Sunday, May 1, 2011

FInal Project - RENDERING CRAFT: A DIGITAL EXPLORATION

ABSTRACT

This digital design exploration involved an investigation into high quality rendering techniques for furniture textiles and upholstery. Many 3D modeling programs do not have the built-in capabilities for photorealistic renderings. However, the 3D modeling program 3D Studio Max enables designers to produce photorealistic depictions of interior product materials. The realistic depiction is in large part due to the programs dominant use amongst current day animators and video game designers, where there is a push for more realistic illustrations. For this same reason, it has been a successful 3D modeling tool for the furniture design industry. Well executed renderings of designer Patricia Urquiola’s work were developed in 3ds Max, which is why I chose to explore its capability to help portray one of my full scale chair design projects, Cocoon.

I discovered answers to many of my design decisions through the physical making of the chair. However, after its construction, it was realized that multiple design decisions could have been made based on a digital model’s visual representation of the chair. Still, the addition of this exploration and digital process into the design development helped to establish an overall design package for the future marketing and visualization of the chair.

For this current research, the 3D modeling of the chair was explored primary as a postproduction design method. My future research will test it as a preproduction and explore additional 3D modeling methods to continue the investigation of rendering digital craft.

The purpose for my research was to investigate 3D modeling techniques for high quality photorealistic digital renderings of handcrafted, non-rigid, highly detailed, and voluminous chair surfaces covered in furniture upholstery textiles. The exploration stemmed from my thesis topic regarding incorporating fashion apparel fabric manipulation techniques into the fabrication process of a chair’s upholstery and structural systems. The initial goals were to use the digital process to advance my design thinking on the role of a digital model for small production design and include it as a visualization tool for preproduction. However, I completed the chair, Cocoon, early on during the digital exploration of the 3D modeling program. As a result, the research became a postproduction visualization tool.

Figure 1. Cocoon Chair. A full-scale final prototype that was later 3D modeled and rendered with the program 3ds Max.

Figure 2. Cocoon Detail Image. A close-up view of the handcrafted details that need to be 3D modeled.

I chose 3ds Studio Max (3ds Max) as the 3D modeling program to research a photorealistic rendering of the handcrafted Cocoon chair. Many 3D modeling programs do have the built-in capabilities for photorealistic renderings. However, the 3D modeling program 3ds Max enables designers to produce photorealistic depictions of interior product materials. The efficient realistic depiction is in large part due to the programs dominant use amongst current day animators and video game designers, where there has been a push for more realistic depictions. For this same reason, it has been a successful 3D modeling tool for the furniture design industry. Well executed renderings of designer Patricia Urquiola’s work have been developed using 3ds Max, which is why I chose to explore its capability to help portray one of my full scale projects. Ultimately for my work, 3ds Max aided in the simulation of textile and upholstery textures, forms, and features. The goal was to gain as much insight as possible from each step of the process to create a rich and aesthetically beautiful image of my Cocoon chair.


Precedents

The Smock chair by Patricia Urquiola was depicted with a rendered modeled created in 3Ds Max Studio by the website Turbo Squid. Overall, the visual presentation of the handcrafted smocking technique was successful. There are some differences in the visual patterns of the 3D model and the full-scale works, but the overall rendered depiction was extremely close to illustration the pattern details.

Figure 3. Smock Chair Rendering. A 3ds Max rendering by turbosuqid.com of Patricia Urquiola’s chair that incorporated the fabric manipulation technique of smocking. Moroso Smock Chair by VizPeople. (n.d.). Turbosquid. Retrieved April 27, 2011, from http://www.turbosquid.com/3d-models/maya-moroso-smock-chair/433767

Figure 4. Smock Chair, the Physical Model. Smock: Designer Patricia Urquiola. (n.d.). Moroso. Retrieved April 12, 2011, from http://www.moroso.it/home_moroso.php?n=products&model=143&l=en

LITERATURE REVIEW

Art-based rendering

The construction of my chair was based on handcrafted sewing techniques that help to create voluminous upholstery fabric patterns. Therefore it was important to address rendering techniques that simulate handmade art-based projects.

