Problem

For this final project, I decided to take an earlier model we used in class, a simple human model created out of blocks, then rig and animate this model. I wanted to animate this model to jump several hurdles, getting experience with animating running and jumping, as well as several other poses at the beginning and end of the animation, such as the model performing several workout stretches or falling onto the ground and getting back up. Here is a link to my final Maya file.

Solution

For my approach on this project, I started by rigging the model. First, I created the skeleton, where each piece of the block man got a separate bone to control its orientation. Once I had created the skeleton, I started setting up controllers and IK handles to help with animation. I ended up adding IK handles to the legs, which, combined with the “sticky” attribute allowed me to move the root bone of the skeleton as a “center of gravity” controller, as the feet would attempt to stay in place unless I moved them directly. This ended up helping immensely when animating the model, as it prevented any issues with the feet sliding when attempting to animate the run cycle. Once this was done, I ended up setting up a few derived keys to control the spine and neck movement, as I thought this would allow me to easily control common spine movements with a single attribute. While this worked well, it had a few unintended consequences that made animation much more difficult later (see Challenges). Beyond this, I set up the arms to have IK controls, but I ended up deleting these and reverting to FK because I didn’t like the way the controls felt when moving the arms the way I would need to for the animation. Lastly, the head, wrists, and feet all had orientation controllers that let me control their rotation independent of the other controllers.

Next, I started the animation of the project. This required several iterations of my rig, but eventually I decided to just move forward with the rig that I had (see Challenges). I ended up hand keying the entire animation (I didn’t copy or paste any keyframes for the running). Ultimately I think I had to rush a bit too much for the animation, and this left a lot of the animation pretty choppy. Also, because of the challenges with my rig, I was unable to do all of the poses that I wanted to do at the beginning and end of the animation, so the animation ended up quite a bit shorter than I had intended. Overall, I think several of the poses in the animation were strong, but with more time and a more complete rig, I would be able to make a much better animation.

Images

Here are several frames from the animation showing off key poses.

Video

Challenges

This project had a lot of issues that were largely avoidable, and most of them can be summarized with “Don’t take shortcuts with rigging”. Several issues occurred with the rig, notable with my derived keys for the back and neck. These keys ended up preventing me from rotating the back and neck normally, meaning that, hilariously, I was unable to rotate the entire rig with the root controller, leading to some unfortunate limitations in what I could actually do with my final rig (but this was my fourth or fifth attempt at the rig, and I decided going forward at that point was probably for the best). Ultimately, this made animating the model WAY harder than it needed to be, leading to a lot of the cut poses and choppy keyframes in the animation.

Another challenge was with the model itself. I assumed that using the premade model would have saved me a lot of trouble, but because I didn’t pay attention to the units and orientation of the human model, when I imported it into my scene, it ended up being very large (the model was in feet, my scene was in centimeters) and pointed in the wrong direction (which was only an issue because of my rig). I eventually sorted this out, but it did waste a lot of time while I tried to figure out what was wrong, and how to fix it.

Lastly, I left myself very little time to render because of everything else, so I divided up rendering between three computers. Unfortunately, there was a discrepancy between the first computer and the other two, leading to my first 150 frames appearing darker than the last 330. I don’t know think it was a difference in my Maya settings, as I used the same scene when rendering from all computers, so I’m not sure what caused this lighting discrepancy. By the time I noticed the difference, I didn’t have enough time to re-render the first 150 frames, so I left them in while putting the video together.

I feel like I learned a lot while making this project (especially about rigging), and I think that I could improve this project significantly if I were to make it again.

Problem

For this project, I was given a model of a unicycle, and I had to surface, rig, and animate this unicycle. The animation had to be at least 10 seconds, with 800×500 resolution at 30 fps minimum specs. The maya scene and animation can be found here. (let me know if you need permissions via email!)

Solution

I surfaced the unicycle and the room with several different Arnold standard surface shader. For lighting, I simply used one Arnold area light within the room, and used a physical sky for the outside light. All of the room modeling was done using simple primitives, and was fairly easily modeled (though admittedly, I didn’t know what to do for the outdoor portion of the animation, so I just used a plane).

