Week 5 and 6 – Flower Asset

This week I want to complete the flower animation asset.

Research

Through the various sources that I collected, I determined how I may model and arrange the composition of the flower asset.

This model is a great reference, however, I do not want it to be this simple. At the least it should have more texture detail.

Considering they how only single colours to the flowers are added, this reference is aesthetically acceptable nonetheless. It is the fact that there are multiple flowers in an organised composition that draws focus to the collection rather than individual small detail.

I collected these two videos while looking for references to use while animating.

Looking at these videos of flowers blowing in the wind, it shows how each flower can be influenced differently depending on their shape and weight. If the flower is heavy, with many petals, the stem is thin or long, then the stem will be more influenced by the wind.

I found this reference on how to make the flower petals, with methodological and realistic accuracy. For example. even how the petals were turned at a slight angle, or the positioning of the petals against each other. For now, I will try to animate them normally (with a rig and controls), and if I have time, I want to try this.

In the video they used deformers and lattice or ‘group’ and curve the petals, something that I never knew about. The way that you can uncurl the petal and curl them again, this would be a great way to animate the petals in the wind, easy and quick. One the other hand, I am not sure if I can animate these settings, plus, the petals would move together and not separately.

Flower Design

My personal opinion on these designs is that the stem shouldn’t be straight. I like the curved and hanging stem in design no.7. Or the no.1, 9 and 10’s stem design where there are multiple. My favourite flowers designs, from the ones above are no. 1, 3, 7, 10 and 12.

I kept the designs simple: not too many petals, no more than three stems and a few leaves. The stem should perhaps be the only thing that moves in the wind. I would prefer not to more the petals or leaves. While it would look more aesthetically pleasing and realistic, it will take more time.

Group Member Flower Choices:

David: 5.

Hish: 7, 8, 9, 10.

Sophia: The whole middle row

Mesha: 1, 10.

We discussed using a single flower, a couple of different kinds of flowers (so that there is a variety), or a few flowers on a single stem. A variety would mean that there isn’t a single kind of flower that the bee collects the pollen from This changes things up a little, so that it is less repetitive. I decided to complete one flower for now, and nearing the end of my project, if I have time I will model another species of flower. At which point, I should be able to reuse the rig, as long as I keep the second flower model similar to the first. The final flowers of choice were number 7 first and also number 8.

Rig Design and Movement

I plan to make a few tests on the flower stem rig with a few variations. The flower animations I want to complete were initially only going to consist of the stem swaying in the wind, however, after speaking to my teammates, they asked that I also have wind in the petals, otherwise it would be too static. Although I have made a few rigs myself in the past, I want to experiment with some methods and effects before I make the final rig for the model.

Rigging Research

I found a video that may show what I wished to do with my rig – the curling. By selecting multiple controls at once, the child controls curl with the parent. This is perfect for what I want to do.

Rig Tests

I needed to understand the kind of effects that different types of rigs will produce, learn out how they appear in the animation. I experimented with varieties of joint number and positions, IK handles, and controls.

Used an offset on rig numbers 7 and 9:

Each control, in the ‘tower’ of controls on the stem, is offset by two, starting from the ‘head’ control at the bottom.

My teammates and myself decided that the last rig, number 9 was the best of the tests to use. They all complimented my work so far, and noted that the ones with an offset, numbers 7 and 9, were the best ones.

Flower 3D Model

Reminder: the flower chosen to model first was number 7 from the designs above. As the first experiment, I decided to try out modelling the petals with n-cloth and a passive collider on a mould to help shape it. Like so:

This should produce most of the petals curved shape as well as the creases in my 2D design.

I tried about 70×70 divisions on the fabric first.

I tried changing up the shape of the mould so that it curves at places were the sides of petals may be. I made these different sizes, this way I can hopefully make multiple petals at once with the n-cloth method. I increased the gravity of the n-cloth nucleus, as the the fabric model seemed to fall extremely slowly. Even at 3000 frames it had not assumes the correct position on the mould. I tried 20-50 units of gravity.

One issue with this method was the the fabric did not drape correctly on the curve of the mould. I think if I remake the original n-cloth model so that it just fits off the edge of the mould, this would produce a better effect. And I wouldn’t have to delete any faces later. The extra fabric that hangs off of the mould, drags the rest down because of the t-shirt pre-set I added, this is why it doesn’t drape along the curve of the mould.

