The making of 'A cup of tea?'
Yu Sun works through the process behind his traditional artist-inspired image, made using 3ds Max, ZBrush, Photoshop, and V-Ray…
Sometimes I wonder what the characters of great artworks from history would do in the modern age. So I created this picture: A cup of tea? I wanted them to look like they were experiencing the charm of Chinese Kung Fu tea in a Chinese teahouse, but the man who makes their tea looks like a newbie.
So, next I'll describe the creation of the artwork and talk about my general production process.
Planning time and collecting data
A cup of tea? is my latest work, and it took about two and a half months in total to finish. Most 3D practitioners know that planning a cycle to control the quality is important. We can't avoid the problem; all we can do is take a short amount of time to plan, in order to achieve the quality and efficiency of a win-win situation. So when I started this project, I set up an overall time estimate and began to plan a general plan as follows:
• Weeks 1-2: (The early stage of the design, collecting data)
• Weeks 3-4: (The early modeling, sculptures of the high poly model)
• Weeks 5-8: (Attention to detail , texturing)
• Weeks 9-10: (Hair, rendering, post processing)
The first stage, I mainly figure is more in accordance with your own goals, so choose photographs and inspirations as you prefer for your image. Then you can draw a rough design. If you prefer, you can also use 3D software such as 3ds Max or ZBrush to rough out a simple simulation of the human body.
Because the action in the scene is a bit much and I can't take on all of the accuracy and integrity of each individual all at once, I go by the philosophy ‘divide and conquer'. When I reach a certain point with each model, I set the benchmark and continue onto the next one.
So to start, I began with the production of Van Gogh and his sunflower. Because this is a famous art painting, the search is reveals relatively good data. At the same time though, the production is made more challenging because the image must match that data perfectly.
Modeling Van Gogh
The key to modeling the sunflower is to use the Move brush and sculpt from the inside to the outside. The key to modeling Van Gogh is depict the feeling of the Van Gogh's nervousness, and show that he wants to protect his
The pores of the sculpture were made with the DamStandard brush with four kinds of alpha. I draw these out by hand, and then carved them into the face using reference images. Although it takes some time, it can create a very
Modeling Van Gogh's clothes
The production of Van Gogh's clothing as a whole was a little different. For this image I tested some new software called Marvelous Designer (MD) to create the clothing, but because it was my first attempt, it did not create a particularly good effect and actually increased the testing time a lot. All in all though, it's a really good software and is worth a try.
Now I can use MD smoothly though, and I hope I can enlighten you with its techniques in the future. This time though, because the software didn't work for me, I used the old way of finding a good reference to carve the clothing from. Where do you get the references from? You don't need to buy new clothes for this; the internet is full of clothing reference images you can browse through.
The Van Gogh model was then basically complete, so I created all the other models on the same principle. Though the other characters followed much the same pattern, I've outlined some of the more complex techniques used in the creation of their individual characteristics.
First, the sweater weave. The sweater was created with a curve in 3ds max. I then copied it out, baked the model and saved it out as a height map in ZBrush. I then imported the height map into Photoshop to create the pattern of the sweater, then pushed it back into ZBrush with the model, and started masking by color. Then I used an Inflat deformation with a moderate setting to create the woolen sweater effect. You could either follow these steps, or simply find a good sweater texture.
The dragon pattern on David's shirt was created simply by painting in the shapes, stroke-by-stroke. There is no simple way to speed this process up, and if you want to save time, you have to improve your proficiency.
After painting the design into a Bump map, you can use it to reapply over the surface, as we did with the sweater, if you need to.
The water was modeled using DynaMesh in ZBrush and then using the Move and Insert brushes to arrange
After I sculpted all the models to the same level of detail, I began mapping and applying texturing materials. The size of textures in my works is usually 4096x4096.
When texturing, my habit is to start the topology and split the UV when the model basically completed. I used the UV layout in Maya to unwrap the UVs, though for the few smaller items, I split the UV using the ZBrush UV master.
When the UVs were done, I began to paint the textures. I started with the Van Gogh model again. For the skin, I used the light and shadow functions in Photoshop to remove the shadows and highlights and then texture-mapped the skin in Mudbox (because it retains the same detail as ZBrush), modified it back in Photoshop, and finally repaired the
seams in Mudbox.
The Diffuse map must be very flat (with no light or shadow). That way you won't have a problem with the lighting and shadow when you render the image. Once the texture is finished, you can check the material effect and move onto adding the detail.
Adding the detail primarily consists of making the Bump map. The Normal map helps restore the general model, and the role of the Bump map is to add more height variety to the surface of the skin.
You can create a Bump map by desaturating the image, then adding curves and levels in Photoshop.
Materials and lighting
I used the VRayMtl shaders to simulate normal objects, and the VRayFastSSS2 shader to simulate objects with an SSS material effect. The image here shows the skin materials and cloth material settings. For the cloth, I used the Falloff in the self-illumination layer.
Test lighting and rendering
First of all, I found suitable HDR images, and added a test light rig as shown here.
You'll find that you might need to test light/render hundreds of times to get the look you want, but it's always worth it.
All hair was done using the Hairtrix (Ornatrix) plug-in in 3ds Max. The brush is very useful, and you only need to adjust its settings to 0.1. Before you start teasing out hair, you'd better plan out and know what kind of hairstyle you want to create. Don't forget to pay attention to the uniformity of the UVs too.
After the preparation work was done, I could start work on the final lighting test. The overall rendering, while it initially wasted some time getting the right one, had a good outcome in the end. Once I'd rendered out the model, I rendered out various channel layers at over 5000 pixels each.
Because I already knew at the start that I wanted a certain effect, I could move into Photoshop with a clear idea of my final look. First, I made a simple fog effect, and worked on the depth of field, air effect, etc. I also adjusted the color, contrast, brightness adjustment levels (see the image here).
I also adjusted the sharpness using the Unsharp mask, and compressed the image size (to fit the recommended storage format for the web) before sharpening again.
After that, my entire production process was over. Thanks for reading about my experiences with this project and I hope everybody can learn something from it.