The Journey of Nova is a miniature version of a graduate research project. The goal was to design a character in 2D with the intentions of 3D printing the character in the end to give it physical form. This mini project was driven by an intense passion for character design paired with the desire to bring the character from a page or screen to the palm of a hand. By the end of the project, there is a greater understanding of the character and the process by experiencing trial and error and reflecting upon each stage of the project.
STAGE 1: CHARACTER
Art & character originally created by a friend.
Description: anthropomorphic, digitigrade creature that exists within dreams. Simple design and color palette so the mind of the dreamer is not overwhelmed
Occupation / Ability: travels in the minds of those asleep, replacing nightmares with pleasant dreams as it passes through
Personality: quiet, benevolent, calm, peaceful
The character was originally designed by a friend. The first two drawings on the far left are drawn by her. The "T poses" are done by Amanda Grace.
This turnaround drawing is extremely useful to have as a reference when 3D modeling the character. It helps pin down the character’s form, shapes and proportions so that the 3D model is more accurate. Nova’s color palette and design are simple, therefore making it easier to translate to a 3D model.
Drawing Nova in a t pose front and back was very helpful to determine how the shapes should look when modeling the character in the same pose in those views. The original quarter views serve as an in between for the poses to inform more about the anatomy. For example, the arms in the quarter poses appear much larger compared to the t-pose arms. Drawing a straight back view without the tail is useful because when 3D modeling the character, I have to model what is behind the tail (which isn’t completely visible in the original concept art).
The mane is another example of something that is shaped differently from the front and back views. The t-pose side view helps make 3D modeling the legs easier. Nova’s legs are the most difficult part of the body to draw and model. This is because they are digitigrade legs that need to support a human proportioned torso. I referenced how other artists have studied characters with body types such as Nova’s to get a better grip on the body proportions.
Nova’s legs are a hybrid of D and G according to this leg anatomy reference. It would seem that digitigrade legs would not support the weight of Nova’s torso and posture. However, Nova’s design balances itself by having larger arms and a tail that help distribute the character’s body weight evenly. If necessary, I will be modeling a base for Nova to stand on to ensure extra stability for the sake of 3D printing in the end.
STAGE 2: BLOCKING
Block out form of creature with basic shapes in Maya 3D software
The blocking out stage began before the 2D reference was made. The first pass at the model was not accurate and needed to be fixed to fit the right shapes and silhouette. It would have saved time by having the reference inside the 3D software to follow instead of “eyeballing” the concept art from a different computer screen in the beginning.
Adjusted block out based on reference turnaround
Made model adjustments to front and side views to match up with 2D reference. Added more subdivisions to smooth surfaces. Deleted eyebrows temporarily. Added the arm. Side view is lined up nicely with added arm. Completely lined up and ready to start blocking out the mane, chest fur and tail.
For the sake of understanding how the eyelid should fit onto the head, I had to shape the actual eye to exist behind the lid, so that it is more believable that there are eyes behind those lids. In the middle of setting the eye, I also discovered that the head itself had no “eye sockets” visible in the 2D references. So I pushed the mesh inward in order to create a dent for the eyes to populate. This helps show that the character has a skull underneath. The eyebrow was then added after that. The eyes, eyelids and eyebrows all deviated from the 2D references. In other words, these parts did not line up and I had to “eyeball” them to get them to look right from all angles.
Blocking out fur was not as difficult because the 2D style of the character has a sharper silhouette making it easy to block out using sharp geometry. The gem was modeled to line up with the 2D reference. However, a mistake was found in the accuracy of the 2D drawing. It was not clear in 2D what the facets would look like. In 3D the gem is corrected with more gem like facets based on its cut.
The tail had to be modeled as if it had no fur first. Basically, if the character had no fur on their tail, this is how it would look. The fur is blocked out below.
Fur is blocked out and added below the tail first to match with the 2D silhouette. Then the same fur was duplicated and moved to add more dimension which is not shown in 2D. The same thing was realized during the block out of the mane.
