Hi! I'm Guido Ekker and I am a Senior VFX artist and Filmmaker living in Amsterdam, The Netherlands.
Working as Senior Visual Effects Artist for Post Panic. I am closely involved in pre-production and concept development from the start of the commercial or film, with previs and on-set VFX Supervision. In Part 1 of this Process Technique article, I'll be talking about setting up a 3D scene in Maya with Redshift to enable a 3D paint over as this is a great technique to know!
Let’s get to it!
Painting over 3D renders is a very useful technique to save time in the overall 3D process. It helps to quickly visualize something that might have taken more time using a more traditional 3D route. I am using this method on a daily basis to work on my DMPs or concept stills.
There are multiple ways of doing it and also multiple software available but in this tutorial I’ll explain my workflow using Maya, Redshift, Nuke and Photoshop. In this demo I will be using Kitbash3D models to be able to flash out a quick scene.
"When you start this process, try to make-up a story that fits well within your scene"
For this practice I wanted to focus on something different, no wide angle lensin\. Something that would fill our frame, something condensed that goes well with the architecture. I wanted to take that look a bit further, creating a more futuristic slum with unified colors. When you start this process, try to make-up a story that fits well within your scene. The story will be the foundation. That way you can explain what we are looking at, and why things are in a certain way, such as lighting, texturing, etc.
To start the visual process I usually like to work with a pinterest board to get inspiration flowing before I start to dive into 3D. These are some of the building references I got together. When working in VFX and creating a full CG shot, reference is always key, it's the one thing that shows you how our reality is and we can always fall back on that.
Here I am setting up the scene, using a mix of different buildings and doing the street layout. This is the foundation of the shot. As a guideline try and think of your scene in a specific area on our planet. Go there on google maps and see how the streets are formed, watch how buildings intersect and try and keep this in mind while creating your layout.
These 3D setups help a lot to establish your camera lensing, lighting and texture. Exploring around and looking at my ref images I went for a very long lens, 200mm, to get that condensed feeling of all these apartments in one image.
"You can find all sensor sizes including the respective resolution"
Talking about lenses during the 3D layout part, it’s good to always keep in mind the technicality of a real camera. Often we just pick a millimeter but it's good to know that camera’s do have more to offer, especially working within VFX we always stay true to the Shot Camera on set. In real life you would pick a specific camera, a RED, Alexa or maybe a 5D, etc. These cameras also come with a certain sensor size. A good site for this is vfxcamdb.com. That’s where you can find all sensor sizes including the respective resolution.
In this case I went for Arri Mini, (21.12) 42.250 mm x 17.70 mm. The horizontal aperture is multiplied. To get the correct anamorphic fIlm aspect of 2.39, we have to stretch that number. These values you can enter over here.
"Lighting is one of the most essential parts as an artist. Not just within 3D but also within photography"
Lighting the scene
Lighting is, for me, one of the most essential parts as an artist. Not just within 3D but also within photography. If the lighting is good, and it feels nicely balanced you can steer the eye of the viewer. Even if the models or textures are not that great, with proper lighting you can get away with much more!
For this image, I wanted to go with a warm and hazy cityscape. This will be one of the main aspects to focus on during the lighting process. I'm using multiple light sources, but the main fill light is a dome light with an EXR image, used as my HDR. For this one I'm leaving the settings as is.
The other three light sources are the ones that will create the light direction and the warmth. For these I'm using Redshift physical lights.
- One directly behind the buildings, as main driver, with intensity of 5.
- Top light with intensity at 2.
- And one light opposite site, behind a building, to fill in and cast some shadow on 0.8.
This image shows the positions of the lights and the settings of the main physical light:
Since the lens is quite long, creating a more condensed flat image I wanted some atmospherics to be picked up by the lighting in order to layer the buildings. This will also then establish the warmth and pollution of the city. You can enable it in the output menu under Environment. The following settings I used for this:
"For the hanging cloth I went with an overall red texture, I quite liked the unified feeling of the slum"
With the premade textures hooked up in my redshift material. I am only adjusting IOR values to correct the material fall-off. IOR values you can find online for any specific material! For the hanging cloth I went with an overall red texture, I quite liked the unified feeling of the slum. Could be somewhere in an alternate reality. Let’s say a rebellion army owning a part of the slum!
From the current lighting setup I am able to judge how much extra texture work I need to do in Photoshop. So for this demo I will not dive into more texturing before rendering, we sort of skip this pipeline step. Our texturing will be more focused during the paint over stage in Photoshop.
Let’s take a closer look at the render settings and sampling. Usually for quick previews I’ll keep my adaptive error threshold at 0.100. This works efficiently when adjusting lighting. Once we’re happy with the lighting I will switch to a more production ready threshold. Lowering it down to 0.010. This will get rid of almost all the noise in the render.
The Min Samples and Max Samples I adjusted a little to get a more refined sampling, especially on the spec. We will talk about that a little more when we set up the passes. In the Common tab you can set your image resolution of the render. Our resolution will be related to the sensor size, aspect ratio of 2.39. With a width of 2048 and height of 858.
Under the AOV tab we will select the passes we want to output. I usually have a standard output of passes that I will use to tweak my Beauty render in Nuke or any relevant compositing software. The reason I am separating them out, and not have one beauty image, is to have full control in the compositing software. This way we can balance the image and push certain passes.
The main passes that I want are, Diffuse, Direct, Global Illumination, Specular and Reflection. Also Cryptomatte is very useful to isolate objects. On top of that I have the Z pass twice, one anti aliased and one aliased. These are good for the depth of field and possible atmosphere. Some other passes are also on, but we mainly will focus on the main passes that are mentioned above.
"From here onwards I switch to my Nuke/Photoshop workflow"
From here onwards I switch to my Nuke/Photoshop workflow which will be covered in the next part. In Nuke we will be combining our passes to create a good looking beauty render. We will take the most out of our individual passes, we look at the order of compositing and finish it all off, taking the final beauty render a step further in Photoshop. Painting our textures and increasing the photorealism.
So there you have it. How to set up the scene! In Part 2 of this Series I'll be showing you how to continue in Photoshop to get to the Final Image so stay tuned!
Thanks for reading and catch you next week! For more of my work and to connect with me visit me below.
~ Guido Ekker