Voodoo Kitchen: a making of
Step 1: Idea, concept and layout
You can always find inspiration in everything that surrounds you. For this particular image the inspiration came after reading an article on VICE News which had some good photographs of messed-up rooms and kitchens.
When the basic idea came to my mind, I started to collect a lot of reference images. This is the most important step because the best place to find variety and details is the real world. References are an extremely important part of the job, and a good software to manage them is PureRef, which you can download here:
After I've started doing a very basic layout and lighting, I like to paint over my renders in Photoshop, using grey values just to understand where I can put objects and how the overall feeling of the image is going. This is a good habit that I use until the end of the job.
Step 2: Composition
Having a good composition is crucial to obtaining a strong result in every form of visual art. One of the most simple rules of composition is the "rule of thirds", often used in photography and cinematography.
Basically we divide our image into nine equal parts using two equally spaced horizontal lines and two equally spaced vertical lines. The most important features of the image should be arranged on these lines and their points of intersection. This way we can decide the "visual flow" of the composition and lead the viewer's eye where we want.
Even a human face is divided in three equal parts (chin to nose tip, nose tip to eyebrows, eyebrows to hairline). You can always break this convention to achieve your desired effect, but that2s something that should be done with full awareness of the rules.
Step 3: Modeling and set dressing
For the modeling, I've used a combination of simple Editable Poly techniques in 3ds Max and sculpting in ZBrush. The assets imported back from ZBrush were decimated or remeshed. We don't have to worry about topology since this is an environment with only static assets in it.
I won't go into depth about specific modeling techniques and workflows, but I'll share some thoughts on the general process. For a scene like this, it's really important to model all the final assets in separate files, so we can evaluate them better and only later include them in the final scene.
Once all the major assets are in place, we can add richness and chaos to the room. Instead of manually placing the dishes and the objects I use the powerful MassFX tool inside 3ds Max. The drawer is an example of how you can fill a concave object in a natural fashion using simple MassFX simulations - you can find some useful tutorials online about this technique. Advanced Painter is another great plug-in to place objects in your scene.
Step 4: Texturing and UVs
Everyone hates doing the tedious process of UVW unwrapping, which is why I tend to use simple UVW box mapping or ZBrush features such UV Master, Spotlight and Polypaint everywhere I can.
When I need cleaner UVs, I use UVLayout to do my unwrapping since I feel to have more control and visual feedback on how my UV shells are going to flatten. 3ds Max has some great tools for map extraction and I often use the "Render surface map" panel to bake my occlusion and cavity maps. These maps are great to add details to our textures and to create efficient edge masks.
I often use tools like the Mix and Color Correction nodes inside 3ds Max. For me, the best textures online are the ones from www.3dtotal.com and www.cgtextures.com, which I use in combination with hand-painted details.
Step 5: Shading
Before doing any shading, I export my objects and merge them in a "look development" scene (a file with a studio light setup, so I can focus on every single object that needs materials and textures).
It's very important to have specular and reflection maps in order to break the perfect look of the CG. Experiment with maps such VRayDirt to create effects like dust on cavities.
VRayBlendMtl is a powerful tool to layer complex materials on top of each other, using black and white masks to control the transition between the shaders.
I usually paint my dirt masks inside 3ds Max using Viewport Canvas, but I also use a lot of textures from the 3dtotal dirt collection. Some objects, like the piece of meat, have slightly more complex shaders to simulate translucency. The curtain uses a VRay2SidedMtl - a very efficient way to "fake" translucency where we have two-sided objects.
Step 6: Lighting
Lighting and composition are the vital parts of an image. I do my lighting very early and keep working on it until the end. I try to balance the exposure inside the camera using VRay RT to have immediate feedback on what I'm doing. White balance is important, so for this I've created a white sphere inside the room and picked the color of it from the white balance option in VRayPhysicalCam.
The main lighting is made by a single HDRI image applied to a VRay Dome Light placed outside. I've changed the gamma of it inside the VRayHDRI node to have the strongest light and shadow contrast. VRay Plane Lights were used inside the room to enhance the details.
I don't care about having a strictly physical lighting; instead I use lights to "paint" the scene and convey the mood I want. I also use some invisible planes to "model" the light and bend it to my needs.
Step 7: Scene management
Having a lot of objects in the scene can be a problem, which is why it's important to divide the scene into layers. This way we can always keep track of our lights, basic room layout, details and so on. I usually use a file for scene assembly, which is the final file, and several files where I do my look-dev and texturing. This is really useful when we want a clean workflow.
You should really learn about V-Ray's DMC sampler and how to use it to speed up your renderings. Test renders can be done very quickly if we know what we're doing, and the final render time will be nothing compared to the time saved during the tests. V-Ray 3.0 has a great sampler called the Progressive Image Sampler, which can show a decent result in a matter of minutes while we're testing. The same goes for VRay RT.
Step 8: Post-production
To achieve the best result in post-production, you should really use MultiPass compositing technique. I keep my project in linear space until the final render (gamma 1.0). In the V-Ray options I use color mapping with gamma 2.2, but instead of burning it in the image I use it just to do the calculation and reduce noise.
The final render was saved in OpenEXR 32-bit color depth to have the maximum color range available. The OpenEXR format can include in one file all the passes we need for compositing purposes. I used V-Ray raw and filter passes to reconstruct the beauty render, plus a VRayEnvironmentFog pass done afterward.
In Fusion, I set my LUT to Gamut view with 2.2 gamma. Before saving, I disabled my LUT and placed a Gamut tool with gamma 2.2 at the end of the flow. (Note that in the first image the LUT is active, while the second, final one is the result of the Gamut tool.) This way I kept a linear workflow until the end of the project and saved my image with the standard sRGB profile applied to it.
You will find out for yourself the benefits of using a pure linear workflow when it comes to post-production.