There was an interesting effect when I set the gravity to 250 units, if I stopped it at the correct point, I could use that model for my petal models.

The model usually creases along the areas that I have creased on the mould, which is not what I desire. (See below).

I ran a simulations with various settings to find the result that I wanted, and duplicated the models I liked to make petals with later. Looking at the model in the state it is above, I believe I should double the number of divisions to something like 150×150… possibly.

Next I tried a disc polygon primitive, the rough size of the mould, so that the n-cloth model stops at the edge.

I had some software issues so I had to restart. I decided to remake the mould in to a cleaner mesh, as you can see above. The mould is wider, so that the petals are not the same width.

The disc shaped n-cloth model had some strange results in the simulation. However, even if I do not fix this, it doesn’t matter. I had been planning to delete the faces at the top anyway, just as long as the faces further down do not get influenced too much.

The circular model shape makes the creases better than the square plane.

This is the result. I like the result except the indents made by the edge of the mould bottom. Decreasing the size of the fabric model more appropriately solved this.

This is the result. This is perfect to use. I can incorporate all of the duplicate models that I collected throughout the process. I went back used the old mould with the new fabric, to produce some further different models.

To separate the mesh into petal sized segments I used the multi-cut tool. Dragging along the mesh, then using target weld to smooth out the outer edge flow. While I like this n-cloth and collision method, I realised that it took quite some time to clean up the mesh after I separated it into segments after the animation. While the results are appropriate, the time mass clean up time is a waste. I should have just made individual models from the start. In fact, I will do that now. Making individual petals before making the n-cloth animation would be faster than cleaning up all of the mesh anyway.

I just have to stop the animation at the right point, not the let the petal model slide off of the mould. Although, using induvial petals may not produce the crease/crinkles that I want, as the model is always falling. (Perhaps decreasing the gravity will allow the model to rest on the edge.

As I thought, the shape of the petal model is fine (because of the mould), but the creases are horizontal and not vertical. I had another idea to fix this. This method will require much less mesh clean up.

This is my solution. The bridged material in-between should produce the same results as the first n-cloth method (the tablecloth looking method), however, this time I will have much less mesh to clean up.

This is the result. Although I prefer the original method’s result, this one doesn’t need as much mesh clean up. With the multiple stencils, and different side widths that I can place the n-cloth model on, I will produce many different and appropriate results. Applying various n-cloth pre-sets to the model makes various effects, since the petal models currently don’t look very interesting (in comparison to the first method).

Final Petals

Making the Stem

I used n-cloth on the flower bud centre to vary the texture more:

With no gravity nor wind speed, only wind noise at 10 units, did the model still hold its shape enough but was influenced by the turbulence to produce a wavy model geometry.

Arrangement of petals and centre flower bud.

The double layered petals are more aesthetically pleasing, the team agreed. However, it will take more time to animate the petals with both sides if the weight painting is not done correctly. Considering I simply duplicated and rotated the first layer of petals, I would just reuse the rig made for the first layer. Thinking about it now, I should be able to rig and animate one layer and duplicate the model and animation on top like the image above. So long as the layers do not conflict/collide/ go through each other. I may need to edit the animations, but this should be a quicker method than making everything for the petal layers individually.

To add extra stems onto the flower stem I used extrude along a curve and Boolean to make geometry to connect them together.

I used ‘union’, see image above. Although, ‘difference’ would have worked. When I cycles through them, I figured that with union, it should be quicker and easier, since I already have the start point for the stem. All I need to do now is fix the geometry by target welding and then bridge the gap between stems.

This method was too messy and I had to add too many edge loops. This is not a clean method that I would have liked to use. Therefore, I restarted on the stem model all together. Instead this time, I extruded the whole stem along a curve- for a cleaner mesh. Then made places suitable for the extra stems to extrude from. Then, I will extrude these along curved lines also. Like so:

This was quite messy. I decided to change my technique and instead extrude the edge of the hole, detach those faces from the main mesh, and bridge the two connecting edges.

I modelled a few leaves which I can sculp to produce additional varieties of leaves.

