After the end of rendering, it enters the stage of post-color mixing, which is an important link to ensure that the color of the picture is uniform, beautiful, and in line with the mood and theme of the work.

First of all, it is necessary to pay attention to the unity of the overall tone, adjust the exposure, contrast and saturation to highlight the main body and the sense of hierarchy, and clarify the overall mood of the scene. Here, my animation mainly needs warm sunlight, so I will use the warm tone for Zelda’s scene style as my reference screen and color reference standard.

At the same time, the white balance is corrected to avoid color bias. Adjust color tendencies according to the mood of the scene, enhance light and shadow details such as highlights, shadows and light effects, and add stylised effects such as particles and halos if necessary. Ensure that it is done on professional color calibration equipment and compared with the reference work or the original image to ensure that the desired effect is achieved, while selecting the right color space and high-quality output format to preserve detail.

In terms of post-production music and sound effects, since my animation does not include dialogue, I only needed to record some vocalizations and breathing sounds.

Sound effects were selected based on specific scenes. For example, Harlow’s entrance is always accompanied by the sound of bells, inspired by the Tinker Bell movies. For the cat’s appearance, I used a heavy thud sound effect to emphasize the ground-shaking impact.

Regarding background music, the story’s tone shifts between relaxed, exciting, lighthearted, and comedic over a duration of one minute and forty seconds. Finding copyright-free music that perfectly matched was challenging.

I decided to use Suno AI, a music creation platform. My experience with it is similar to using MidJourney for AI-generated images. Suno AI can produce music with or without lyrics and allows for keyword adjustments to extend or modify existing results, making it a powerful AI productivity tool.

What’s holding me back on rendering is motion blur,
First of all, motion blur is a kind of visual effect commonly used in animation or film and television production, which simulates the blur effect produced by fast-moving objects in actual photography. When the camera’s shutter is open, objects in motion move within the shutter time, creating a stretching or blurring effect in a single frame image. Motion blur adds realism and fluidity to the picture, especially when representing fast movements.

As mentioned in several previous blogs, my character moves very quickly in some shots, such as 008 and 017, and this motion blur is not limited to the character, in shot 010, the character and the camera are relatively still, so the motion blur here is reflected in the background

However, when I was rendering, I noticed that there was no motion blur in the originally rendered shot, and I filled in the instructions for motion blur in the Console Variables with a value of 4, but it didn’t have the effect I wanted. But after I kept checking, I found that first of all, in the camera Settings to turn on the motion blur option, you can set the MIN and MAX values, and their matching frame rate. After turning this option on, my motion blur was rendered successfully

I didn’t adjust too much on bloom and lens flare, because it didn’t affect my overall effect much, and the motion blur turned on May have partially affected bloom and lens flare’s effect

Lighting is also important in the scene. First of all, my scene is an outdoor scene with SkyLight and DirectionalSunLight, which can make my scene have a bright atmosphere and appropriate shadows, which is of great help for the rendering of scene shots. When I need to focus the camera on harlow’s face, there is usually no light on harlow’s face, resulting in her face being dull, which is also related to my character and action design

So my method is to light each shot one by one, and avoid harsh light. The light type I use is spotlight, so that I can achieve soft light effect according to the adjustment of the light source distance, and because my role does not have the effect of Subsurface Scattering Shaders, So I controlled the color of the light between skin tone and light pink to simulate the feeling of ruddy skin

This is before and after the light

Once all the animation was completed in Maya, the next step was rendering. I used Alembic as the file format to save and load the animation. Before exporting, I made sure that the model’s geometry didn’t have any faces with more than four edges. In Maya, I combined the geometry in face mode to ensure that the Alembic file retained the materials.

For the scene selection, since my character is a cartoon, and the scene is a combination of a forest and grassland, I designed the environment to be more stylized. I built the first version of the scene when I completed the first animation shot and performed some test renders to ensure the project was moving smoothly.

However, when I continued to promote my animation and collected a lot of data, I felt that my scene lacked details, my texture was too rough, the scene light was too stiff, and the style of the flowers in the foreground was inconsistent with the trees in the background, so I quickly began to conceive a new scene.

I have considered it based on my second role; It looked like an alien cat to set up a semi-alien world with an alien forest of fantastical plants, but when I confidently adjusted the camera and introduced Harlow into the scene from maya, my aesthetics told me that the scene was too realistic and had too much textural detail for my characters to fit in.

So I have the scene now, in the stylized grass with detailed trees and many kinds of flowers, the perfect realization of my story. In the construction of this scene, I refer to the animation of Miyazaki Hayao and the scenes in Zelda games. At the same time, I also use some assets to build my scene quickly and efficiently.