The mane is another part of the character that was revealed to not match well with the 2D reference. This is because the front 2D view was a guess of how the mane would spread out in 3D space. Once modeled, it doesn’t match the 2D reference, but that is okay. Like the tail fur, the mane had to have another “layer” of fur between the middle and outer fur to fill the 2D reference void (what you can’t see). This helps make the mane look more full.
This image shows the final version of the character fully blocked out and ready to be retopologized.
STAGE 3: RETOPOLOGY
I discovered that I can use most of the blocked out model. Blocking out a model usually means that what you made needs to be redone with retopology to make the model easier to animate and add texture down the road. The only areas that need retopping are the whole head and areas where each piece meets. For example, where the arm intersects the torso, retopology on that area will help connect the two parts without having to retop the whole torso or arm. Same applies to where the leg meets the lower body and where the tail meets the back.
I also realize that the mane will have to become its own separate piece in order to save time. The mane could be attached to the back of the head, neck and back, but for the sake of completing the model in a timely manner, that is not necessary. Though, because the mane will be separate, it will need adjustments such as polishing the overall look of the mane and combining sections to make it easier when it comes time to do UVs and posing.
In green, these areas serve as connectors. The arms, legs and tail are all separate and need to be connected together with even topology. These connectors lined up well. The areas that were underneath were deleted and replaced with these parts to connect the body together later.
In the blocked out version of the head, the topology is too complex and does not match up with the ears, neck and torso. The eyes, eyelids and eyebrows also are separate and don’t line up. These parts are connected using retopology and now flow well shown in the second image.
After all parts are connected and retopologized, the middle edge is centered so that when the half is mirrored, it will line up correctly to be combined. The gems and the mane will remain separate objects.
The mane was difficult to figure out since the 2D style reference of Nova does not give enough dimension to properly show how the mane will work in 3D. Improvisation was necessary to get the mane to look full enough by using multiple pieces and layers of hair chunks.
The tail has a similar amount of improv. In a side view, it matches well. However, just like with the mane, the tail should have more fur to transition to the sides in order to make it look more full.
Final 3D model in Maya, front, side and perspective view with the 2D references behind them. Model is ready to be UVed and textured.
STAGE 4: UV & TEXTURE
Cutting and Unfolding
The UV sections are cut where the white lines appear. These cuts help spread out the model parts so that the texture can be more evenly distributed. The checkered pattern is spread throughout the model in Maya after cutting the model into unfoldable sections. This pattern shows if there are any areas that stretch too much or may need to be cut into more pieces so that future texture will not be skewed. Overall, the cuts I made were satisfactory. Here the UVs are shown cut up, unfolded and organized onto a flat plane. Each piece is intended to be a bit apart so that the textures for each piece do not overlap later.
Painting and Editing
Once the model was exported out of Maya and imported into Zbrush, the UV map information follows it so I can start to paint directly onto the model. The 2D reference is set in the background for color matching. The colors appear darker on the model because of the 3D material that captures light and shadow information.
The flat color option in Zbrush disregards all light and shadow. So the colors here are shown as the correct colors based on the 2D reference of Nova. After painting is done, the new painted texture can be exported out in the same format and organization as the original UV layout. However, the texture must be edited in Photoshop to address any issues of blurry areas or possible overlap.
The texture is edited in Photoshop for cleaning up areas that could not be reached by the paint brushes. It is also referenced inside of the Maya project file so that the model updates after every edit.
Final texture applied to the final model in Maya. The gems and mane did not need to be textured since they were only one color. Materials were assigned to the mane and gems inside of Maya to give them their color. At this time, a gem shaped base is added for 3D printing support. Model is at this point completely ready to be posed using tools called blend shapes in Maya.
STAGE 5: POSING
Pose concepts with notes. On the left, the pose is designed to be grounded on a base specifically for 3D printing. The pose on the right is strictly for Maya Arnold renders. The dream catcher circle is added as a cherry on top for the final render to look more in character. Nova as a character has the ability to weave light into shapes such as a dream catcher that looks similar to a magic circle. The dream catcher will be a model that will serve as a glowing light source for the final render.
