a long time painting in and looking at references (The only really challenging part here is to keep a good organization as these projects can have up to 400 Layers) One important lesson I learned (and still learning) over these 3 projects was to clearly separate your scene in FG, MG and BG. Especially with fog elements (or using your Z channel) can really help achieve this. .
After finishing the MP, we can use this one for creating an animation. We have to separate the MP in layers we like to project on the Geo in Clarisse. But first we have to think about, if we need to animate anything in Nuke (waterfalls, people, boats, birds, etc.). In the first project, I only had lights and people to animate, but the second one was quite a lot of work more due to all the waterfalls and birds (I attached a screenshot of how I did that). Each Layer we saved from Photoshop (or animation from Nuke), has to be projected and rendered out separately in Clarisse. This can be quite tedious, especially if you have tons of waterfalls, edges or cables, which require additional cards to not be projected wrongly (for clarification: you do not have to render each bird separately, rather group your layers in Ps in FG, MG and BG so they are not overlapping) . We have to animated our camera next. The more movement we have, the more overpaint we have to do later. After projecting, we can see on the last frame how much information is missing. You can create a mask for the missing part and overpaint it in PS. Go back to Clarisse and blend the two Matte Paintings together with a blend node ( I had quite some troubles with the antialiasing causing on pixel lines between the two projections. Nobody could really help me with this problem and they (support team at Clarisse) thought is was a bug in Clarisse. However, that problem cost me at least a month of extra work, probably even more. Specially in the second project. I had around 20 tracks for each part of the animation to paint over in Nuke to fix this issue (I found a solution for the first project, which worked better). I was following a long (24h) tutorial from Emilis Baltrusaitis (I have to give him a lot of credit for my work and he was also very helpful whenever I had questions), who showed this Clarisse/Photoshop technique, who had no problems what so ever. But I learned a ton in Nuke, so after all it wasn't to bad).
So after rendering all the layers, we can go back into Nuke to assemble our scene. If you don't have the problem I have described above, this is just basic compositing. Just a normal projection setup for the sky and the different fog elements, which we can project on some cards. I rendered one deep image to create a LowPoly model (with DeepToPoints) of my scene to know exactly where to place my cards (My scene is far to big to import in Nuke). You could also just use your Z pass to cut the parts out you don't need and even multiply it with some fog Stock footage. Another thing I did, which helps the shot is to create God Rays in nuke and multiply it with Stock footage (There is a great Tutorial on that from Hugo Guerra: (Hugo's desk) https://www.youtube.com/watch?v=thXx_LNlPSU&t=229s).
At the end I added CC and different effects (grain, chromatic aberration, blur, sharpen, motion blur, etc., depends on what you need)
That's pretty much a rough description of my workflow at the moment. Thank you for taking the time to read through this and maybe you have even learned something. This is just one approach for a project like this and if you are interested in learning more about this technique, check out Emilis Blatrusaitis's tutorial "Monsoon" on Gumroad.
My Demo Reel:
Below you can see the concept with an early stage block-out, the node tree in Nuke from the first project, textures, some references I was using, the problems I had to fix between the two projections and two screenshot from my scene in Clarisse (I used the cards on the bridge to project the animated people on) :
Below are some early renders with overpainted ideas on what I wanted to improve and a few references I used for guidance and for overpainting:
I had no concept for this one. But I liked the idea creating an abandoned futuristic city in a huge cave with tons of waterfalls. At that point I was following the tutorial from Emilis and I learned how to animate waterfalls in nuke, so it made sense to use that knowledge before I forget it.
This project was actually my first big project using the techniques I explained above. This time I used the textures on the buildings as a base color. I textured the head, the structure behind the head and the mountains in Clarisse. I believe I used the plants from Megascans. The buildings are from KitBash3d. I modelled just some basic parts on the FG building, on the water and the head (and the robes). But to construct the city, I made scatter systems for the FG,MG and BG (with approximately 20 KitBash3d models as my base geo). At that point it was more playing around with the settings, till I liked the shape of a building and moving it in place. When I was happy with the composition, I moved on to hand placing objects to give the scene more life and believability. Then I created another scatterer for the plants and placed some trees from Speedtree in the FG and MG.
The following steps are pretty much the same as I explained in 1.1 ...
I learned a ton with this project. Probably most of the things I know in Clarisse and Photoshop (Thanks again to Emilis Baltrusaitis). Even the problems I've faced were worth it and I hope I can figure out what's causing them, without having to fix them in post. Anyways, these methods allowed me to reduce the render time to a minimum. To render each frame was impossible (as I currently have one PC for working and rendering).
I was looking through my images for ideas and pretty much had it in my head when I saw this picture from Iceland (photobash). I immediately started working on the MP for the background, followed by the 3d scene.
This project took by far the least time of the three. I had two weeks to finish, before we had to submit our final Demo Reel. The scene took me around three days to be render ready and another 2 days of compositing. I had to render two times. The first time took 6 days, but was too noisy with quite a lot of fireflies. The second time, I was forced to use a Renderfarm, which finished just in time. I tweaked the settings as much as I could for the time available.
The only real struggle in this project were the quality of the renders. I tried to fix as much as I could in compositing. I had a gizmo in nuke for the fireflies but it killed all the highlights, which was not worth it for me (I cut out some parts where the fireflies were to strong with a roto and used the gizmo). I had to comp quite a lot with this one to get the fog affect from the waterfalls I wanted. I projected the waterfalls in Clarisse to have the reflections in the water. For the MP, I build a simple scene in Maya, which was probably not worth it as there is almost no parallax (Cards in nuke would have been enough). Again, the birds are animated and also projected in nuke. At the end and in between, I did quite a lot of CC make it fit better in the scene and cover up some render imperfections :D
All the environment assets are from Megascans. I handplaced them in Clarisse and scattered rocks and plants on top. I prefer painting the points with the brush by myself instead of covering everything and erasing the one I don't want. The hero asset + the animation are from Mixamo ( I just retextured it). I actually had 3d fog in the scene but there was no way I could render it in time. So I had to go with my Depth pass (I closed the cave with a plane so I can use the Depth channel for placing my cards (saved so much time)).
So that's it with the presentation of my latest work. I hope you enjoyed this little tour and might even leave me with feedback. Thank you very much and if you wanna stay in touch you will find some information about me below.
LinkedIn: Niklas Werthner
About me: - My name is Niklas Werthner and I'm 24 years old. I'm born in Austria and moved, in end of 2019, to Germany (Cologne) to learn more about Visual Effects. My whole life was dominated by sports (especially Track and Field) and is still a big part in my life, but nothing I ever really wanted to pursue as a profession. I never considered an artistic job, even though I really enjoyed art class in school. I remember seeing Avengers - Infinity War (2018) and asked my brother how movies like this were done. We both had no clue. This topic intrigued me but at that time I didn't even know how a PC worked. In mid 2019 I opened my first 3d/Lookdev softwares: SketchUp, Lumion. And in October I started my education at Pixl Visn. Three years later, I feel confident explaining my brother how these movies are done :)