Rigging

I started by rigging the unicycle model. For this, I decided to give the unicycle 4 main controllers: the root controller, the wheel controller, the seat controller, and the tilt controller. I also added pedal controllers, which allow the foot pedals to be rotated. However, I didn’t end up using these very much for the animation.

The Root controller is responsible for the rotation and translation of the main model. This was used extensively during the animation (for fairly obvious reasons).

The Wheel controller is responsible for rotating the wheel of the unicycle. I animated this by hand primarily, which was actually a lot of fun to get its rotation to line up with the movement of the unicycle.

The Seat controller is the “head controller,” as the seat worked really well to help the character emote, which was really fun to use in the animation as well. Overall, I think this controller was the best implemented, and has the fewest issues.

Lastly, the Tilt controller used a set driven key to manipulate the shaft of the unicycle, allowing me to tilt the unicycle’s shaft forwards and backwards, giving it a spine like movement. This was exceptionally useful for both emoting, follow through, anticipation, etc. (It basically allowed the model to animate in a much more interesting way from an animation principles standpoint). This controller could have been made even better by adding a left/right tilt option as well, which would have helped the model emote even more.

To improve the rig, I would add left and right tilt functionality for the Tilt controller. Additionally, I would add a deformer for squash and stretch, for either the wheel or the whole model. Otherwise, I’m fairly proud of how the rig turned out.

Animation

For the animation, I tried to focus on follow through, anticipation, slow in slow out, exaggeration, staging, and appeal.

I think that most of the animation works well, but I think I need to adjust the timing for a lot of the actions, as most of them play out too quickly, which makes the animation a lot less effective. Next, because my rig wasn’t set up to deform the model, I didn’t have any squash and stretch, which would’ve helped sell the animation whenever the unicycle crashed or jumped. Lastly, I think that near the end, I started getting sloppy with my follow through for each action (though the animation started getting a bit too long for constraints of the project.) Overall, I am pretty happy with the animation, though I could have spent a lot more time working on it to make it better.

Conclusion

This project was a lot of fun! I really enjoyed thinking about the rig and doing the animation. I think that I could make the rig a bit better for animating the unicycle to give it a better range of expression, and if I messed around with the animation a bit more, I could make it even better. Overall, I think this was my favorite project so far!

Images

Video

Challenges

I had several challenges that made completing this project difficult.

First, I had trouble setting up my rig and getting all of the parenting relationships correct, which caused me to do a bit of (probably) unnecessary weight painting and unbinding/rebinding the skeleton. Later on, I noticed that I had set up my root controller improperly, and that this had screwed up the unicycle model. This forced me to go back to a previous version of my rig and caused me to lose all of my progress on animating (which thankfully wasn’t too much).

My next challenge involved my lights and render settings. No matter what I did, I couldn’t seem to get rid of the strange artifacts with my realistic sky light. You can see this pretty clearly outside during my animation, where the outdoor light causes “pixels” to pop up and vanish quickly. This also occurs with the unicycle’s specular highlights from both lights, though it is harder to see clearly. I tried messing with the light and camera sampling, but I couldn’t find a good fix.

My final, and greatest, challenge to completing the project was rendering. I have no idea if I had a setting turned on or what, but rendering this animation out took almost 20 hours (each frame took about 2-4 minutes to render). Admittedly, this was partially my fault for creating such a long animation (and for waiting so long to start rendering), but I am genuinely curious if someone can find a reason why my renders had taken so long (I compared render timings with a classmate, and he said that his whole animation took 5 hours to render).

Either way, I think that the project still went well in spite of these setbacks, and I had a lot of fun!

Problem

For this exercise, I had to find a Pluralsight tutorial that related to rigging or animating in Maya 2019, then record three lessons that I thought were particularly useful.

Solution

For this exercise, I chose to watch the Maya 2019 Fundamentals: Rigging and Animating course on Pluralsight. Specifically, I focused on tips that would be useful for setting up a good, animation friendly rig with good control set ups and parent relationships. I will group the takeaways I had based on their topics.

Setting Up Good Controls for your Rig

When setting up controls for a rig, NURBS Curves are used because they are objects that can be manipulated, but will not appear when rendering, making them functionally invisible handles for an animator to use.