First image shows the results of using QuickRig in Maya that generates a simple rig based on a humanoid skeleton and joints. Because this is a quick set up for moving the model, the influence of each part is not perfect. This results is folds, dents and jagged areas to appear as the rig is being used for posing. In the images above, examples of this happening are most noticeably the mid area (the belly button that is not supposed to be there) and the folding of the hips and upper thigh.
Blend shapes are tools in Maya mainly used for animating a model quickly. For this project however, they are used to fix parts of the model that are broken from the quick rig. Above, the hips are smoothed using a blend shape. The dent on the mid section is also fixed.
The tail was not recognized by the quick rig in the beginning because the rig is catered to human proportions. In order to move the tail, a second blend shape is used by creating a lattice deformer to push the tail down.
Final 2 poses using the combination of quick rig and blend shapes. Dreamcatcher model was downloaded for free and later edited in Maya.
STAGE 6: RENDERS
All renders are rendered using Arnold Renderer in Maya. An appropriate amount of lights and cameras were set in the scene to capture the compositions.
3D Print Pose
STAGE 7: 3D PRINTING
3D printing preparation and process overseen and guided by Lee Cherry in North Carolina State University Design IT Lab
The 3D printer of choice is the Formlabs Form 2 resin printer. I have printed using the Fusion 3 F410 before for past prints that were simpler in design. Because Nova is the most complex model I have made for printing, I decided to try resin printing. The 3D printer of choice is the Formlabs Form 2 resin printer. Trying resin printing not only broadens my knowledge of printing options and prcoesses but also informs me of what type of printer is best to use next time I decide to print a model for figurine making.
Autodesk Netfabb is used to repair the model before the model can be positioned for the print.
Preform software from Formlabs is then used to position the model inside the constraints of the printer. It is also used to calculate and generate the best way to print supports to ensure that the model is stable while being printed. In a resin printer, typically models are printed upside down, letting gravity do its best.
Printing job took a little over 8 hours. Model finished printing and is removed carefully.
The print is extremely messy and sticky, like glue.
Cleaning Off Leftover Resin
Using a 99% Isopropyl Alcohol bath, the model is emerged in the machine for 45 minutes. The alcohol and subtle movement generated by the machine cleans off the sticky left over resin. An additional 20 minutes was needed to completely clean it all off.
Cutting Off Supports
After drying off, using a small pair of side cutters, the supports are cut off as close to the model as possible. However, there are bumps left that need to be sanded or carved off using an exacto blade after the curing process.
Curing In UV Light
Curing is the last part of the post clean up required for this print. Curing the model for 45 minutes in a UV light machine ensures that the model will harden and therefore be best suited for sanding and carving areas that need attention such as the leftover bumps from the cut supports.
Sanding & Cutting
Using a small sheet of fine grain sanding paper, I sanded down the areas that were uneven due to the leftover support bumps. After sanding, an exacto blade is carefully used to chisel and carve off remaining bumps. Running water over the model in between several passes with the blade is good practice to get rid of shavings. After all bumps are carved off, the model is sanded once more.
Stage 8: FUTURE
For the project, the goal was to finish the model as a successful 3D print to complete the journey from 2D to 3D. The goal was reached, but it could continue. I have plans to obtain a drimmel tool to have more fine tune control over sanding areas that were too small to carve with the blade or reach with the sand paper. After that I plan to prime the model with a medium grey spray paint and eventually hand paint it with acrylics and small brushes.
I also plan to send the model with its texture to i.materialize, which is an online 3D printing service, to get the model printed using a Polyjet that prints in color. After I recieve it, I will update this page with a pros and cons comparison of painted versus printed in regards to applying color to a print.
As for the digital 3D model, there is much that is possible. The quick rig that the model is set up with is a good start for animation. The paint weights would need accurate adjusting in order for the model to not fold in on itself as it is moved. After paint weights are fixed, there are many ways Nova can be animated. A walk or run cycle, jump animation or simple idle animations would be possible. The model and texture are simple and low poly enough that it could even be used as a playable game character. The journey of Nova from 2D to 3D is never ending.