Rig Complete

Rig Animation Test:

Petal Movement Prep Notes

I am a little unsure of how I should animate the petal movements, however I do not want to spend too much longer on this as I will go over into the next assets time. I spoke to my tutor and his suggestion was to first look at a reference. But basically the movement I should make with main petal control is: Orig position>forward>up>Orig position.

Try add squash and stretch to make a little more interesting. The down position would be squashed and up position would be stretched. If this is even possible with the controls that I have made.. i will try. A little cartoony?

Texturing

While looking at tutorial videos I realised that by the time they started rigging, they had already completed their texturing. Therefore, to ensure I was following the correct workflow, I stopped where I was and worked on the textures in Substance Painter.

Colours

I used the original design sources as colour references, but I used the photoshop colour library to choose similar ones that fit well together from Photoshop (swatches above).

Substance Painter

While I have made simple textures before I prefer the method that substance painter uses where I can paint straight onto the model and Uv’s. However, I am not very experienced with this software, so it took a little time to understand it.

I found that using texture sets would stop this from happening. The model textures set were in one piece, so the centre flower bud would be painted when I made the petals.

I used this useful video to fix this issue, and separate the parts that needed to be painted separately. I am slowly learning how to use Substance Painter as I go.

Exporting/ Importing the Texture Maps

I only added the base colour map, normal and roughness map. There didn’t seem to be anything else that is necessary.

Textures Finished

Weight painting

Three sets of petals are controlled to one rig section, which is controlled by a control. As you can see above.

Result of weight painting. Errors since the petals are double sided. I cant move the petals too much. I have spent too much time on this already. I will just move the petals to just before they deform. I can also add a small amount of squash and stretch, as I intended. This should be enough, unfortunately this is not what I intended.

Animating

To speed up the animating I used a rotoscoped from a reference video. A side and front view, using different flowers in the same video, to get slightly different wind effects that should move at the same time (wind noise).

My animation workflow for this asset includes:

• X-axis rotation of the stem (Front view).
• Z-axis rotation of the stem (Side view).
• Add more rotation to the flowerhead – swaying that fits with the wind that impact the stem.
• Add offset to the step controls.
• Edit graph editor curves so far.
• Petal translation (slight movement that doesn’t effect the weight painting imperfections).
• Squash and stretch for the petals (stretch when moving up, and squash when the petals move down).
• Graph editor.

I will smooth out the curves and make them more gradual, and speed up the down parts of the arcs a bit. This should make it seem that wind impacts it. Considering I used rotoscoping, the reference had slight wind only. I can make the graph editor curve units to increase speed (make a few samples to show the team).

I animated very limited movement on the petals. They are too stiff because of the messed up weight painting, which I do not have time to intricately figure out. I should have made the petals as single layered objects (planes). there would be less problems, but the petal would have to be flat.

I cannot move the petals as much as I may have want to, this is a large issue if the team does not like the animation, since I need to move onto the next asset promptly. I produced about 12 seconds of animation, using offset on the stem, and squash and stretch on the petals. I varied the wind strength, in how much the flower swayed by making multiple versions and moving groups of key frame units in the graph editor. They I made versions of wind speed, how quickly the flower swayed, by squashing or stretching the key frames on the animation timelines.

The feedback from the team, one said that it was ‘perfect’, visually and fit for the purpose. However, there was the issue with the petals mentioned. Therefore, if I have time at the end of the project I can edit the model, textures and painted weights.

Animation Play Blasts

These are all of my tests with the animation, the last few are were I experimented with the length/speed of the animation, as well as wind strength (extent of movement).

Asset Overview

Overall, this asset has been quite successful, while fulfilling the necessary requirement of the game animation asset. While, I had some complications, the animation outcome is suitable for the purpose, and I received positive feedback from my teammates. If I had to do it again, I would have liked to try animating the petals like in the link below. I think these petals would have been very aesthetically pleasing.

This animation asset took me longer that I would have hoped, a week in fact, instead of it taking me a single week. I need to move on quicker than this with the other assets. When I can help it, I should try to use premade models and rigs from online, otherwise it will take me too long. The bee wing animation should not be as complicated as I had originally hoped. I will make the simple wing movement and allow time for other important things, like figuring out how to do the hair on the bee.

Final Render