The Search of Hard Surface Perfection in “Éxodo” Environment

3D artist Julián Rojas Millán talked about the creation of a very detailed 3d scene “Éxodo”, which features a lot of various techniques. The author goes into the tiniest detail explaining the production of his project.

3D artist Julián Rojas Millán (portfolio) talked about the creation of a very detailed 3d scene “Éxodo“, which features a lot of various techniques. The author goes into the tiniest detail explaining the production of his project with Arnold Renderer and Photoshop.

UPDATE: We should mention that the scene is heavily inspired by the works of Tor Frick – a wonderful artist from Machinegames. He makes some amazing stuff. You should definitely check out his portfolio.

“Éxodo” by Julián Rojas Millán

Introduction

My name is Julián Rojas Millán. I’m a prop and environment artist in the making. I’m from Venezuela, currently living between Caracas and my birthplace hometown. Right now I work for two companies: Silvermile, which is a Finnish mobile focused company founded by ex Rovio’s guys, and Singular – a venezuelan based augmented reality startup. I work as a 3D and technical artist, mostly focused in created 3D asset from concepts optimized for mobile platforms. That means I have to watch out for the precise polycount and creating really tight UV layouts to reduce wasting texture or unnecessary materials.

In Silvermile I’m working alongside people with a lot of experience in the mobile field, both from Venezuela and Finland, so I have the chance to get deep into the process of optimizing shaders, tools and gameplay elements inside Unity. in Singular I will also start to rig soon, and I hope to script tools as well, so I’ll make sure we have easy to use but fast rigs and an optimal exporting pipeline for Unity. Both jobs are handled remotely, which has it set of challenges, but I’m willing to relocate outside the country if the chance arises inside the industry.

I am also a mathematician, I got a bachelor in math focused in computer graphics from Universidad Central de Venezuela, which is one of the biggest universities in my country. I worked as a research assistant in computer graphics and teacher for a long time. My main objective is to become a technical artist, because by now I managed to grow in the math and artistic side of things!

From Programming to Art

I really like video games, I have studied the game design foundations of them from a mathematicsl perspective and now I have experience creating assets for them, but I always see the artist that influence me: Vitaly Bulgarov, Paul Pepera, David Lesperance, Josh Herman, Fausto De Martini, Alexey Pyatov, Gurmukh Bhasin, Ethan Hiley, Tim Bergholz, Darren Bartley and a bunch of others (the list goes on and on) and I say to myself “I will never be that talented” and I usually lose my motivation. But I end up feeling worst because I’m not trying something in the computer so I go back and try to do something instead. It is a personal and internal struggle because I don’t see myself as an artist.

Creating a routine is the hardest part. The best way to fight the fear, at least for me, is to create a daily discipline of sitting down and doing my job, which I really enjoy. And in the same way devote as much time as i can get to work in my personal projects to achieve the best of me in an step by step process.

For me, it is a long way track to become an artist, the one i want to be, and it is only starting.

Building Éxodo Scene

It started as a way to improve my speed modeling skills, so I drew upon a work of the fantastic artist Tor Frick (you can check out his portfolio here) as a reference (he actually noticed the heavy resemblance from one of his piece when I published a WIP in the Ten Thousand Hours group in Facebook, I forgot to mention his work, I was really embarrassed). As I said, I’m only starting and I’m currently in search of my own style. I look up for the artists that influence me so I have a path to follow. Gathering references is one of the most important steps of any creative process, but the way you transform them is what defines your style. I’m in the search of that style.

By the time I was glad with the first asset, which was like a couple of hour of work, I thought: “well, this is already a modular piece, let’s duplicate it and see what happens if a try some composition here”. Few hours more and I had blocked out an scene with some lighting in it.

When the second pass of details was ongoing It was a fully fleshed project, but meant more as a concept piece, like an plot suggestion for a game, than an environment intended for a game engine, i wasn’t paying attention to polycount, UVs or anything like that, just the final result.