Authors Kowalski, Markosian, Northrup, Bourdev, Barzel, Holden, and Hughes discuss translating art and illustrations into a 3D model. In their article, they work to address the creation of algorithms and software protocol referencing the rendering of stroke-based texture and a program interface based on free form modeling that lends itself to altering and assigning procedural textures. They express their goals as, to allow for an artist’s style to be depicted in renderings and knock down the scene complexities by coinciding rendering with modeling.

OpenGL is the system Kowalski et al. use for their software framework. The authors explain that the system lets designers apply textures to surface regions. Color and ID reference images are incorporated into the system, “these are off screen renderings of the scene, subsequently read from frame-buffer memory to main memory and made available to the procedural textures” (Kowalski et al., n.d., p. 3).

Kowalski’s et al. graftal textures focus on aesthetic effect. It is in this portion of the article they begin to address the rendering of fur with reference to graftals. Graftals superimpose a 2D image on to a 3D surface and the image is comprised of points (Kowalski’s et al., n.d.) The authors state that graftals must be assigned using controlled screen-space density that aesthetically mimics a specific texture, while also achieving depth through parallax and generating interframe coherence.

To meet these requirements, we have adapted the “difference im- age” algorithm (DIA) used by Salisbury et al. [14] to produce pen- and-ink-style drawing from grayscale images. Their algorithm con- trols the density of hatching strokes in order to match the gray tones of the target image. For each output stroke drawn, a blurred im- age of the stroke is subtracted from a “difference” image (initially the input image). The next output stroke is placed by searching in the difference image for that pixel most (proportionally) in need of darkening, and initiating a stroke there. The resulting image con- sists of marks whose density conveys the gray tones of the original.

Surface Design Modeling

In order to achieve a high quality photorealistic rendering of the Cocoon chair, it was essential for me to consider the techniques I used for surface design modeling. Wenfeng, Zhenyu, and Dingfang’s article regarding 3D modeling surface design helped to reinforce the importance of rendering an object’s surface and it highlighted the methods involved in the process. The authors emphasized the necessity of marring surface and structural designs.

Wenfeng et. al (2004) discuss the roles of appearance, figure simulation and aided design, and texture mapping. “Appearance refers to the surface configuration of an object, especially a commercial product, and it is existing physicality” (Wenfeng et. al, 2004, p. 268). The authors describe figure simulation and aided design as embodying the attributes and beauty of the desired material. In addition, they stress the importance of collecting, processing, synthesizing, and simulating images and material textures for the success of figure simulation and aided design. Wenfeng et. al (2004) note inventor Catmull’s 1970s introduction of texture mapping into 3d modeling programs. The article defines texture mapping as a technique that generates details that mimic the surface of a material.

Deformable Geometry Maps

The surfaces of the Cocoon chair’s inside seat and outside back are entirely comprised of undulated shapes, the inside a fabric surface, the outside resined fiberglass. In the physical model the undulated surface appears plush. In order to achieve a closely identical surface from the physical model to the 3D model I search for articles relating to realistic renderings of fabrics. Liu, Prakash, and Srinivasan’s article Interactive Deformable Geometry Maps helped to answer many reason why I had difficulty 3D rendering the chair’s undulated surfaces. The article discusses parameterized meshes and NURBS surfaces for simulation of textile surfaces in 3D modeling programs. They note the issue of the difficulty in distinguishing the location of mesh points and contact points. As a result, they stress the importance of the modeling programs to be interactive with the users. In addition, Liu et. al (2006) identify the necessity of being able to place constraints on mesh surfaces. They explain that when a user moves a selected point on a mesh, the typical result yields movement of random additional points to various locations. They propose the idea of setting constraints and collision detection on mesh points in order to successfully 3D model non-rigid surfaces. Liu et. al describes applying constraints so that only a certain perimeter points around the selected contact point can move with it and at a specific distance and direction to that point. The authors put forward the idea of collision detection to demonstrate the advantages of being able to identify when the movement of a contact point collides with an unwanted point or surface. During the time of my research, I was unable to determine if 3ds Max included such features.

