How to Enter the Top 100 of the Internet's Largest CG Challenge by Pwnisher

So, you want to know how to crack the top 100 in Pwnisher's 'Endless Engines'? This isn't just any competition; it's the Super Bowl of 3D artistry. Every year, it pushes the limits of what's possible in CG. You could be a contender with the right tools, techniques, and of course, a little magic. The magic here is Fox Renderfarm, the leading render farm in the CG industry that provides cloud rendering services. Now, let's break down how to turn your 3D dreams into reality.

Spearheaded by the industry legend Pwnisher (Clinton Jones), this prestigious annual competition has been setting the bar for 3D artists worldwide for six years running. Recognized as the pinnacle of creativity and technical prowess in the CG community, this challenge is a must-enter for any serious 3D artist.

Having admired the creative brilliance showcased in previous years, our team at Rayfire Art Center was eager to join the competition. Among thousands of entries, our work managed to secure a spot in the top 100, a testament to our team's dedication and creativity. We are thrilled with this achievement.

Final Result:

Today, we're going to share the behind-the-scenes story of our competition entry, detailing our creative process and production methods. We hope our experiences can inspire you, and maybe you'll be the next name on the top 100 list.

Preparation of the Work

The rules for this 3D rendering challenge are actually quite simple. The official organizers will provide a scene with a camera already set up, and all participants need to do is replace the assets within the scene while keeping the camera and its lens settings unchanged. The real challenge lies in how to showcase your creativity and technical skills within just a few seconds of animation. It's worth thinking carefully about this before you start production. The theme for this year's 3D rendering challenge is 'vehicles,' and the official prototype animation is shown in the image below:

Challenge Template © Pwnisher

Based on the theme and rules of this competition, we've outlined the following plan:

Scene Models: We will search for available resources.

Main Character and Vehicle Models: These will be created independently.

Animation: We'll create the animation independently, render it frame by frame using Arnold, and then use After Effects for special effects and post-production.

Theme

The next step is to brainstorm and determine the theme of our animated short. We have a few creative concepts in mind,

Given the recent surge in popularity of AI-generated art, we decided to conduct some preliminary experiments using AI to generate initial concepts.

AI-Generated Images

AI-Generated Images

Analyzing the winning submissions from past competitions, we've concluded that a compelling narrative and innovative ideas are key to success. Therefore, we'll focus on crafting a captivating story within 4 seconds. Our final storyline is as follows:

1. A robber swerves into the scene on a motorcycle, sending bystanders into a frenzy.
2. A police car pursues the robber, only to be met with gunfire.
3. The robber disables the police car and makes a getaway.

We've looked to similar films and TV shows for inspiration, hoping to capture the visual style that we want to achieve.

Images sourced from Google

Images sourced from Google

Images sourced from Google

Next, we will analyze the concept scenes.

Reference:

© Pi Xiu

© Pi Xiu

With the story finalized, we dived into creating our initial concept draft:

© Pi Xiu

Character

Initially, we conceived the characters as a pair of somewhat comical robbers with a touch of clumsiness. However, we later realized that such characters might be too monotonous, lacking in storytelling depth and interpretive potential. So, we continued brainstorming and decided to design the two characters as a fat man and a thin man, adding contrast and humor. Since we had a thin character and they were comical robbers, we immediately thought of the Joker (Heath Ledger) as a prototype. But later, we felt that a female Joker would better convey a sense of wit, acuity, and a touch of madness. Then we thought of Jinx and Harley Quinn. To further enhance the characters' appeal and because we personally liked the idea of ordinary people cosplaying as superheroes/villains from the movie 'Kick-Ass,' we decided to design the characters as ordinary robbers cosplaying as superheroes/villains. Ultimately, we designed the female robber to look like this. The makeup was inspired by the makeup in the movie 'Joker,' the hairstyle referenced a cosplay look similar to Harley Quinn, and the clothing color scheme followed the classic Harley Quinn palette.

© Pi Xiu

Moreover, subsequent animation of the character revealed challenges in rigging the coat for realistic movement. To address this, cloth simulation was implemented in Marvelous Designer during post-production.

© Pi Xiu

To differentiate the heavier male character from the female robber, we designed him as a somewhat dim-witted, obese figure. Continuing with our previous concept, if he were to cosplay as a superhero or villain, it would naturally be Batman, the Joker's long-time nemesis. The added contrast of the small mask instantly created a comical effect.

© Pi Xiu

Vehicle

As vehicles are the central focus of this animation, we paid special attention to their design. To maintain a consistent visual style throughout the scenes, we designed the robbers' vehicle as a charming tricycle, which aligns with the female character's personality and interests. We also incorporated more rounded shapes into its design.

© Pi Xiu

We originally planned to customize the passenger seat with various anime and game logos, but IP concerns forced us to simplify the design. We kept only a large JINX logo, later modified to 'JINY'.

© Pi Xiu

We wanted the cop car to look cool but serious. Since you can't really see it up close and we were running out of time, we just did one quick design.

© Pi Xiu

Background Scenes

We determined the building silhouettes and adjusted their density to match the video's pace. To enhance the scene, we added foreground details, like overpasses and shopping malls, and varied the heights of the buildings. We also included a hidden logo. After experimenting with other elements, such as oncoming traffic, we decided to remove them to focus on the main theme.

© Pi Xiu

After that, we started making the models look really good. Because we were short on time, we used these repeating textures and added lots of little details to the models. We made the buildings in pieces so we could put them together in different ways. To make the buildings look different, we used this cool software called Arnold to mix materials. It saved us a lot of time! And of course, we used a bunch of ready-made materials too.

