Making of : Draped Art
The Project
The project is an art foundation in the outskirts of Strasbourg. It’s a personal project I worked on as a student. The building is composed of an underground part with a library, museum and lecture hall and a copper-clad tower emerging from the forest around it with apartments for the residents as well as conference rooms. The verticality is ubiquitous in the concept hence the vegetal columns slicing through the space and bringing light. Materials wise, what is interesting here is the discrepancy between organic shapes (roots, soil, drapes) and the more rigid parts (concrete walls, wood floor, glass). It creates a nice interaction between light effects and hues, the core of the image here. Here’s how I proceeded step by step. Software used in this project was SketchUp, V-Ray for 3ds Max and Photoshop.
3D Model
The 3D model was done mostly in SketchUp. Even though this software is not the best when it comes to exporting to external software for rendering, I actually found that there can be quite an efficient workflow once you know what to be aware of. You might have another workflow here, but if you’re looking for an efficient and simple one, here is how I proceed :
1. In SketchUp, model your project with as many details as you can but don’t insert furniture or high poly objects yet.
2. When in monochrome mode, check that all your faces are facing outward (look white). If they are blue, it means they are backward. To correct this, just right click on the face and choose reverse face.
3. Differentiate all the objects that are going to have a different texture later on. To do so, create a material (a simple color, the uglier the better), give it a name, and then apply it on the appropriate faces.
4. File>Export>3d model.
5. Export in .3ds and in the options tab choose “export by layer”.
6. Import into 3ds Max. Once in 3ds Max, you’ll see that all your objects (meshes) are imported with the materials’ names (quite handy) and also they’re grouped in the same way they were grouped in SketchUp. Only then can you start to import furniture and high poly objects. Don’t forget to use instance and not copies so that you don’t use too much RAM for no reason and reduce the render time at the end of the day. The drapes were made using a cloth modifier with a simple plane set as a cloth and my furniture set as a collision object. Also, the floor was made using the Floor Generator plugin by CG-Source. Really efficient and easy to use. A must have for 3ds Max users.
Texture the Model
Texturing the model was also pretty straightforward. Just don’t forget to apply a uvw map modifier on each of your mesh so that you can scale and place your texture properly.
Floor Material
The floor was done with the Floor Generator plugin as mentioned earlier, and I used the MultiTexture Map with several wooden maps from cg-source as well. A bit of reflection was added (glossiness set to 0.6) as well as a bit of bump map.
Glass Material
The glass material is a simple V-Ray material with full reflectivity and refraction. A greenish tint is added as a fog color (R:54 G:61 B:45) and the fog multiplier is set to 0,005. This basically creates the thickness of your glass when visible (typically visible on handrails or coffee tables) Finally I generate different noise maps (by changing the phase in the noise parameters) and use them as bump maps on my glass to distort the reflection a little. I then plug them to a Multi/Sub Object and apply this object to my glass panels. Don’t forget to add a material by element modifier on each of your panel and set the frequency of the material to 25.0 (since there are 4 kinds of noise) so that 3ds Max knows how to apply the aforementioned material.
We could go further into the realism, but we already have quite a realistic glass effect there.
Drapes
The drapes are done using a VRay2sidedMtl but plugging only one material and leaving the other one blank so that we get this translucent effect. There is no diffuse map, only a refraction map (cloth texture) on a falloff map. You just need a bit of tweaking on the falloff map to get the exact translucency you’re looking for.
Soil
Nothing fancy here, just a displacement map directly taken from the diffuse map.
White Walls
No diffuse map here, just a little bit of noise as a bump map to make it a bit irregular.
Concrete
Diffuse map, subtle bump map and a reflection with glossiness set to 0,65.
Render the Base
The base is not really fancy to say the least. Using a VraySky and VraySun I can get nice cast shadows which is actually the only thing I really need since most of the light effects and color work are going to be done in Photoshop later on. Some render settings :
- GI + Irradiance Map (min rate : -3 ; max rate : 0) + Light Cache (600 subdivs)
- Resolution : 2400×1800
- Render additional passes (reflection, refraction, zdepth and materialID passes)
- Material subdivs are individually set to at least 16 subdivs or 32
- Render time : a bit more than 2 hours
Photoshop – tweak errors
Before even beginning to work, I always turn my image into black and white mode by placing a hue/saturation adjustment layer on top of everything, lowering the saturation to zero. (you can also use a channel mixer, or a gradient map, it doesn’t matter that much). The idea here being that colors are always gonna be off in your base render, but there’s no point in focusing on colors if your image is not even looking good in black and white. As long as your image doesn’t look good in black and white, don’t even bother thinking about colors yet. I don’t like spending much time tweaking my model and render to be perfect on 3ds Max since most of the time it would take just a minute to adjust it in Photoshop. Here for instance, there was a couple of things missing or just not clean :
- The soil columns were not looking good (at all) on the edges, since the displacement map only work in one direction. Using a stamp brush tool, I cleaned this area in a few seconds.
