Gabriel Gibbon's profile

Projection Mapping Project: Big Head

The fusion of 3D physical models and 2D digital images create sculptures and artworks that blend and overlap the digital space and the physical world. Digital designs, renders, models and videos can be brought into the real world through projection. This project showed how real-life objects or people can be scanned into digital models, manipulated through computer programs like Blender and Solidworks and then recreated in the physical world through digital fabrication techniques such as laser cutting and 3D printing but most importantly at any scale. A head that can fit in the palm of your hand or a head that can take up an entire room. When these sculptures are projected back onto, they blend virtual space and reality seamlessly.
This project culminated in the form of a 2.5 meter (8.2 Feet) model of my face and head. This Behance entry will show the entire process and can be followed as a step by step guide, bearing in mind I have access to some incredible pieces of equipment through my university that would not normally be easily accessible. The final sculpture was built entirely by me and in order to learn its construction and the best method for making it a variety of prototypes and mock-ups were made. 

Sketching
Completed Models
Head Scanning
A 3D scanning company was used to scan my face. They used a handheld scanner and moved it around me. It emitted a harsh bright light as I was scanned. The final scan was exported as an STL. The scanner was unable to pick up facial hair and head hair. In order to create a full head model, the face scan was combined with an existing head model. Unfortunately, I am unable to locate the 3D model that I used, but it was downloaded from GrabCAD. 
I used Blender to combine the models and fix any imperfections. I used MeshMixer to further smooth and refine the model.MeshMixer was also used to fill the model to make it solid and suitable for 3D printing. 
Combining scans in Blender
Completed Model
Lasercut Cardboard Models
A method of bringing the 3D model into the physical realm is using a laser cutter. The software: Slicer for Fusion 360 allows the user to slice the model into a series of profiles that can be cut out of flat material and stacked onto of one another. 
Paper model made using Pepakura
Pepakura is a software that allows the user to ‘unfold’ a 3D model into a series of flat triangles (https://tamasoft.co.jp/pepakura-en/ ). These individual pieces can then be printed out, cut out and then refolded and stuck together to recreate the 3-Dimensional model in real life. While in principle this is very simple it is actually very difficult to do. Most 3D models (stl, obj files) tend to have lots of curves. Due to paper only being able to fold in 1 direction the software unfolds every single curve into a huge array of overlapping triangles that can’t actually be cut out and folded together 
To account for this the polymer count of the model needs to be reduced, ie the model needs to become simplified and more geometric. The term Low-poly is used. In Blender the command Decimate allows the user to simplify the model by specifying the number of polygons the model is made from. Too few and the model loses all detail and becomes an abstracted form, too many and the Pepakura file cannot be cut and folded. 
The assembly of these various separate triangles is a very straight forward process but is very tedious as accuracy is essential in cutting, folding and gluing. Small imperfections add up and can ruin the assembly as parts will no longer fit together. The model of the first head was comprised of 15 pages.
In previous Pepakura projects a craft knife and a metal ruler was used successfully to cut out parts but this is incredibly time consuming and not aways accurate. A modern electronic vinyl cutter is a far faster and more accurate automated solution. The Silhouette Cameo 4 vinyl cutter was used to cut out every part. The pages where first printed at Postnet on Heavy 8 cardboard at R8.00 per page. In order to align the cutting software and machine with the printed pages a software add on is required: Pepakura Viewer 4 for Silhouette Cameo 4. This exports the pdf file of the parts with registration marks in the corner of each page. These registration marks match the printed marks on the card pages. The software also exports the file as a DXF which the vinyl cutter is compatible with. The machine is then able to scan these marks to cut along the correct edges corresponding with both the dxf file and the printed card. Each page needs to be manually attached to the Vinyl cutter’s adhesive cutting mat and the correct file loaded. It took about 60 seconds per sheet as opposed to 1-2 hours of hand cutting it would usually take if a manual approach was taken. 
3D Printed Model
FDM 3D Printing (Fused deposition modelling) deposits a small layer of plastic on a surface through a nozzle. The movement of this nozzle and the print bed create a 3D part through the process of additive manufacturing. The Stl file is imported into a slicing software Cura. This software places the model on the print bed and helps generate supports for the model. The supports are plastic frameworks that help support overhangs allowing a successful print without sagging or drooping elements. These supports are designed to be snapped off of the part once complete. Supports were needed underneath the chin as this was a large overhang. The head model was printed on an Ender 5 3D printer by a fellow student and cost R200.00.
​​​​​​​
Projection Experiments
Lasercut models from perspex
Unfortunately, the laser did not fully cut the Perspex and removing the parts caused the brittle material to crack.
BIG HEAD
Laser cutting pieces 
The large model was created virtually the same as the smaller model but on a significantly larger scale. The size of the model was limited by the door size at the University of Johannesburg as this sculpture was assembled in the workshop and then transported to the studio for the future projection projects. The size of the sculpture was then scaled to be 2.5m tall, 1.6 metres wide and 1,15m deep. 3mm corrugated cardboard was chosen to be the most suitable material for this process as firstly it is cheap, at R10.00 per 1100mm by 900mm sheet, the material is strong but lightweight and finally it is able to be cut on a laser cutter. With these parameters in place the model could be modified to become suitable for this form of assembly. The model was sliced in front of the ear leaving just the face to reduce the size, scale and scope of this project. The file was then imported into Blender and decimated to reduce the polygon count. This obj. file was then imported into Pepakura and unfolded. The cut lines need to be manually chosen and are illustrated below as orange lines.
Keeping the pieces as large as possible is very useful to reduce assembly time, keeping in mind the size of the cardboard sheet that they must fit onto. Pieces that are too thin, delicate or unnecessary complicated were simplified using the smooth edge and join/disjoin face commands in Pepakura. Once each separate piece was created, they were then organised to fit onto the cardboard sheet’s pages in the software. Filing up the the page with as many pieces as possible while leaving 15-20mm gap between each for easier cutting. 
As the eventual model needs to be folded along demarcated fold lines there are two methods that were explored. Firstly, the laser was set to half the standard power to score along the fold lines and full power for the edge. This would work well for a solid material but as corrugated cardboard has a series of ridges throughput its inside the depth of the score line was uneven and folding was not always accurate. The second method was far simpler and more successful. All fold lines were converted into dashed lines. 20mm dash followed by a 20mm gap and then a 20mm dash. Then all lines were cut on the laser cutter. The dashed lines guided the fold but had enough material between each piece to add strength and accuracy to the model.
Due to the 3mm thickness of the cardboard flaps would not be suitable for the assembly as they were when paper was being used. The flaps would raise the mating faces too much creating an inaccurate model. Instead, the flaps were removed and hot glue was used. The numbers used to align the pieces were unable to be cut/scored in the cardboard so as soon as each cardboard piece was cut, they were all hand labelled according to the digital document. This process was essential as assembly would be impossible without them.
The assembly of the model was straight forward and followed the following steps.
1. Two mating cardboard pieces were found
2. Their matching pieces were located digitally 
3. Each piece was folded according to the digital template
4. Masking tape was placed on the outside of the pieces 
5. Hot glue was used to attach the edges together 
6. Small strips (100mm by 50mm) of cardboard were placed along each edge to reinforce the seam. The cardboard pieces needed to be folded along their length in line with the corrugations to properly fit the folds of the model and not distort it. 
7. This process was repeated for every part. 

