Realistic Visuals: Tips, Tools, and Tricks

Introduction

Hi there! My name is Rajko Stijakovic, and I am the founder of Studio Heisenberg and 3D artist at 3D plans. I come from Bosnia and Herzegovina and love to travel and discover new cultures! I started playing with 3D in my early high school when I discovered Sketchup. In those times, I was a hardcore gamer, spending a lot of the time on my console, and the idea to get behind the scene and make worlds in 3D was stunning. But it will be later, around 4-5 years from that moment when I started taking 3D visualization more seriously and learned it more in-depth. In the meantime, I followed the career of a volleyball coach. The main breakpoint was discovering Cinema 4D and Vray, where I realized what are the real possibilities of the CGI. From that moment, I have worked for over 50 clients till the date and started my own learning platform on Patreon for Cinema 4D and Corona/Octane users. Combining my 3D and coaching knowledge I also try to give value to the CGI community on my Instagram and YouЕube pages. 

The Principles of the Realistic Visuals

My approach is simple, but we all know that simple things are often underestimated, and many tend to skip those. The main thing is to observe carefully the world around you, taking notes, photos of the things you find inspirational, and if you need to make something that you cannot observe in person, then use the internet to collect as much reference as possible. Today is easier than ever to make stuff more realistic, and tomorrow will be even easier, as tools and software constantly improve with new functions. Especially, with the introduction of Substance tools and Quixel Megascans, they made texturing so much easier, and they are affordable even for freelancers.

Modeling

As with material, I try to model everything as close as possible to the real-world, it can be time-consuming, but it pays off in terms of better visuals most of the time. Luckily, as opposed to modeling for the game industry, modeling for architectural or product visuals does not limit us with poly count, so we can really go more in details, but on the other hand, it can drastically increase production time, so it is good to find some nice balance of what is needed to be achieved with modeling, and what we can add later with normal and displacement maps. I model everything with C4D, then use the bridge to Rizom UV to make UV’s when needed and bring it back to Cinema 4D for further use. In certain situations, I polish my models inside ZBrush or do some sculpting if needed. 

The Approach to Composition

I think that every 3D artist should learn photography as much as possible, as it can be a game-changer for the level of the visualization quality. I learn a lot about the composition and cameras watching movies and, in last years, just scrolling through Instagram where you can find great photographers who share their everyday work. But maybe most important thing is to grab your smartphone or even better get a digital camera and practice because when you master modeling materials and lighting, you are nothing else but photographer inside 3D space with virtual camera, and to understand exposure, depth of field, and different focal length there is no better way to learn than to play with actual cameras.

Working with Materials

Materials are indeed one of the most important factors in order to achieve photorealism. Where most people make mistake is the fact that there is no perfectly flat surfaces and no perfectly clean materials, so you need to add small imperfection to gloss to make it look more realistic, also adding some details in the normal map to break flat surface can help make the illusion. The human eye is well-trained in looking the real world every day, so the viewer will easily spot anything that is different, and he might not know to explain what exactly makes some render not looking photorealistic, but he will deeply know that something is wrong. Luckily, most of the renderers and software out there have adopted PBR workflow, so by using some classic slots for textures, you can easily achieve everything mentioned above. It is important to get the physics basics right, to know what is Fresnel, ior, and some main material qualities, after that you can start building some imperfections on top of it, adding scratches, dust, fingerprints, etc. But be aware, don’t overdo!

The Workflow in Substance

I love using Substance tools and introducing them to my workflow more and more recently. For base materials, I use either Substance Source, which is a great library of PBR materials, and most of them can be tweaked to get the result needed, as most of them are made with Substance Designer and you can download native files. If I am not using source materials, I use Substance Alchemist to make all the needed maps from the material photo or reference I find, it is a fun and easy way to make any material, and it is getting more and more powerful. After that, I load them all inside the Substance Painter and then use some smart masks to add imperfections and make materials more lifelike. The only thing I would like in this phase is some kind of bridge between C4D and Painter, where we could send an object with one click back and forth.

Rendering

I currently stick to Corona and Octane with Cinema 4D, but extensively learning and practicing UE4. With Corona for Cinema 4D, I feel most confident as I have been using it since the alpha phase of development and I really like how it behaves with interiors, the other reason I am using it is for tutorials I make for my Patreon page, where I run Mastering Corona for C4D course atm. On the other hand, I like Octane and its speed, the intuitive node material editor and for sure interactivity. I have tested it both in the same scene and achieved pretty much the same result, so it is basically the matter of project preferences for which I will use more. If I need some quick renders with a lot of iterations and changes during it, I will use Octane and take advantage of its speed and versatility. If I need to build a big scene with a lot of objects and work both on interior details and exterior foliage, etc., I will use Corona. I have tested the Redshift with the same scene as Octane and Corona, and the only drawback I had was so many tweaking to get the result I need, which is, on the other hand, good because it gives you much more control. I currently only have a license for Corona and Octane, but as Redshift is now part of Maxon, I think I will have to get it as well. Unreal Engine is for me the thing that will be one of the main tools for all the archviz stuff in the future, and it is developing so fast that is hard to keep pace, and it could be a subject to its own topic maybe in the future when I implement it in my everyday work.

Lighting

As other things discussed, I tend to rely on real-world values and examples, so I use image-based lighting whenever possible, and places like HDRI heaven are great for testing different options! For architectural visualization, it is important to understand where the lights will be in real life, but also being aware that we can bend the reality a bit in order to get a more appealing result is important. Also learning the studio lighting that photographers use and their tricks they use on their sets is important as we don’t have to buy expensive equipment in order to get a better lit scene, it is one click away. Lighting is, besides materials, one of the most crucial things in visualization, as it can be used to hide or reveal good and bad modeling, or good and bad materials, and if you know how to do that, you can spare some time and make it your benefit. But I think each detail matters, so if you have time, dedicate it to make everything looks nice, even if it will be in shadow. 

Rajko Stijakovic, 3D Artist

Interview conducted by Kirill Tokarev