The Search of Hard Surface Perfection in

So I made a plan to achieve some personal goals:

  • To get heavy on the details.
  • Use Arnold Renderer as a way to learn it.
  • Study composition.
  • Get my hands dirty in the post process phase, both in color correction and Photoshop overpaint.
  • Try to convey a story, a plot of some sort.

The Search of Hard Surface Perfection in

I really enjoy the cluttered look of details being layered one over the other, which can be exhausting for an eye to follow, I really like that sensation of overwhelming that a person can get if the scene has so many moving parts. But in creating a story it is a balanced. That balance was an iterative process that took me a while to figure out, and it was really fun. It was like researching in math, programming a solver for linear systems, where you have to think about each tiny step so the numerical error doesn’t blow up your solution.

The Search of Hard Surface Perfection in

The post process phase was also an iterative process where I jumped back and forth between Arnold and Photoshop looking for the right initial material and light setup. As a concept piece, I knew the image would change a lot in Photoshop, but having a good lighting setup from the very beginning was a way, in this case, to learn Arnold. Now I know how to get fast light and material blockouts I can mockup in Photoshop.

The Search of Hard Surface Perfection in

The project took me like 2 weeks to complete. Sadly, after some unfortunate events I lost most of the final files, so I only have a very early blockout of the scene. I provided the WIP pictures of a previous project, which follows the same pipeline but instead of Arnold I used Keyshot.

Working in Hard Surface

1 of 7

Working in Hard Surface is a lot like playing with Legos, with the added feature of polygonal modeling, so for me it is easy to get lost duplicating, transforming and moving details. I enjoy the process. It’s quite relaxing for me and I really like the overwhelming feeling of a lot of details clustered in a scene so it is easy to get caught just piling up details.

The Search of Hard Surface Perfection in

Neil Blevins has a amazing article about Primary, Secondary, and Tertiary Shapes in environment and Hard Surface design. I really dig into that workflow but I like to blow up the details, so there are primary, secondary and tertiary shapes but in a exaggerated manner, where looking for a place to rest is almost like a game.

In that regard, you can emphasize those shapes not only by silhouette, but by lighting, color scheme and texture/overpaint detail. For example, if you have a bunch of tiny pieces but one of them is red and the rest is blue, the blue ones are processed by the viewer even like a part of the background in many cases, a part of a bigger shape, or if you composition lines are literally the edges of a big piece of your environment, the floor or the ceiling. This helps the viewer to know where the primary, secondary and tertiary shapes are located. It’s a process of balancing, in my case that means erasing a lot of detail that I would like to leave there, but they are just not reading well.

The Search of Hard Surface Perfection in

As I said, hard surface modeling is an iterative process, a back and forth dance, making changes in lighting and color until each piece works: based on silhouette, inner detail, color scheme or material. It is slow at first, but you can get really fast in that process by practicing. You develop a sense of what will work and what won’t.

Another problem is over detailing. In a cluttered scene like the ones I like to develop, much of the dense detail is lost because of the number of pieces, so even though it’s a cool process to make each piece as detailed as you can, you need to finish the scene at some point. In that sense having a good blockout of the geometry and the camera really helps to know where the details work best, helping to focus your efforts where it matters the most.

The Search of Hard Surface Perfection in

I also relied a lot on kitbashing. The process works like this: I modeled the big shapes and modular components, detail them just a bit so they say something by themselves, duplicate them and transform them until they work composition wise, and then use a library of pieces so I can kitbash the rest of the details. Those kitbash pieces can be created by you, but you can find them on the internet, both as free resources or paid ones. After a few projects the idea is to have a kitbash library with your own created or modified pieces plus the downloaded ones.

Kitbashing is the most LEGO-like part of the process, and even though you end up editing most of the pieces to fit your scene, besides designing new ones, kitbashing is a way to save time you can use not only to be more productive but to work your narrative and composition in the image.

Having your kitbash library is crucial if you enjoy hard surface modeling. There is no such thing like enough kitbash pieces.

Setting Up The Lighting

As it ended up being more like a concept piece that a real time environment, it is a mix between Arnold, After Effects and Photoshop. I really enjoy this process of modeling and overpainting. I want to move towards concepting as I gain experience.

The Search of Hard Surface Perfection in

I took the time to read about Arnold and create a few practice scenes where I grasped the basic workflow, learned about balancing sampling for camera, lights and materials and how to achieve fast renders through that balance. Then I went to the main scene, blocked out some lights using Maya Renderer and then used Arnold real time feature to tweak a first pass. I then bounced between modeling in Maya and nudging the lights while building the scene until the last piece of geometry was laid out.

The scene has 4 area lights and a glow material for the planes that represent the light panels. It’s really simple. A similar approach can be done in Keyshot, moreover now that the latest iteration has an interior lighting mode, but having in mind that Keyshot needs geometry to put the light material on it.

When that process was finished, I started then putting the result from the Render Viewport to Photoshop so I can mockup a mood tweaking some filters, then go back to Arnold and adjust the light until I was happy. That step isn’t needed as you can change as much as you want in Photoshop, but I was practicing the lighting setup in Arnold.

The Search of Hard Surface Perfection in

Two volume passes were then setup, one for the key light and one for the remaining ones. The glow on the lights was enhanced quickly using After Effects, which has some nice effects that for some reason aren’t in Photoshop.

I don’t like how Arnold manages Ambient Occlusion, so I made that pass using mental Ray which I used extensively when I worked for the TV network. I also created some masks to manage layers in Photoshop.

I made some scripts to automate the process in Arnold (for example to assign flat shaders to obtain a material masks pass) but those passes come by default in Keyshot, alongside others like a depth map. Unfortunately, Keyshot doesn’t have volumetric effects yet, but that can be somehow emulated in Photoshop from the depth map for example

Using all of that I started the overpaint process and a heavy postprocess phase when I adjusted the lighting in a very dramatic way. Initially it was a lot darker, but most of the detail got lost so I went for a more “Clinic lighting” look where everything is visible but is so evenly lit and somehow pretty clear that is like an hospital, not mainly terrorizing but more like unsettling. That was the main mood goal after previous attempts. I might look forward to continue this trend in future pieces.

Now that I got some experience, I will use Arnold alongside Keyshot, depending of the asset or main goal, to render out this concept pieces. Both are fast, easy and highly customizable.

Material Creation

There are no textures, the materials have a solid color or a Maya ramp color node that just shifts different values. I worked all of the “texture” detail through overpainting in Photoshop. A beauty pass was used as a base, a lot of photographs were transformed, cropped, changed and used in different layer modes to detail the scene. Thought like a concept piece, I allowed myself to illustrate over it and treated differently than a in game asset. I’m working on a new piece, now that I have the time, when I will lay out UV in the fastest possible way and use a combination of texturing and overpainting.

For the materials, I created a new scene to sketch them, following Arnold’s Renderer Online Help where they show some basic material setup. In that very scene I balanced the sampling input so I wouldn’t have to do it in the main scene which was quite heavy. If you dont take care about that early your render time will skyrocket. I then reopened the main scene, combined geometry according with the material they will have, exported the material, assign them and tested out a first beauty pass. That was the fastest process of all, I didn’t spend that much time on it, the final scene has only 4 materials: plastic, chrome, painted metal and a diffuse glow for lights. You can do a very similar setup in Keyshot.

***

Be sure to check the second part of our interview with Julián, where he will talk more about his experience working with Unity.

Julián Rojas Millán, 3d artist

Join discussion

Comments 0

    You might also like

    We need your consent

    We use cookies on this website to make your browsing experience better. By using the site you agree to our use of cookies.Learn more