Crafting of a steel frame to reinforce the head
Due to the scale and weight of the cardboard sculpture, it was unable to be self-supporting. The sculpture was then clamped to a heavy wooden display stand for both assembly and viewing. To address this lack of strength a framework needed to be created. The model sagged and the cardboard flexed so an accurate framework needed to be created before attaching it to the sculpture. The bottom neck perimeter was able to be cut out on the laser cutter and attached using a series of folded 90-degree pieces of cardboard. This reinforced the shape of the neck but also made sure it didn’t flex or warp. The same process could not be made for the back of the head due to its significant scale. Instead, the perimeter of the back of the face was placed onto 28 separate pieces of A3 paper. Registration marks were placed on each page’s corner for later alignment. Once printed the pages were then attached together using masking tape and taped to the workshop floor. Tension was created in each corner with tape to make sure the layout was flat and that the paper was not lifting up which would affect the scale and accuracy of the final framework. 

3mm by 32mm flat mild steel bar was chosen as the frame material as it was readily available, The University of Johannesburg Industrial Design Department had large amounts on hand. The workshop also had a metal shear, metal brake and a 5mm metal punch. This allowed the entire fabrication of the framework to be done manually and without any electronic tools. Once laid out on the floor the outer perimeter was divided into various straight sections and individually measured. These measurements were duplicated onto the steel and marked with a set square and a permanent marker. Each measurement was individually labelled to make sure bends were done correctly. 
A sliding bevel square was used to match the folds on the paper to the folds on the steel. The steel was placed in the steel brake and was folded until it corresponded with the bend on the square. A large hammer was used to persuade the metal to match the correct bend angle. The max length of the steel sections was 3000mm so 4 pieces needed to be used to make the entire perimeter. They were joined together using nuts and bolts through punched holes in the steel. Each joint was made 1 section larger so that there was always a minimum overlap of 200mm. 