© Pi Xiu

We made it night time, so we put up a bunch of those bright neon signs to make it feel cool. And we made the ground look more interesting by mixing different textures.

© Pi Xiu

Animation

First, we checked to see if the animation on the model looked okay. To make sure that if we changed the size of the model later, the animation wouldn't get messed up, we put all the important animation stuff at the very top of the model's hierarchy. Then, we put the smaller parts of the animation lower down. This way, if we change the model, the animation stays the same.

© Pi Xiu

We quickly got the skin on the model using a plugin called BonePro and started animating right away. After we finished the animation, we went back and made the skin fit the animation better so we could change it easily. When we looked at the first version, it was too much stuff happening in just 4 seconds. So we focused on the gun shooting and the motorcycle moving around. We thought the ending with the police car in the middle was kinda boring, so we changed it.

© Pi Xiu

We changed the ending. The second police car crashes into a wall and goes off the screen just as the robber gets away. This leaves the screen empty, which makes the story feel more finished.

© Pi Xiu

The primary focus of the animation was placed on the female character, with key actions such as firing a weapon and head turns. The male character's animation was simplified to provide a contrast. A second review revealed a lack of dynamism in the motorcycle animation, which appeared rigid and block-like. To address this, additional animations were created for the fenders, body, and wheels, enhancing the overall visual appeal.

Rendering

Arnold Maya was selected for its offline rendering capabilities, which aligned with our team's expertise. The renderer's flexibility allowed for seamless adjustments to material networks, lighting conditions, and post-render layer separation. A referencing workflow was implemented for efficient layer management. 

A nighttime atmosphere was established using a low-intensity, cold-warm contrast HDRI. To enhance the scene's depth and define light outlines, atmospheric volume fog was introduced. Neon signage was realized through a combination of self-illumination and MeshLights, while AreaLights were employed to accentuate building silhouettes.

Following lighting environment optimization and the application of camera motion blur, keyframes were rendered for evaluation. Iterative refinements were made to the lighting until a satisfactory result was achieved. Layer-based rendering was employed to produce the final image sequence. Individual layers representing the street, ground, vehicles, lights, vehicle reflections, volume fog, and Z-depth were individually adjusted.

© Pi Xiu

My computer couldn't handle rendering this, so I tried using a bunch of computers at once, but it was still too slow. I ended up using a supercomputer farm to finish it. Next year, I'm going to try using Blender or Unreal Engine 5 instead.

Visual Effects

Before we started making the cool special effects, we drew some pictures to get a better idea of what we wanted. We drew the car exploding, the bad guy shooting, and how the cars would move around.

© Pi Xiu

For rapid iteration, a proof-of-concept was generated by combining the completed model and animation. Subsequent modifications were made based on this preview. Initially, Unity's Recorder plugin was employed to output an alpha channel, but due to the limitations of the URP, this was unsuccessful. A green screen workflow was adopted, enabling accurate extraction of the green channel in After Effects. However, the resulting composite exhibited aliasing artifacts in the special effects. To address this, a dual-layer compositing approach was implemented. The primary layer was generated by manually extracting an alpha channel using thresholding techniques to preserve the main subject, while a secondary layer was created by removing the black background and applying a screen blend mode to enhance the bloom effect.

© Pi Xiu

While working on the explosion sequence, I considered how to create a memorable, high-energy impact. I drew inspiration from Yuji Nakamura, whose work on high-energy action sequences in anime like 'One Punch Man' features a signature "black and white flash" technique. This technique involves filtering out excess color information to create a striking visual contrast between black and white. This technique has become increasingly popular in recent game promotional videos. While there are various methods to achieve this effect, such as real-time screen color capture and conversion in Unity or hand-drawn animation, I opted for post-production in After Effects for greater flexibility and ease of adjustment. However, given sufficient time, a hand-drawn approach would have been ideal for expressing the desired visual style.

© Pi Xiu

By rendering individual layers with a Z-depth channel in EXR format, we facilitated a highly flexible post-production workflow. This enabled precise control over depth of field, motion blur, and lighting adjustments. The Magic Bullet Suite was instrumental in achieving the desired aesthetic, providing a comprehensive suite of tools for color grading, effects, and image restoration. The plugin's extensive library of pre-sets and customizable parameters significantly accelerated the post-production process.

© Pi Xiu

It took us two weeks to bring this project to life, and we're thrilled with the results! Seeing the amazing work of our fellow competitors inspired us to push our boundaries even further. We've identified areas where we can improve and have a clear plan for leveling up our skills. Next year, we're excited to dive into Blender or Unreal Engine 5 and create even more awesome stuff!

New Pwnisher's CG Challenge: Kinetic Rush Challenge

This project was a thrilling journey that pushed our creative and technical limits. Speaking of pushing boundaries, have you checked out Pwnisher's 'Kinetic Rush'? It's the internet's biggest and most exciting 3D challenge yet! As the leading render farm in the CG industry, Fox Renderfarm is proud to be a sponsor, offering creators fast and reliable cloud rendering support to bring their wildest imaginations to life. Don't miss your chance to shine!

Conclusion

Mastering the art of 3D creation is a journey that demands dedication and innovation. By breaking down the process step-by-step and leveraging powerful tools like the render farm, you can elevate your projects to new heights. Remember, every masterpiece starts with a single frame. Fox Renderfarm is here to empower your creative vision and accelerate your workflow. So, are you ready to make your mark on the upcoming Pwnisher's 3D challenge, 'Kinetic Rush'?

Source: Art Center of Thunder Fire