Making of Making Of Ø-616 Recovery Center by Milan Stevanović.
About Milan Stevanović.
My name is Milan Stevanović. I am a young Serbian architect and freelance 3D artist, with a soft spot for Scandinavian architecture and lifestyle. I got my Master’s Degree in Architecture from Niš University in 2013. Prior to my graduation I was involved in numerous projects in Serbia and abroad, from small residential houses to government project such as Science and Technology Park in Niš. I am equally skilled for working independently and on a team and I’m open for collaboration as an architect as well as a 3D artist.
First of all, I would like to thank CG Workshop for giving me an opportunity to showcase my work throughout this making of. It is both a great honor and pleasure for me. Ø-616 recovery center is a project done for a client in association with my partners from 3d-norge.no, with whom I hope to collaborate for a long time. Project is a reconstruction and expansion of an existing building and its redevelopment into a recovery center. Plans and facades were provided by the client, as well as information about materials in the scene. Client wanted a warm evening scene.
The project was modeled and rendered out using 3ds Max and Vray, and postproduction was done in Adobe Photoshop.
In this making of I will be more focused on post production part of work then on working in 3ds Max, because for me that is where the magic happens.
Having drawings modeling wasn’t especially hard part. Building is more or less simple, orthogonal in its base. After importing a line drawing into 3ds Max base geometry was created with simple editable poly modeling and lots of slicing and shell modifiers. After finishing base model geometry I started modeling details. I used Floor Generator script for modeling white wood cladding on the facades. Since Floor Generator isn’t able to work on vertical surfaces I had to detach as clone sides of facades and rotate them by 90 degrees so they became horizontal surfaces so could apply Floor Generator script and create cladding. After that I just repeat process backwards to put cladding at its place. Same principle was used for modeling a small red building in front of a main building.
The ext step was modeling windows and doors, wooden details around the cladding and adding gutters. With that the modeling of the main building was basically done. All that was left to do was surrounding and some interior details. Since I couldn’t gather as much information as I needed from plans I got, I used Google Maps to get more familiar with the site. Wood planking in foreground was modeled out with a Floor generator also. For tree models I used recently released Laubwerk plugin and Itoo Forest Pro for grass.
Materials in the scene are very simple, noting too special about them. Traditional approach with diffuse, reflect and bump maps on most of them. Glass material is also straight forward, black diffuse color with almost white refract color and dark grey/white falloff for refraction. For wood planks material I used Multitexture Map Plugin for geometry done with Floor Generator and for none Floor Generator geometry I applied Unique Material ID script so that Multitexture can be used on those objects too.
I would like to point out one material in particular. It is a roof material. While working on a project I had a dilemma about whether I should model the roof tiles or just apply a material on a roof slopes. I wanted to achieve a bit old and used look of a roof tiles so I decided that is faster and more efficient solution to find suitable texture. I have found what I needed on CG Textures. After downloading the texture I modified it in Photoshop and made it seamless and darker a bit, and with PixPlant plugin exported Specular and Normal maps. Diffuse map value was set to 100, Reflection and RGlossiness maps value to 25 and for Normal map value of 30.
For lighting of the scene HDRI map was used for exterior lighting and Vray plane and sphere lights for interior. HDRI map that was used was made by Bob Groothius and was downloaded as a free HDRi from here.
Instead of using Vray dome light combined with HDRI map in it, I just putted the HDRI map in the Environment map slot. Reason for doing so was to get an Alpha map directly from the frame buffer since it will be needed in postproduction stage of work. When using Vray dome light you have to make it invisible in order to get Alpha map. Another reason for this is that if you use HDRI map in Environment map slot you get a lot softer lighting of the scene which is quite good for evening scene.
Render setting are pretty common Vray settings. I work in Exponential color mapping with Dark and Bright multiplier set to 1.0 and Gamma on 2.2. I use Adaptive DMC image sampler and Area antialiasing filter, which turned out to be a best solution for me and my workflow. As for Indirect illumination Irradiance map for Primary bounces and Light Cache for Secondary bounces. I tend to finish with rendering as quick as possible so I can get to the part that is much more interesting to me and that is post production. My goal is to get a good quality base render from 3ds Max that I can alter latter in Photoshop. Because of that I extract all of the render passes just in case.