- Adjust the drape with the smudge tool to make it look more natural. Even though there was many subdivisions, you can still see a bit of the polygons which is definitely not what you want on organic shapes like these.
Add Details
Vegetation
Even though displacement maps can sometimes do wonders, it was among my first attempt here and there was plenty more to do there before it actually looked as expected. The idea here is to have a column of soil and vegetation cutting through the space. The upper part would be of vegetation, roots and whatnot and the lower part more rocky. Roots are cut out from photographs of mine or from the internet. Mainly using the pen tool, you don’t have to be that accurate since you can just blend it with a soft brush, painting black on your vector mask. Also don’t forget to adjust the levels so that it blends in properly. You don’t want the roots to be lit from under when the light is coming from above. Colors will be taken care of later on. The ivy is from my vegetation library and is just a png file with transparent background. Nothing fancy in terms of technique here so, just the idea of making it look more random and natural, in order to contrast with the rigidity of the suspended ceiling grid.
Wall Art/Photographs
I had to add the little detail of the string detectors thingy in front of each photograph or painting. Also, putting in the photographs is pretty straightforward. Just drag and drop image in Photoshop, distort using Ctrl+T, right click and distort or perspective, and add a vector mask to hide what needs to be hidden, nothing fancy here.
Adjust Lighting
In this project lighting was paramount, not only because it is a museum, but because we have such a discrepancy between materials that light is going to behave in distinctive and interesting ways whether it is on a really smooth surface or a rocky one. The V-Ray sun just gave me the cast shadows I needed, and making the sun bigger (size multiplier in the modify tab) I had less sharp shadow edges which is much more realistic and soft.
To make the lighting more interesting and less dead, I create subtle rays of light in the room. First, you need to be aware of why we can see rays of light. Often we see renders with massive beams of light whereas there’s no reason for this phenomenon to even exist in the scene. For beams of light to be visible, the atmosphere has to be filled with dense clouds of particles (either water droplets, or dust particles). If there is no such thing, then you shouldn’t even consider using this effect. Here is a quick method to do that. This is the pure Photoshop technique…..here’s how to proceed:
- Select the area supposed to be lit with a polygonal lasso tool (it doesn’t have to be that accurate).
- Create a curve adjustment layer (or level adjustment layer, whatever you’re more comfortable with)
- The area not selected is automatically masked by a vector mask
- Tweak the curve to make the light brighter (don’t hesitate to make it a bit too much, you’ll adjust it later on if need be)
- Add a gaussian blur (filter>blur>gaussian blur) to blur the edges of your light and make them super soft
- Depending on the source of light, you can redo the same procedure but reduce the area of your light, blur a little less, and increase the light intensity. This is more suitable for an artificial light.
- Coming back to my first remark, to add a bit of realism you need to do two more things
- Paint dust particles in lit areas. These particles, depending on their size, are the ones catching the light and making it visible. You can either find particles brush on the internet, or just tweak your brush presets (scattering, opacity, angle, etc.) to get the same effect.
- You can add clouds in the vector mask of your light to simulate a more random propagation of light in the space.
After that, I can take care of ceiling which was supposed to be backlit, so I adjusted it using a level layer adjustment and made it brighter. The drapes needed to look more spectral or surreal, therefore I pushed up the white a little using a level adjustment layer here too. I also needed to bring more light in the library in the background. Finally, to add a little bit more of contrast I also used the burn/dodge method. To do it in a non destructive way, just create a layer filled with 50% grey and set it to overlay. I usually duplicate it, one for burning the other one for dodging. Then just put this layer on top of your image and work on these layers instead of the actual render base.
Lights get more intense and you can create bloom effect on edges.
Populate
As you might have noticed if you watched the video, I first had more people in the image but it felt overcrowded and the sense of calmness was not really there anymore. Therefore I only kept a couple of the former cutout. The woman guarding the museum, a couple of people in the library or the mezzanine. And the child in the foreground with her busy mother (potentially) in the background. We could almost remove the girl on the right side of the image but then the right part would really feel underpopulated so I kept it. The key here is about storytelling, and not overcrowding the place so not to overshadow the architecture. Quite a tightrope to walk on. In this regard and also from a more technical point of view, here is a complete breakdown of the procedure I usually follow when I populate my render. There’s even a script freebie at the end to make you save time on the next time you have to populate your renders. One little thing in this image : the boy’s shadow. Be careful with the junction between horizontal and vertical planes since the deformation is going to be different. On horizontal plane the shadow is stretched whereas on vertical plane the shadow is a simple translation of the object. Another final thing to integrate is your people properly : don’t forget to render your reflection pass in whatever render software you’re using. This way you can put this layer over people who are behind glass. Set this layer to something similar to soft light, and adjust opacity. The effect has to be subtle, but has to be there to make your render more realistic and your people blend in more properly.