METHOD

Learning Curve

Initially, the methods for developing a 3D model in 3ds Max, included: (a) learning 3ds Max, (b) hands-on experience with pattern making and sewing, (c) fabric and substructure material selection, (d) sketching design ideas, (e) building a full-scale model for reference (f) hand rendering, and (g) designing the product in 3ds Max. I was to learn the program’s interface, in addition to modeling, materials, lighting and rendering tools and techniques. However, from the project workflow diagram listed blow, it is apparent that some of the steps were changed and some were added.

Figure 5. Project Workflow Diagram. A comparison of the anticipated workflow to actual final workflow.

Assumptions

My assumptions were focused on the capabilities of 3ds Max for modeling, design revisions, and designing simultaneously or before full-scale fabrication. I was presuming that I would only make use of 3ds Max and that the program may include rendering and material limitations that I am not yet aware of, but will be once I’ve begun modeling. I assumed revisions to the design must be made once I created the product three dimensionally as compared to when it was sketched or hand rendered, or even when small mockups were done. Also, I considered my 3D model to be an initial design tool before final development of the physical chair.

Interpretations / Assessment

The assessment was not on the accuracy of the geometric dimensions of the upholstery pattern, but rather based on how well the photorealistic rendering captures the visual concept of the hand sewn fabric manipulation method. This was determined by how well the pattern was mimicked in the 3D environment from the real world. Thus a comparison of the 3D model was made to the completed full-scale model.

There was tremendous progress from the first test run of the 3D model in 3ds Max to the final. The advancements were with the visual representation and my understanding of the program.

Figure 6. Test. The first test run in 3ds Max for the project. Experimentation with lofting and texture mapping.

Figure 7. Second test. A second attempt at creating a photorealistic rendering of the Cocoon chair’s undulated surface. Experimentation with bump maps, bit maps, and an editable mesh.


Additional Computer Programs

A tutoring session with another graduate student presented the idea that much of the early form development of the 3D model could be accomplished easier with using and additional program. A major addition to the development of my research was incorporating Auto CAD into the design development of the 3ds Max model. The role of Auto CAD in the research was to establish a lofted splined surface, which I later imported into 3ds Max. Auto CAD allowed me to import photographic images of the finished physical full-scale model, draw reference lines on its profile and faces, and loft the splines together.

Figure 8. First Auto CAD Splines. First attempt at drawing splines on image in Auto CAD.

Figure 9. First Auto CAD Loft. The first result of lofting of the splines in Auto CAD.

Figure 10. Second Auto CAD Splines. The second attempt of creating splines from a photographic image. This time number of splines were increased.

Figure 11. Splines from Second Attempt. An Image of the splines for the top and bottom surfaces.

Figure 12. Second Set of Lofted Splines. On left a lofted image of the inside seat, on the right, a lofted image of the outside chair back.

Figure 13. From Auto CAD to 3ds Max. The imported inside seat image from figure 12 with added texture bump and bit maps.

Photoshop was another program essential to the development of the 3ds Max model. In order to create texture bump maps for 3ds Max, I had to import the images of fabric and fiberglass materials into Photoshop. In order to reveal a detail teture map surface in 3ds Max, the brightness and contrast of the images were adjusted in Photoshop. The images were then imported into 3ds Max as bump maps, and a slight blur was added to knock down the high level of grittiness. These images along with the texture bitmap images of the materials rendered results closely related to the physical materials.

Figure 14. Inside Seat Materials. On left scanned fabric bit image, on right grayscale fabric bump map image.

Figure 15. Outside Seat Materials. On top fiberglass photo bit image, on right grayscale fiberglass photo bump map image.

Limitations

There were several limitations I encountered. Roadblocks occurred when I struggle to figure out why a particular operation I’m performing is not working. This was the case when using the smooth modifier. Though the modifier was successful in rendering smooth edge treatments for the inside seat of the chair, the outside fiberglass surface rejected the operation. Another limitation was the in working with an editable mesh and NURBS in order to create a 3D model of the visually non-rigid chair form. I experienced the problems authors Lui et. al (2006) described in their article. These inability to distinguish and select specific points, the collision of points and surfaces, and the necessity to move each point individually to create the undulated surface. In addition, there was a time constraint for the development of the 3D model. During this time, in order to determine how the hand-sewn technique would appear, not only did a 3D model need to be built, but also I first had to work on mockups of the pattern and construct the final chair form to determine its overall shape.