Shots 016, 018, and 019 focus on close-up shots of the cat’s face. I’d like to provide a detailed explanation for shot 018. In this shot, I modified the original wide shot because I felt that working with a wide shot would require considering the entire body twist of the cat, its center of gravity, and the order of movements, which would be time-consuming and lack impact. Instead, I drew inspiration from scenes in Japanese anime Haikyuu!! and Spy x Family, as Japanese animation is highly skilled at expressing tension through animation. Below is the reference video I used.

For expressing tension, what came to mind were camera shakes and strong winds. To achieve this in my animation, I used a ragdoll dynamic. After attaching the corresponding physical simulation capsules to the ears’ controllers, I applied air and turbulence forces. The air effect provided wind strength and direction for the capsules— the direction of the lines indicated the wind’s direction, while the length of the lines represented the wind’s intensity. The turbulence was used to disturb the existing wind lines, adding an extra layer of dynamic motion to the shot.

This can be seen more clearly when the geometry is removed

Shot 21 can be considered a looping animation of the two characters. I continued the earlier Harlow flight animation, modifying only the face to save time. For the cat’s animation, I used reference footage of a tiger running in a loop, as the size and weight of the tiger more closely match the characteristics of my character. This helped me achieve a more believable and dynamic movement for the cat.

Shots 12-17 represent a shift in the emotional tone of the animation, moving from tension to calmness to joy, and ultimately leading up to a high-energy soccer kick. Personally, I’m not very skilled at creating subtle body movements that reflect emotional shifts, so I made a lot of adjustments.

In the initial blocking phase, I didn’t use reference material or shoot any reference footage. Instead, I relied on my own ideas to pose a few key positions. I felt the timing was close to what I envisioned, so I considered the early stage complete.

However, upon reviewing the initial blocking, I noticed many issues, including the camera rotation, as well as the rotation and follow-through of Harlow’s spine and arms. To address this, I decided to record my own reference video, which significantly helped me in refining the animation.

This is a comparison video showing the differences before and after recording the reference footage.

Shot 017 was primarily based on the reference video I recorded, combined with curve adjustments. For the reversal part, I used frame-by-frame keyframing to ensure smooth animation. During the soccer kick, since Harlow and the ball were positioned far apart and there needed to be a quick camera movement to showcase the power of the kick, I specifically adjusted the waist controller. This helped me achieve the desired effect and meet the ideal standards for the shot.

For Animation 005, I continued to use the approach of starting with a cube. However, for Shot 006, where the cat takes a step forward and roars, the cube approach became less applicable.

For Shot 006, I adopted a method of building the animation layer by layer using controllers. First, I clarified the timing and distance of the cat’s two forward steps, as well as the duration of the preparation and roaring phases.

Since it was very difficult to find a reference for this animation, I recorded my own reference video, imagining myself as a monster emerging from behind a cliff.

Shots 007, 008, and 010 all feature Harlow’s flight animations. The total frame count is very short, but the shots are highly expressive.

First, for shot 007, Harlow is escaping. Since the entry direction for the animation in shot 008 is from right to left, Harlow’s exit direction in shot 007 should also be to the right. During the turning animation, Maya’s built-in breakdown algorithm proved ineffective, so I ultimately chose to pose the movements frame by frame. I carefully drew the curves for the head, hands, and feet to ensure that the details captured by the audience’s eyes in this extremely short time frame remain smooth.

For shots 008 and 010, I started by creating a stationary flight loop animation. Since the essence of these two animations lies in variations based on the base animation, I placed the flight animation in the base layer of the animation stack. Then, I selected the necessary controllers and added new animation layers to make the modifications required for these two animations.

Shot 009 is a rather challenging one, requiring a solid understanding of the fundamentals of quadruped animation. I repeatedly revised the feedback George provided, but the timing was still considered too fast. George suggested starting this shot with a cube as well. Key curves to pay attention to include the torso, all four paws, the tail’s tip, and the head.

This is my reference

Shot 011 can be viewed as a death animation where the character interacts with the ground. It’s important to focus on the speed of the character’s forward slide, ensuring it gradually decelerates before coming to a stop. The swinging of the legs should be staggered in position, and the force from the character bouncing back off the ground should also decrease over time.

001,002,004, these three lenses are mainly in the middle and long distance of the focal length, which can show the details of the flowers and harlow’s overall movement trajectory. Due to the short shot time, the character moves fast. So I don’t think the blocking-spline animation method I’ve been using is as good in terms of progress and effect as the one George taught about using a cube as a way to observe the timing and spacing of the entire animation.

So I started my animation with cube. I wanted to fix my timing and spacing as early as possible, but even so, in the process of improving the animation, I occasionally broke or overruled the timing to ensure smooth animation.

This is a screenshot of different completions in the same frame

This is a video comparison of the entire animation from cube to polish stage