When initially setting up a control handle, you can go into control vertex mode and change the orientation of the control handle itself without changing the orientation of the pivot point when in object mode, which can help you make the control handle have the proper alignment without messing up the pivot alignment.

Also, it is helpful to have the pivot point aligned with the actual bone that the controller is moving. To do this, simple hit D to edit the pivot point, then use V + Middle Mouse button to snap the pivot point to the location of the the bone itself.

Lastly, the controls should always have frozen transformations. To help with this, putting each control handle in a group that you move around to get it to its default position helps a lot with setting the controls up initially without messing with the handle’s transformation, rotation, and scale. Lastly, this group is very useful when the control needs to be moved by a parent control, such as the root: if you parent constrain the group to the root’s movement, the child control handle will still move without having its translation and rotation attributes affected by the move.

Using Set Driven Keys with a Rig

You can set up controls that drive a set driven key relationship with any number of joints. Those joints can only be moved with this set driven key relationship, but its parent and children joints can be rigged normally. To set up multiple driven joints for one driver, you only need to load each driven object and key the driven attributes’ values for any relevant values of the driver.

A cool trick to use for set driven keys is to make an attribute up that controls them. To do this, go into the channel box, edit, add attribute. Here, you can create an attribute in the channel box and change many parameters, including maximum and minimum values, and whether or not it is keyable. Setting this up as the control for the set driven key relationship can be very nice, as it means you don’t have to edit the control handle’s position or rotation to affect the position of the joints it’s affecting.

Cleaning Up a Scene

Several cool tips for cleaning up a scene and making sure it is animator friendly.

Make sure that your rig is properly parented: the root control handle should not only move the root joint, but all of the other controls so that their relative positions to the model stay constant. This should be true for all control handles that edit the position of another control handle’s joints.

Lock and hide all attributes that the controller shouldn’t be affecting. This keeps things easier and safer for the animator, so that things can’t accidentally break or be incorrectly selected during the animation process.

Name all of your joints and controls something that will make sense.

Once you are finished setting up the rig, group all of your control handles into one group. Then, do the same for the skeleton and mesh (if they weren’t already grouped like that already). Lastly, group the skeleton, mesh, and control handle groups together into one large group that is properly named and easy to navigate for anyone looking at your scene.

Bonus: Useful Hotkeys

Ctrl + 1 = toggle solo mode

Hold V + Middle Mouse = snap to vertex

Hold X + Middle Mouse = snap to grid

G = repeat last command (useful for parent constraint, etc.)

Problem

For this project, I needed to create a simple walk cycle that was stylized to convey an emotion. We were given a model that we rigged in class, and that was the base that I was supposed to use for this exercise.

Solution

For my walk cycle, I decided to make a sad walk cycle. For this, I started by setting the initial pose of the model and keyframing them at frame 0. For this, I keyframed the position of the root, the IK rig of each leg, the back, neck, and each shoulder. I hunched the back and had the shoulders hang forward, and I set each foot in the proper place. Then, I duplicated this frame at frame 40 so that I have the same start and end position for a loop.

After that, I created the top of the first step at frame 10, the end of the first step at frame 20, and the top of the second step at frame 30. For each frame, I payed attention to the rotation of the body, the height of each step, the slouch of the back and neck, and, of course, the position of the legs and feet. After I was happy with the result, I increased the space between the end of each step and the start of the next by 10 frames, adding more anticipation and weight to each step.

To improve this further, I need to make the upper body move a bit more, and having a bit of follow through by having the arms swing more would help make the animation look more fluid. Additionally, I tried making some changes to the graph editor, but I didn’t really like what it did with the flow of the animation, so I decided against it. I think that if I messed around with the tangents more, it would add more weight to the animation.

Video

Problem

For this project, I needed to find an image of a scene and attempt to recreate it as closely as possible in Maya by modeling, surfacing, lighting, and rendering. My model can be found here (email me if you need access to the drive folder).

Solution

For my solution, I started by modeling the tables that all of the objects were placed on, and arranging them so that they would be lined up properly to recreate the shot. The tables themselves were fairly simple, only comprised of a series of cylinders and toruses that were slightly modified.

After the tables were complete, I moved on to the flask/vase (?) on the bottom left table. I modeled this out of a sphere, slightly deforming it until it was in the shape that I wanted. Then, I modeled the candle, which was three different cylinders. Next, I modeled the clay vase, which was another cylinder that I shaped, then added ridges to the outside of so that it would look more like the reference image.

Next, I modeled the glass bowl, which was the second hardest object to model. I started with a cylinder and created the overall shape of the bowl, then I poked all of the faces around the outside of the bowl, creating a diagonal pattern along the outside. Then, I added a few more edge loops that gave me the geometry to create the alternating divets of the bowl from the reference photo, this was challenging, and the bowl didn’t end up exactly right, but it was fairly close to the reference photo.

After this, I modeled the glass pitcher, which was fairly easy. I once again started with a cylinder, got the basic shape, then prepped it for smoothed mode. Then I modeled the brush, which I created by using a sphere for the handle, then a cylinder for the brush itself. Lastly, I modeled the copper pot, which I had the hardest time modeling. I started with cylinder for the base, used a sphere and a cylinder for the lid, and two cubes that I extruded to create the handle and its mounting bracket. I was trying really hard to get the pot to look clean and well defined, especially the spout. In the end, I mostly succeeded, and the topology stayed relatively clean around the spout. The last modeling I had to do was to add five planes for the floor and the four walls of the room. With that, the modeling was complete

Next, I added a few lights to the scene for help seeing my shaders, then moved on to shading. For this project, I primarily used Arnold’s aiStandardShader, for all of my objects. I started by creating a glass shader for the glass bowl and pitcher. I did this by making the glass fully transmissive and decreasing the roughness of the specular highlight. This achieved the effect I wanted really well, creating a clear glass that distorted what was behind it. I also attempted to add caustics with this glass, which partially worked (see challenges). After the glass, I created a red liquid material for the liquid that (I assumed) was in the glass bowl, to help get the red coloration you see in the reference image (see challenges). This material was very similar to the glass, with less transmissive color, and slightly more roughness.

After those shaders, I moved on to create shaders for the tables, creating a smooth ceramic for the table tops and a painted metal shader for the table legs.

After that, I created a more silver metal for the table feet. Following that, I made a brass material for the flask and a copper material for the pot. These three materials were very similar, mostly differing only in their base color.

After that I textured the clay pot, which I made very rough so that it appeared more diffuse. I also applied a bump map out of fractal noise to make the pot look rougher and unfinished, like in the reference. After that, I textured the brush handle and the handle of the copper pot. I used a black plastic looking shader with medium roughness to diffuse the specular highlight, which seemed to work well for the material. After that, I tried to texture the brush in a way to preserve the appearance of bristles (to little success). To do this, I applied a texture with a high roughness, then linked the color to a fractal noise that I distorted to give the appearance of random, mostly horizontal lines around the outside (this worked out very interestingly. It looked more like wood grain, but it was procedural and very easy to make, so this is a pretty neat trick! Unfortunately, it didn’t work quite as well as bristles). I then created another fractal noise with the same parameters, and linked it as a bump map to give the bristles texture. This didn’t work as well as I had hoped, but I couldn’t think of a better solution.

After that, I only had to texture the floor and the walls. These were both actually really fun, as they had a pseudo noise pattern that was fun to try to get right. For the floor, I ended up using the marble texture scaled up to get the pattern of the floor, then I gave the floor a medium roughness to get the reflections of the tables to look right. For the walls, I used the cloud texture, which I manipulated to have similar splotches to the wall from the reference image, and I gave the walls a high roughness so they wouldn’t reflect anything. With that, all I had to do was prep the scene to be rendered.

First, I had to set up my lights, which was quite challenging, as I couldn’t quite tell what lights the photographer used for their setup. It really seemed like they only used one light, but I couldn’t achieve a similar light setup without multiple lights. In the end, I used two Arnold area lights, with only one light casting shadows. After I did this, I created four different cameras to render each of my rendered shots for the scene.

I rendered using the Arnold Renderer in Maya. For my render settings, I chose a sampling of 6/6/3/3/3/3 (I increased camera to increase the quality of the anti aliasing, and diffuse to get better caustics, which only sort of worked). However, the first image took a very long time to render, so I lowered this to 3/3/3/3/3/3 after the first render (which was based on the reference image). I also rendered all images as HD 1080 and had all of my light sampling set to 5.