RESULTS / OUTCOMES

Initially the process was difficult. The majority of the issues were based on the fact that 3ds Max Studio was I program I had not begun to learn until my current research. The first test 3D model I composed centered on displaying the pattern with bitmaps and bumps. I was able to scan in a mockup image, grayscale it and manipulate the contrast in Photoshop, and then bitmap and bump the mage in 3ds Max. The issues of working with NURBS, assigning different image patterns on the front of the seat and back of the seat, and attaining the proper color were issues I struggled to overcome.

Figure 16. Final Renderings. Final rendering perspective views of the chair’s inside seat, outside back, and base.

As I moved forward in my exploration, I discovered the result of marring the Auto Cad and 3ds Max allowed me to work with a 3D form, drawn from my photo images of the physical model, in 3ds Max from the beginning. By the time of the final iteration I had overcome the initial problems of texture mapping and assigning materials. The focus turned toward problem solving issues of modeling photorealistic shapes for the undulated surfaces.

CONCLUSIONS

Observations / Reflection

My comparisons of Patricia Urquiola’s Smock chair 3D model, literary research, and hands on physical and digital model explorations created an awareness for me as a designer. The research shined light on the fact that even though my initial reaction to 3D modeling was to depict every aspect of an upholstery pattern accurately, I saw it as not being required for the final rendered outcomes success. Instead, the importance was in creating a visually similar illusion of the pattern.

I became more familiar with 3ds Max and the advantages of pairing it with an additional program, like Auto CAD. I feel that by starting with a program I was already familiar with, Auto CAD, I used my time more efficiently. In addition, I was able to focus more on the modeling techniques of 3ds Max that I hope to explore at the beginning of the project. One technique I had more time to focus on was manipulating the surface of the model by working with 3ds Max modifiers.

Since the Cocoon chair 3D model contained undulated surfaces on the front and back, it was important to add and place proper lights. The lights needed to reveal the contours of the back surfaces and the seat concave seat. A spotlight was directed at the back of the chair and positioned low, yet slightly angled upward. As a result a variation of low and highlight helped contour the surface. An omni light was placed above the chair and positioned of center from it. This created lowlight shadows to emphasize the concave seat.

Also, reflecting on the project from my initial exploration to its final development enlightened me on additional significance of the application of the 3D model. I moved from viewing the model as a visualization tool for me as the designer, to the idea of incorporating it as tool for clients to visualize various materials and their properties on a chair they would not otherwise have the opportunity to physically see. As a result, the ability to successfully render a high quality photorealistic product image, is a valuable marketing tool.

The Future

The future research of the project reflects on many aspects of my exploration that I feel need further development. The areas to reinvestigate include working with a polygonal model, the smooth modifier, and bitmaps. The polygonal model would allow me to approach the 3D model development in a manner that may prove more successful. The continued research of the smoothing modifier and bitmap cold help me to nail down a more successful surface, texture, and finish for the Cocoon and furture chair projects.

In the future, I envision a program where I could create the pattern in real-time, with my hands, needle, and thread, just as I did with the full-scale model. Hopefully the current emphasis in society on crafting will help to identify the need for 3D modeling programs or program capabilities to realistically render craft.


REFERENCES

Kowalski, M., Markosian, L., Northrup, J., Bourdev, L., Barzel, R., Holden, L., & Hughes, J. (n.d.). Art-Based Rendering of Fur, Grass, and Trees. SIGGRAPH 99 conference proceedings.

Liu, Q., Prakash, E. C., & Srinivasan, M. A. (2007). Interactive Deformable Geometry Maps: Efficient Modeling for Interactive Deformation of Non-Rigid 3D Objects. Visual Comput, 23, 119-131.

Wenfeng, L., Zhenyu, W., & Dingfang, C. (2004). Modeling and Simulation of Product’s Surface Design. Computers & Industrial Engineering, 46, 267-273.



APPENDIX A

PRESENTATION SLIDES











No comments:

Post a Comment