Adjust Colors and Saturation
Now that the render looks pretty cool in black and white, we can deactivate the black and white layer and start focusing on saturation colors.
Saturation
As we can see, the image is way too saturated. To correct this we have to understand that an image can be divided in three values: Hue, Saturation and Brightness (or Luminosity). We’ve seen in the previous tutorials that we can isolate Hue by neutralizing Luminosity. To do so we filled a layer with pure red (R:255 G:0 B:0) and set its blending mode to luminosity. In order to isolate saturation, we have to neutralize hue as well. To neutralize hue, we have to fill another layer with pure red (R:255 G:0 B:0) and set its blending mode to Hue. Hue and Luminosity being neutralized when we turn both of these layers on, the only information left is the saturation. To bring it up a bit more, we can use a hue/saturation adjustment layer and increase the saturation so that the discrepancy between different levels of saturation are emphasized.
Using hue/saturation adjustment layer, we can start to change the saturation of each area that need to be reworked. Namely, the overall image, and then each cutout people.
Here is the image without the saturation pass. It looks much softer, but still we have to work on the colours now.
Colors
Saturation has been taken care of, we can now proceed to the color adjustments. As usual, I make a little color palette and put it on top of my image to know which kind of hues I’m aiming for. To isolate hue and work in a more “scientific” or “pragmatic” way at first, here’s the method I use :
- Create a layer filled with pure red (R:255 G:0 B:0)
- Set it to Luminosity blending mode
- Add a hue/saturation adjustment layer and increase a bit the saturation.
- Put these two layers in a group and move this group over the whole image in your layer hierarchy
You now only see the hues of your image. It now becomes much simpler to adjust each area of your image to match what colors you were looking for. In my case, the idea here was to play with shades of blue, orange and a bit of green instead of my base render being mostly just orange. To change the hue, I just apply a hue/saturation adjustment layer (all grouped in a single COLOR ADJUSTMENT group) and paint the area I want to change the color of in my vector mask.
- The wood flooring was supposed to be a bit orange
- The soil column mostly in the green hues
- All the walls and drapes should have blueish hues.
Be careful that areas you’re working on are coherently segmented. Depending on the orientation of the wall in our image for instance, it will either have a blue hue or a more orange one. Be careful not to simply fill an area precisely (using alt+delete to fill for example). You rather want to use a soft brush and paint areas in a more subtle way. Keep in mind that transitions in hues are never abrupt, there is always a tiny gradient somewhere in between two different hues since the colors of close objects will be affecting the color of each other.
Adjust Composition with Crop Tool
The image is done… or is it? When I adjust my camera in 3ds Max, I tend to render larger area than what I would actually need, just to be safe. Here, the composition was good but there was something not really convincing about the way the image was cropped. On our image before cropping, here is what was happening :
- Horizontally everything is working all right. We have one third on the right with its own atmosphere. The soil column makes a good transition and the perspective closes the scene on the left while the light penetrating on the left makes it the center of the image.
- Vertically, it is not quite there. Everything is a bit floating. The dark area at the very bottom is not interesting, the very upper part as well is not that good since we have a little bit of the spotlight on the right which is a bit weird
Taking these things into account, we can notice that we could rely on an interesting composition trick which is the “frame within a frame”.
- As I said earlier, the soil column acts as a partition in the composition, and so does the darker area in the middle. Playing on this composition, we can crop our image so that we keep the same ratio between the scene on the left with the boy in the light, and the overall image.
- This way we can get rid of the upper part and the very lower part. We also have to tweak just a little the drape on the right to take away the dark part which is breaking our light composition.
We have two composition frames. What is under the main rectangle is dark and takes away the attention (we can crop it but we have to stamp tool the darker part of the drape on the bottom right corner), what is above the main rectangle is also of negligible importance therefore we can crop them to have the same ratio between the smaller and the wider frame.
Now we can call it a day! I hope you enjoyed this making of! There will be many more to come so don’t forget to subscribe to our newsletter and like our Facebook page! For those of you who haven’t seen it yet, you should check out the breakdown video for this render here to get extra details. Also C&C are always more than welcome so that I can improve parts that are still unclear to you.
Thanks for reading!
Simon Oudiette
Simon Oudiette graduated with a Masters degree in Architecture from the École National Supérieure d’Architecture de Strasbourg. In 2015, he founded architecture visualization studio Horoma. To see more of Simon’s please visit his website Horoma.