Pictures

Challenges

Over all, this project went fairly well. I only had some minor challenges crop up while I was working, but I couldn’t figure out good work-arounds for a few of them.

For modeling, I had several small issues. First, I misjudged the foreshortening of the table legs with the front table, causing the legs to be a bit too far apart. Second, I made a few minor modeling mistakes with the glass bowl, causing the dimples to not be perfectly uniform between rows. I also realized too late that, in the reference photo, the glass bowl doesn’t have any liquid in it, and the glass itself has a gradient color to it. I could probably have fixed this by using a ramp shader on top of the glass shader, but I had noticed this too late to fix it in time. Lastly, the copper pot gave me a lot of issues when I modeled it, and while I fixed many of the topology issues, there are still a few minor problems with it. For example, I had to add several edge loops to define the edge of the spout of the pot, but I couldn’t add these without causing hard lines along the length of the pot (because I had to add these edge loops, they affected the geometry where I didn’t want them to, but if I tried to isolate the edges I was adding, that would have caused several other topology issues). This didn’t affect the model too much, but it is noticeable if you look at the pot along the side that the spigot is attached.

Texturing went surprisingly well this time. I used the Arnold shaders, and looked up a few tutorials on how to achiever certain textures, and I got pretty close with most of them. The hardest one was the brush, and I couldn’t figure out how to get the bristles to look right with the methods I chose. I probably would have had to either look up how to do hair simulation, or maybe used a texture map to get closer to the product, but I don’t think I could get the brush to look much better with my current approach.

I also didn’t leave myself enough time for the lighting, so I realize now that the lights aren’t very accurate to the reference scene. I tried using Maya spot lights at first, and had the lighting set up fairly well with them, but I couldn’t get them to work well with shadows. Because of this, I tried using Arnold area lights, which worked a bit better, but it was harder to get the lights to work exactly the way that I wanted them to (orienting them and controlling where they stopped was confusing, so that is probably the cause of most of the lighting mistakes in the scene). In the end, shadows still proved difficult (I couldn’t get rid of the hard edges on them), and the lights themselves didn’t cover the whole scene as well as I had wanted (for example, the copper pot ended up pretty dark).

I also tried to get caustics to work, but the render times to get decent caustics ended up being too long to get any good images of them. I think my caustics would have turned out better if I had gotten rid of the liquid in the glass bowl, but I was never able to get my caustics to look like the reference image.

If I had left myself more time to try rendering out the caustics and play around with lighting, I might have been able to fix these issues, but I forgot to account for how long rendering can take.

Problem

For this exercise, I had to watch a movie and identify three examples of principles of animation used by the animators of the movie. I chose to watch the stop motion movie Coraline.

Solution

Example 1: Exaggeration

This first clip utilizes exaggeration with the piano. The piano hands move in a very exaggerated way, popping into the foreground of the shot. Additionally, the way that they make Coraline’s father turn around is also heavily exaggerated, giving a lot of energy to their movement in this clip.

Example 2: Overlap and Follow-Through

In this clip, Mr. B (the guy crawling on the rafters) is demonstrating a lot of secondary action with the small tassel on his jacket and the movement of his hat, keeping him a very interesting subject during the clip. Additionally, when he falls, there is a lot of follow through and overlap where his head hits the ground first, then his legs and coat tails flop to the ground. This makes his movement seem alien and uncomfortable, but still very realistic.

Example 3: Anticipation

This short clip shows a good example of anticipation and squash and stretch as the cat jumps onto the higher branch. He stops, quickly squashes himself to prepare for the jump, extends as he makes the jump, then squashes again slightly as he lands. This happens quickly, but it makes the jump look both fast and realistic.

Problem

For this project, I had to take the car model I had made previously, then place it on a table or plane, add materials to it, light it, and render it.

Solution

To complete this exercise, I first needed to fix up several of the issues I had left with my car model. Notably, I finished the undercarriage of the car to the best of my ability without reference, then I added windows to the car so that you couldn’t see the unfinished inside.

Once I had done that, I moved on to surfacing the car. First, I created a Blinn shader and made it a red color that matched the car’s body. Then, I made a Phong Shader for the headlight. For this shader, I changed the color to red and increased the transparency, then I increased the shader’s luminence to make the headlight seem like it was on.

I created four more Blinn shaders, one for the gold metal of the car, one for the silver metal of the car, one for the black windows of the car, and one for the tires. I was debating adding the gold detailing to the tires, but I couldn’t think of a way to do that easily without creating clean UVs for each tire.

Once I was done with the shaders, I moved on to lighting. I added two small point lights for the headlights that were colored red. After that, I set up my three point lighting scheme with a key light, fill light, and back light. Each of these had an Arnold Exposure of 15 to ensure that they were visible when rendering with Arnold. Lastly, I added an Arnold realistic sky light to provide a background for the render.

Once I had done that, I prepped my render settings and created a turntable camera. I set the image output to TIFF, made sure to set the render length to 60 frames, and I set the image resolution to 540 HD. I did not mess with Arnold’s sampling, because I was a bit worried the renders would take too long. After that I created the turntable render as a render sequence, then I took the image sequence into Adobe After Effects, where I compiled the images together and exported them as an MP4 file.

Video

Problem

For this assignment I needed to make two spheres and shade each one so that one looked like an orange and the other looked like the planet Jupiter.

Solution

For this exercise, I created two spheres and opened hypershade. For the orange, I created a Blinn material and changed the color to a bright, saturated orange. After that, I attached a Brownian noise map to the normal map attribute. After that, I played around with the values to get the the bump depth about right, then I added lights to the scene and rendered the orange.

For Jupiter, I started by creating a lambert, then linking the color to a ramp shader. This ended up not working as I expected, as the pattern of the shader changed with the camera direction (like what happened in class). Eventually, I just switched to a 2D Ramp texture, and that seemed to work as expected. Once I added the various layers to the ramp shader with no interpolation, I added a small amount of noise to the layers to make them a bit closer to the original image. At this point, however, I wasn’t sure how to proceed. Each layer needed more detail, and while I could make each layer itself a different ramp texture with high noise, that didn’t work the way I wanted to (each layer overlapped with the others). Ultimately, I didn’t know how to proceed, so I set up lights and rendered the scene. Strangely, for this object, the image came out a lot darker than the orange, so I ended up doubling the intensity of my lights, and that gave me a decently bright image.

Pictures

Problem

For this project (Due September 25th), I was given a toy car, and I had to create a polygon model in Maya 2019 that was as close as possible to the actual toy. To do this, I had to take reference images of the car, and use them as reference for “box modeling.” the model can be accessed here.

Solution

For my solution, I imported the front, top, and back reference images I took of the car, then went about box modeling, starting with the back wheel and using my reference images to build out half of the car’s body. This took me the longest out of all of the parts, because the car’s shape, especially with the fins in the back, made it especially hard to box model with the references that I had (see Challenges at the bottom).

Once I had done that, I mirrored the body, merged the center vertices, then started working on the asymmetrical parts of the car, notably the headlight and the front guard of the engine.

Once this was complete, I moved on to the other parts of the model, namely the front grill, tail lights, and main engine components, all of which were fairly simple to model. After that, I attempted several times to create the bottom suspension of the car, however, I was unable to complete this part of the model in time, and so the current model’s suspension is unfinished.

After attempting the suspension, I went and added the wheels and axles to the model, then adding several “filler blocks” that are placeholders to block out the currently unfinished parts of the car.

In the end, I did not have the time to finish the bottom, the interior, and the windows of the car. However, the overall topology is clean, and if I had left myself more time, I would have been able to finish each of these components.

Reference Pictures

Model Pictures

Challenges

Box modeling was quite challenging for me with this project. I initially started modeling with my reference images, but after I had created the side silhouette of the car, I realized that all of my references were off from each other. This forced me to restart several times, and while I adjusted the images so that they were more closely aligned, the perspective distortion of the images was still too great for the references to be very useful. I was able to get the main body of the car fairly easily, but a lot of this model ended up being freehand while looking at the toy itself.

This made the bottom suspension exceptionally tough, since by the time I made it to that portion of the model, I was already vastly off any helpful reference, and every approach I used to model the suspension ended up have considerable issues that I couldn’t figure out how to solve.

Problem

For this exercise, I needed to model a high heeled shoe based on reference, exclusively using NURBS tools in Maya 2019.

Solution

For my solution, I started with the base of the shoe, figuring this would be the most important piece for lining up the rest of the surfaces. I initially drew a shape with the CV curve tool to match the shape of the sole, then duplicated it, moved it up, and used the loft tool to connect them. However, I could not fill in the shape of the shoe itself (as in, the space within the closed curve shape.) I initially tried to use Planar to fill this hole, but the resulting surface was very difficult to move into the proper, angled shape of the shoe’s sole. Instead, I tried to move the curve itself into this position. When I tried this, Planar would no longer create a surface within the curve (as it was no longer a flat curve that Planar could extrapolate). At this point, I was at a loss with how to proceed.

Eventually, I restarted my process, and used a NURBS Cylinder primitive to quickly get the overall shape. However, this too had its problems. I didn’t have the necessary control points to make the surface properly shaped, and much of the sole (and especially the bottom), ended up with incorrect surface details. Because I couldn’t remake the topology of the cylinder, I was at a loss on how to fix this as well. Eventually, I just moved on to the next piece of the shoe.

Next, I began working on the peg on the heel. This was fairly simple: I started with a NURBS circle primitive, then duplicated it several times to get a series of circles of the proper shape. Then, I lofted between each one to create the peg. However, here we had another issue resulting from the cylinder we used from the sole: because of the way that I shaped and angled the heel of the shoe, it was very difficult to properly connect the sole to the shoe itself, so the connection there is misaligned.

I moved on next to the back of the shoe, which was fairly simple at the start: I used the CV Curve tool to make the half pipe shape of the back, then duplicated it a few times to properly scale and adjust the back as it moved up. Then, I used Loft to create the surfaces of the shoe. However, I once again had issues connecting this part of the model with the rest. Because the back of the shoe appears to wrap around the bottom of the shoe, I attempted to do this was well, adding a few Isoparms to my curve to wrap over the shoe and the poor connection with the peg. However, I had a great deal of trouble making the bottom of this piece look correct, and so I tried to make the connection point look as neat as I could (the bottom of the heel turned into a mess on my model), then I moved on.

The straps themselves were fairly easy: I started with a CV Curve tool, made a curve, and placed it into the position of one of the corners of an X-strap. Then, I duplicated the curve three times, placing each at one of the different corners of the strap, then Lofted between each one. However, just like my first attempt at the sole, I was unable to create a cap for the straps, as I couldn’t use Planar to do it. After I had made the first strap, I used duplicate special with a -1 scale in the Z direction to create the other strap. I then made the right strap rest on top of the left strap to match the reference. To finish the connection of the strap to the back of the shoe, I used a NURBS cube that I scaled to fit into place. Lastly, the toe strap was fairly simple, and I once again made a CV Curve, duplicated it three times, and lofted it to create the shape of the strap.

Pictures

Challenges

I definitely struggled with this exercise. I feel that I have a decent understanding of how to create and manipulate Curves in Maya, but I was having a lot of trouble manipulating Surfaces to get them to move the way that I wanted. I was unable to find a way to create a surface within a closed curve unless that curve was completely planar, and this really presented a hurdle to me at many times during the exercise.

Also, the choice to use a NURBS cylinder for the base really hurt my workflow, as I didn’t have many options to manipulate the surface topology in a satisfying way, and it became very difficult to shape many of the other pieces to fit with the shape of the base that I had. In hindsight, there may have been some settings that could have helped me, such as adding divisions to the cap of the cylinder, though I didn’t find that option when I looked for it.

Overall, I ended up with a pretty sloppy model, because I could never figure out a way to fix these issues, so I tried to hide them with other elements, just causing more issues to propagate as I worked longer.

As a question for the future, do you want me to provide pictures of the problems a model had for your reference? I chose pictures that would show off the parts of the model that worked the best this time around, but I specifically avoided giving a picture of the bottom of the shoe, as that is where most of the problems ended up being hidden. Let me know what you would like my pictures to focus on!