Since a lot of work on my renders is done in post production I tend to render my raw images a little bit dark and not with too much contrast so I can keep more information in the pixels. You can easily brighten up an image and add contrast to it, but it doesn’t go that easy other way around if you are not using LWF. I save my raw renders as 16-bit tiff file. While working on projects over the years I came to a conclusion that, for my current work flow, this is quite enough.
This is my raw render. As I mentioned it doesn’t look much now, but that is ok for me because usually when I work I have some vision of my final image in my head and when I estimate that the result is satisfying I render hi res and go to Photoshop. Other times when I am not sure what I want to do with a picture I spend more time in 3ds Max experimenting and trying to get a better raw result.
My basic workflow is similar in every project. There are a certain steps in postproduction that repeat. First step was adding Ambient occlusion pass and in Multiply mode and lowering its Opacity an Fill value, and with it bringing out the details and stronger shadows.
After that I started working on a background part. I changed the sky, added some adjustments to it so it fits better with the rest of a scene. After that I added mountains in a horizon and blend them in. Last was adding an object in the back that I picked up from Google Maps.
Next was adding changes that will affect all previous work. Curves and Brightness/Contrast layers for bringing more contrast and brightness to the image (settings common), and a few Brightness/Contrast layers for details (settings details).
Now that I brightened up overall image I wanted to make lighting coming from the inside more notable. Before I did that I added a few concrete textures over the asphalt and used Free transform so they match the perspective. I lowered their Opacity and Fill and set them to Overlay mode. Then I increase brightness on some windows and added a glow. I also used masked out Vray Raw Lighting pass and Specular pass and set them to Screen mode to enhance light on tree and enviorment.
Stains and leaking details came next. Textures were, same as concrete ones, transformed and positioned with Free transform and then set to varies blending modes, depending on what suited the best. All these added textures were downloaded from CG Textures.
Final details that were added before flattening the image were people. All of them were added in Photoshop. Sometimes when there is a need for major number of people in the scene I try to add most of them in 3ds Max as 2d planes that are facing to camera. In this case there was no need for adding people in 3ds Max. For me personally adding people in post production is much better solution. Same thing is with stains and other details because adding them in Photoshop gives you much greater control over them. Once you position them you can, at any time, make them more or less notable, mask them and everything else without testing and rendering, and results are instant.
To make my work easier before I start adding people I export one additional costume pass from 3ds Max. What I do is import one 2d plane with opacity map in the scene and copy it multiple times on different spots in the scene. Plane is about 175-180 cm which is an average man’s height. This way I am sure that scaling of people in the scene is correct and everything looks natural. For extracting this additional pass I use very useful RenderMask script and for quick changing of wire color Random Wire Color script. Both of these scripts are huge time savers and I would recommend them to anyone who is doing a lot of postproduction in Photoshop.
After extracting the pass is added to Photoshop and set to Screen mode. After that people are added, scaled and adjusted. For adding people the main thing always is for them to have same kind of lighting and to make sure that light on people comes from the same side as in the scene. Smaller changes can be added but it is hard and time vaster trying to put cutout from daylight photography into an evening scene.
I will not explain every single cutout added. There are already a few very good tutorials about this, like VizPeople tutorials and maybe little less known by Piktoforma. Instead of detail explanation below you can find layer combination for the image of a boy on the right and see how they affect it.
When I finished adding people I flattened image. I used various Nik Software Color Efex Pro filters and combined them together with different Opacity to get better contrast and details and high pass over all that to sharpen the image.
Last step was adding sparkle detail in a foreground, final adjustments of an image and vignette.
To enchase sparkling effect for window light reflection in the water I used an image I found on Google. I masked it out and set it to Lighten blend mode.
Costumer wanted a warm evening feeling of the image so for my final adjustments besides Curves I added Hue/Saturation layer on Soft Light mode with very low opacity, Color Balance to give a little bit reddish look to the midtones and Selective color to add some more warmth.
Here is a final Layers combination so you can have a better look at it. I came to understanding that organization is very important when you are working with layers so before starting your work, make sure that you know what you want to do and where to put it.
In a conclusion I would like to add one thing, considering that there is constant debate about photorealism in architectural renderings. Being an architect myself, I tend to give as much information as I can with my renderings, and keep details clean and clear, and not achieve photorealism at all costs. I always found visualizations with a little “artistic touch” more eye-catching than a plain photorealistic render, and also think that an image should tell a story and have a certain context attached to it.
Thank you for reading this and I hope that some of thing above will help you in your work.