Once the frame was created straight sections of steel were needed to prevent warping. The lengths were found using the edge-to-edge measuring tool in Pepakura and were duplicated onto the steel. By folding up each end and punching a hole in both the edge and the framework these stretchers could be bolted into place. This framework was still flexible so triangulation was used to reinforce it. Metal was bent and bolted into place forming triangles throughout the frame. 
To attach the frame to the cardboard sculpture a variety of methods were discussed. From high strength construction adhesive to a series of clamps. Finally, nuts and bolts and custom washers were settled on. The washers were cut from sheet steel on a hydraulic sheet steel shear. 32mm strips were marked out and then divided into 40mm by 32mm rectangles. The centre of each of these were punched followed by being cut on the sheer. This created 50 custom washers. A hammer and an anvil were used to flatten each washer as the force of the punch distorted them. 
50 holes were punched around the framework to accommodate these nuts, bolts and washers. This frame was then placed inside the cardboard head sculpture and temporarily clamped in place. The steel frame was clamped to another display stand to keep it upright as it is front heavy. A drill was used to drill through the punched holes and through the cardboard. Bolts were then placed through the cardboard using a washer to compress the cardboard and to stop the bolt from ripping it.
Once attached the sculpture was far stronger and due to the outer perimeter mimicking the digital model the proportions of the face weere now perfect. The nose and eye area still sagged inward so more reinforcing was required. 3 x 3000mm strips of 22mm pine brandering were attached with bolts, hot glue and screws to both the cardboard and the steel framework to properly reinforce the head. Once this final reinforcing step was finished the head was complete. It took 3 people to carry the head from the workshop to the studio. The head is still very light although due to its large scale it is very cumbersome to move around

Completed head sculpture
Projection Experiments
After the sculpture was completed, it was finally able to be projected onto. The sculpture was placed inside a large studio against the back wall. The studio has a row of windows along one side which let in sun light. A series of display stands were used to block this light from falling onto the sculpture. The studios lights were turned off and all the doors were closed to make the rooms as dark as possible. A standard overhead projector was mounted on a wooden base and focussed on the sculpture. It was placed around 5 metres away from the model and placed on its side so that the beam of light covered the entire head sculpture. A variety of images were projected onto the sculpture. Firstly, my face, of whom the sculpture is based off of was projected onto it. As they are both the same face the eyes, mouth, chin and cheeks all line up creating an enormous 3D model of the person’s head. Various other faces were projected onto the sculpture. The following images show these projections. Some faces align better with the sculpture than other. A template in Photoshop was created using gridlines for the eyes, nose and mouth lines. This allowed all of the various projections to be as correctly aligned as possible.
By projecting flat images of textures onto the model they warped around the face creating depth and a very realistic appearance. Other textures such as fire and gravel were also projected leading to some interesting results. Giving the cardboard the look and texture of other materials.
The cardboard sculpture was created from a 3D scan and model. The face model was then imported into Solidworks visualize and various textures and materials were applied. Including gloss red, blue plastic, cork and steel. The digital model aligned with the physical model allowing the cardboard to appear as other materials. 
Technology lets all avenues of ideation possible and this project was the perfect platform for that. A video was taken of my head from a head-on angle. For 60 seconds I blinked, looked around and make various face movements. Once projected onto the head this video aligned perfectly and the result is a startling and very realistic depiction of a 2.5m tall face. Virtually coming to life. 
Projection experiments at the UJ Gallery
Video projection
Flat 2D video
Projected video appears 3D
Projection-Mapping Project: Big Head
Reflections: Gabriel Gibbon
This project introduced me to projection mapping and the blending of virtual reality; digital art with the physical space and the 3-dimensional world around us. I learned an enormous amount throughout this project. The scanning and digital manipulation of my head in Blender, Solidworks and Mesh mixer introduced me to the digital sculpting of organic shapes. To bring these digital models into the real world I used a variety of methods, many of which were new to me. To create the various stacked and interlocked models I learnt how to use the laser cutter, a tool that I previously thought to be complicated. I have since used this tool on other projects, my own business and I have helped teach the first years how to use it. 
Increasing the scale of the head models required even more tools and techniques to be used. From small scale paper craft using a vinyl cutter to the eventual 2.5m model. This project showed how applying design thinking and a methodical approach helps break down a problem into manageable, actionable steps. A huge point of learning in the project was the blend between digital fabrication techniques and manual craftsmanship. Combining the accuracy and detail of virtual models with the “hands-on” approach to prototyping. 
This has fundamentally shifted both my view of prototyping but also the options available to me. Materials, finishes and designs can be instantly adjusted or updated when using projection mapping. Adding movement creates a further level of creativity and possibilities.  
The building of the Big Head sculpture was the largest project that I have undertaken and built by myself and I am very proud of the result. Not only is it accurate to the initial model, but it also works very well as a platform for projection mapping. I also entered the sculpture into an online art competition which not only won the grand prize but also received 12 000 views, showing just how far-reaching a university project can go. This shifted my perspective on how I approach future UJ projects, as future portfolio pieces rather than standalone briefs. 
I am thrilled to have learned and experimented with projection mapping and I look forward to seeing how far I can take it into future projects and ideas. The possibilities are quite endless. ​​​​​​​
Projection Mapping Project: Big Head
Published:

Projection Mapping Project: Big Head

Published: