Previsualization is the process of making a stripped-down version of your film/animation sequence before actually making it. Usually made within the 3D program, a previs is animated very basically, just enough to get the story across.
This time, we jumped straight into the task, instead of doing any prior research because the task itself involved quite a bit of research. We were given three items: a bicycle, piano and projector, and told to choose one of the three items to work with, the other two being backups. We would then be creating a pre-vis version of the model as well as planning it out as if we would actually be modeling it. The piano drew me in most, since I have always found them to be very beautiful and majestic instruments, and it would be great doing some research on one. My first instinct was to google ‘piano’ and see if anything useful came up. I found a handy guide on the different types of pianos there is, and this came as a surprise to me because there were more than expected and each with its own unique build and purpose.
I knew right away that I would be doing a medium or concert grand piano and the matter was simply of choosing a more specific model, so I looked through a few websites and I found a very professional looking company that specialised in the production of grand pianos. There, I chose my model, and using the website and Google, I found all of the relevant information and images that I wanted for the task. Its name is the Classic Grand and it is model B.
I then looked into grand piano’s more in general, in order to better understand the model that I had chosen. I did this through researching how they are made in the real world, and through my reading, I discovered that it is quite the process, which shouldn’t have been surprising to me. The outside rim is made first, connected to the inside rim, before both of which are then attached to the soundboard, which is made separately and is what allows the instrument to be played. I would like to link the websites that I used since they have useful links within them with extra information that would be too time-consuming to plough through.
————> Yamaha Instrument Guide
How Products are Made <————
I also found some practical images that helped me understand the different components of a piano and would certainly be used as a reference for a 3D modeller in creating one.
The size and scale of the model that I had chosen were all specified on the website and I even found a guide that would help me when modeling.
Next, I looked into how a Grand piano moves and if there was anything about this design that made it unique. Once again, an image came to my rescue. Since I find it much easier to understand and remember information visually, I did not shy away from answering the questions with images sometimes, and this one clearly showed how the movement of the keys related to the inside of the piano.
Model B in particular is an exclusive design that provides the tuning pin with smoother movement under torque, a more uniform retaining action, making it a piano that holds its tuning longer, and a popular medium grand.
In order to help more extensively with our research, which is an essential part of planning a model or 3D scene, we were given prompt questions to answer and delve further into if necessary, which I will put here, but will answer all in one go since that is how I did my research on the day.
My model won’t be animated, however, it will be a hero asset and the main subject of the shot. Because of this, I will need to pay more attention to the details and take care to make it look realistic as if it has been used or is able to be used. Looking at the blueprints and templates that I have found, I believe that between 6 to 10 primitives will be used to create it. This largely depends on how much I want to model, since I could go right down to the small cogs and inner-workings, however, it is not my aim to capture the mechanics since it would be hard to find orthographics for them, it would be difficult to make them accurate and that would be too tedious and time-consuming of a model. The process of building the piano in Maya would be somewhat similar to how it is created in real life, starting from the main outside materials/panels and bringing them together, however, I will not be building the inside entirely separately, since I only want to create what will be visible, not every single inside part. Most likely, the soundboard and visible pins will be extruded, and the strings made separately if straight, or with EP curve if bent/curved.
If I was to animate it, I would make the keys movable, since they are the main parts of the piano that move. However, when being played, the inside keys also move, so I would have to consider linking them somehow.
As in most pianos, tuning planks or pin blocks are usually made from maple and in the better quality grand and upright pianos, the pin blocks will be laminated. Ash and walnut wood, ivory, brass, high carbon steel, wool felt and aluminium are other very common materials found in piano manufacture. So for the materials that I will use, I would like to have a reflective surface so I will be using Phong E or other metallic materials for the main body. For accent details and some internal parts, wood would be a more realistic option, so things like Blinn or Lambert could be used to mimic the effect since they have no highlights.
I plan to use cube primitives for the legs, body and lid, as well as the keys. There aren’t too many round objects in a grand piano, the edges are sleek and curved, so I will be using the verts extensively and maybe even SMP. Symmetry isn’t an option that I will be using much since Grands are asymmetrical, which adds to their beauty. However, the mirror function could come in handy for the music rack. Since it is a simple object, I might find it faster to model it fully. I would also like to create a piano stool if I have time and duplicate can certainly be used here, for the legs, as well as the legs and keys of the piano too. I don’t believe that I have anything that I have modelled before that could be used since I haven’t worked on any piano-related components, but models can be pulled up for reference for certain techniques, such as when I used the verts to create the spout of the Utah teapot, or duplicate special for stairs (could be like keys?).
Now for the task – creating a pre-vis version of the object in Maya. We were originally supposed to have only 10 minutes to roughly lay out the primitive shapes of the model, but we were given some extra time (not that it helped me a lot anyway, my model still turned out pretty poor). The challenge was to use scale, rotate and move only, and no extrusion, since we weren’t actually modeling, but creating simple visualisations instead.
After importing my reference images, I began putting in basic primitives and trying to figure out a good method for this new technique. There are so many things I wish I could have done better, such as the legs – why didn’t I scale parts in slightly so they match the reference better? I simply didn’t know how simple to go and I didn’t know how much time I could spend on one aspect before moving on to the next. Everything was a guess and it was difficult to understand the orthographics when I was trying to build ‘simple’.
Using the verts was probably the only part that I knew well and got right. I added in edge loops along with the cube that I was using for the main body, and moved their verts in along the edges, roughly matching the curve of the body on the reference image. I think I must have scaled the images wrongly on the grid though since they didn’t match up to each other as they should have. I don’t know why I didn’t fix this at the time, I must have been very confused and trying to translate the front, top and side parts to each other.
See here, I added a panel that was completely unnecessary out of confusion. I couldn’t understand why the legs didn’t match up to the main primitive, and why there were extra panels in the front image, so I added and stretched out a primitive, trying to understand how that would work and if it would help. I also had to get to the top so I abandoned that, hoping to come back to it, and quickly made the propped up lid using the same moving-the-verts method. Our time ran out just as I was rotating it into place and I couldn’t get back to the extra primitive, or the other components that I wanted to make. Here are the final screenshots, which I wish I didn’t have to display, because I am not proud of them, however, they are evidence of my work so they need to be presented.
I didn’t like this process at all. Even with the reference, it was difficult to plan because of the great difference in detail. What I mean by this, is if I were to model the piano properly, I would fully, or almost fully complete the shape I was working on before moving on to the next, as I have done with all of my previous models so far. However with this, ‘thinking simple’ meant a lot of jumping around, making different parts whilst previous ones were unfinished. I had to go against my nature in making something look perfect and leave it rough and weird looking, whilst working on some other part. Perhaps this wouldn’t be the case normally but there was something about pre-vis and planning that made me hate what I was doing and the objects I was making were horrid. Perhaps it was the time limit or perhaps it was the fact that I had never done this before, but I was incredibly frustrated and angry and I wanted to delete everything, so I was glad when we moved on to the next part of the lesson.
Intro to nCloth
nCloth is a fast and stable dynamic cloth solution that uses a system of linked particles to simulate a wide variety of dynamic polygon surfaces.
It is flexible enough to simulate many different types of surfaces and to better realise its versatility, we were shown a few showreels and clips of 3D animations that make use of it. Some of the examples included, as well as a few others we thought up of as a group were bedding, clothing, curtains, capes, flags, posters or papers (like money), a suspension bridge, volleyball net, fabric and the sails of a ship. I was excited to hear about all of these options since they showed that nCloth had large capabilities and could add to an environment or model’s realism.
When we got round to trying out this effect for ourselves, we were given a PowerPoint demo to follow, so that we could each work at our own pace, understand the instructions better and refer to the steps as needed.
The first thing I did, following along with the steps in the PowerPoint, was to change the Maya functions from modeling to FX, in the top left bar which determines the workspace functionality.
Next, for the space in which the simulation would take place, I created two poly planes and a sphere, arranging them so that the sphere was sandwiched in between the two planes, with some space between each. Another important thing to note is that the second poly plane was not duplicated, but generated as a separate primitive.
After this, with the top poly plane selected, I went to the attribute editor and clicked on the ‘pPlane2’ tab. From there, I added more subdivisions to the top poly plane by clicking in the subdivisions width box and changing the value to 20. I did the same in the subdivisions height box.
The next step was to change the top poly plane into a ‘nCloth’. To do this I selected the top poly plane and then clicked on the nCloth dropdown menu at the top of the screen and selected ‘create nCloth’. This was the step that was key to the whole simulation since it turned the flat plane into a cloth. The stimulation was already there, but without a solid object for the cloth to interact with, there was no point hitting play just yet.
Afterwards, I selected the bottom poly plane and clicked the ‘nCloth’ dropdown menu at the top of the screen again, this time selecting the ‘Create Passive Collider’ option, in order to give it physical attributes. I did the same thing for the sphere; turning the objects into physical colliders is what allows them and the cloth to interchange since they now have physical value or physical attributes.
It was almost time to animate the cloth. I was first told to ensure that there were enough frames of animation on the bar. Just underneath the time bar to the bottom right of the screen, there were two boxes. I change the first to 600 and pressed enter and the second box changed automatically after this. I then clicked on the symbol of the little man running with a cog over him, which was on the bottom right corner of the screen. A window containing more specific animation playback preferences showed up, where I clicked the playback settings drop-down menu, and selected ‘Play every frame’.
After this, the animation was read, so I clicked play and watched the poly plane fall onto the sphere and wrinkle up as a cloth! It was really impressive and wasn’t as complicated a process as I thought it would be.
Now for the fun part! I selected the cloth and jumped back into the attribute editor, clicking on ‘presets’ to get a list of different materials to appear. These presets are what can be experimented with; choosing different materials creates varied and interesting effects. It was our chance to really see what we could do with the baseline of the function in Maya (the presets are always just scratching the surface, hence why they’re called ‘presets’, designed so that the material acts a certain way which is often used. The names were quite amusing and I had fun scrolling through, just looking at the options, before finally diving in and clicking on the ones that intrigued me most.
My favourite part was definitely experimenting with the presets and applying different options to the cloth. They each altered the way the fabric behaved slightly and it was fun to imagine what situations they could be applied to, such as chain mail, which would obviously be used to metallic netting (it made me think of the scene in Maleficent when she transformed into a dragon and the giant iron net was thrown over her).
After we were satisfied with a particular animation/video, in order to render it and add it to our blogs, we had to right-click on the timeline and tick the box next to ‘playblast’. Without changing any of the settings, we clicked ‘save to file’ and chose a place to save it on the computer and clicked the playblast button straight after. The video played through and once it was done, I dragged it from the desktop into Medi Encoder, so that it would be a compressed file small enough to be put onto my blog.
It was great that we were shown how to do this because I wanted to create more than one render since I enjoyed playing around with the presets and had a couple of favourites. It turns out, I should have been clicking ‘replace’ each time I wanted to change a preset, but instead, I was using the percentage options. I think these were layering on top of each other or mixing together, and making really interesting movements, so I’m glad that I was doing it sort of wrong so that I discovered this. I also discovered that Maya has notes on each preset, which can be found in a small box in the attribute editor, towards the bottom. It was cool being able to read a bit more into the presets and how they worked.
Burlap at 25% – Falls with less bend/less flow
Lava – 50%. I couldn’t fully replace lava since it made the cloth act incredibly bizarrely, not like cloth or any fabric, which makes sense since it’s called ‘lava’ At half the power, it was more realistic.
Flag Task
We created our own simulations next – a flag blowing in the wind, as an example of what can be done with nCloth with these easy steps (applying nCloth, making passive colliders etc.)
First, we created a flag pole with a cylinder and I even added some details towards the top to make it more interesting. We also added a poly plane to the scene, which would be the actual flag. After scaling it up, stretching it out into the stereotypical flag shape and rotating it so that it was on its side, we moved it into position. There was no need for it to be touching the pole yet, but it was next to it, to prepare for the next step. In the view-port that clearly showed the flag from the side (this could be the front or the side on), we clicked wireframe mode in order to be able to clearly see the geometry of the objects, and moved the flag into the pole, but not too far in; about the second line was a good place. Any further apparently causes distortion later on.
Then, with the flag selected, we added ‘nCloth’ to make it a fabric material. In order to connect it to the pole so that it could blow in the wind and not fall straight down, we drag selected all of the verts on the flag touching the pole, and then also shift selected the pole, and went to ‘nConsraint’. In those options we clicked ‘point to surface’, which caused all of the verts to turn green, showing that the flag is now knitted to the pole. With both still selected, we went into nCloth again, this time to turn them both into solid objects by clicking ‘create passive collider’. Now, we could create the effect that we wanted, which was the flag blowing in the wind.
In the outliner, we selected nucleus1 and went to the attribute editor. Amongst all of the other options, we found and focused on wind speed. We ever so slightly ‘crept it up’, increasing the value bit by bit so that Maya didn’t crash. Each time, the wind grew slightly stronger and the flag began to billow more realistically. If it was blowing in the wrong way, we were told to look at the wind direction, which was an option underneath wind speed, and change the axis if necessary. Once we were satisfied, we playblasted and rendered.
To fully complete the flag though, it needed some insignia, which was luckily provided for us as an example of how to apply it. It involved opening up a window I have only used once before called ‘hypershade’, which appears once you click the little green ball icon with a white circle on it.
Once that was open, we could see all of our detailed design options for the different materials. We added a blinn to the lineup, selected the checkered box next to colour and clicked ‘file’, similarly to how we applied a kitchen hdri to the skydome of our teapot models. In the file attributes, we selected the folder next to ‘image plane’ where we could search for and find the flag image. Once that was added to the blinn, we could close the hypershade and go back to the flag. Instead of assigning a new material, we added an existing one and chose the blinn with the insignia (for me this was blinn1). After choosing it, I gave the program a few moments to process the material and request and then clicked ‘6’, and the pattern appeared! It looked awesome, and I finished it all off by adding a material to the pole and poly plane because colour makes everything feel more complete.
Before
After
I love the nCloth portion of the lesson, especially after seeing my flag rendered out and blowing in the wind. It would be cool to create a pirate flag, blowing with the sails of a ship on the sea, or perhaps a war flag, tattered and bloodied, standing crooked in a battlefield. It’s fun to imagine how the techniques we learn could be applied to different environments and add to the feel of a scene. However, those scenes and the models in them, require planning, which is the area that I struggled with most. I need to work on carrying through my skills in researching and being able to effectively show how I would create a model, especially since pre-vis is such an essential part of the process, and will be a requirement to evidence in the brief.