It’s been busy couple months. I’ve been hard at work animating the film. Back in May, I decided to start with the most challenging scene in the film—the big kung fu fight!
I started with this one in part because we’re still finalizing the dialogue recordings with the cast, and there’s minimal dialogue in this scene. However, I also knew this was going to be the most difficult scene to animate in the whole film. Up to eight characters on screen at once, tricky lighting, and over forty shots for less than a minute of footage. I knew it was going to be a technical challenge, but also a ton of fun to animate. I estimated it would take about a month to complete.
Dear reader, it took two.
The animation itself was pretty straightforward. It took a couple of weeks to get all the initial keyframes done for all the shots, but then I needed to work out the rendering. That’s where things initially started to bog down. While render times weren’t a problem (even the most complicated renders only took 20-30 minutes for an entire shot layer), I quickly discovered the the lighting situation for the scene was more complicated than my character rigs could handle easily.
However, before diving in on fixing all that, I decided to finish out the animation by going from the “keyframing” phase where basic poses and timing is worked out, to “tweens” where the in-between frames are added to smooth out the motion. You can see the difference between the two here for this shot:
KEYFRAMES:
TWEENS:
Thankfully going from keyframes to tweening is pretty straightforward and really fun. In the case of the film, I’m animating on “2s” which is a fancy way of saying I’m animating at 12 frames per second (fps). The film itself is 24 fps, but I’ve found that limiting the characters to half that really helps “sell” the 2D animated look.
If you’re not familiar with how 3D animation works, there’s two main parts to the process:
ONE: ANIMATION — Pose the characters, props, and move the camera in 3D space to set up the movement:
TWO: RENDERING — Set up the lighting and render out the final artwork with final colors and lighting, and in this case, ink outlines.
While in some cases you can get by with rendering out a single image with the characters and background completely finished, in practice this is rarely how it works in real life. In the case of PS Chinatown, I’m rendering out a minimum of two layers: one for characters, one for backgrounds, but for the more complicated shots, I’m rendering out literally DOZENS of layers and then assembling them in Adobe After Effects. I even figured out a way to fake a painted look for the backgrounds.
Setting up a shot in Adobe After Effects. This shot took over 80 layers to achieve! 28 of which had to be rendered out of Blender individually.
Once I was happy with how the movement was looking, it was time to turn to rendering, and fix all the problems I had with the first render pass.
After getting an initial render test done for the entire scene in early April, I realized that I needed to make some improvements to the characters to really make it look great. The main culprit was a technique called “rim lighting” when you use a bright light behind a character or object to “pop” it out from the background. The initial character rigs supported this, but the results were a little clunky in the most extreme lighting conditions. In short, I just did not like the results:
Initial render test of the the most complicated shot in the scene. Looks okay, but… the lighting on the characters left something to be desired. And the hair on ponytail guy is awful!
So, I went back to the drawing board and re-thought how to implement a better scheme for creating the lighting on the characters. The answer ended up being incredibly techinical, involving editing the controls for lighting on a limb-by-limb basis (head, torso, arms, and legs), and then creating a separate render layer that can be easily edited in After Effects. Suffice to say it took some work and about a week to implement on all the characters. However, in the end the results were worth it!
Final render of the shot. Background and characters more integrated. However it took implementing an entirely new rim lighting system on the character rigs, which took about a week to figure out. I also had to model new hair for ponytail guy.
However, during the rendering process, I had to completely revamp the work Nina and I had done in After Effects after the first round in late March/early April. And it took a couple of weeks to get all the shots updated and re-rendered. During this time I decided to bite the bullet and purchase a new M4 Max Mac Studio to replace my aging iMac I had been using as my desktop machine because it was starting to choke on the Blender scenes I’d set up (but that’s another blog post entirely.) Thankfully, the new machine has arrived and has sped up the rendering process tremendously (in part because I now easily render in the background while while continuing set up shots either in Blender or After Effects. Heaven!)
I finished rendering and compositing the last shot back in May, and have since passed all the shots along to Nina for cleanup. I’ve been working on action scenes that follow this ever since. I just started in on the gunfight finale and should be finished with that in August. I’ll update y’all on all that later!
Anyway, here’s a few stills from the scene. Enjoy!
Hey, all! It’s been a couple of weeks since the Kickstarter ended, and while I took a little break from updating to recover not only from the campaign, but also a nasty cold and Thanksgiving, pre-production for the film has been humming along in the background.
First, we’re nearly finished with making all the character models! I put the finishing touches on Mike’s friend Kuan and the restaurant Hostess, and then moved on to working on creating background characters that will populate the streets and the restaurant scenes. I also got the first of our “special guests” completed—more on him soon!
Test render of the restaurant’s HostessTest render of Mike’s friend Kuan.The models for our “extras” for the street and restaurant scenes as seen in the Blender viewport, but not yet rendered.
To create all these background characters, I have two “universal rigs”—the same that I used to create the gang members last summer—one male, one female. While there are some generic outfits I can use, I customized these characters with clothes and hair purchased from DAZ3D and exported to Blender. There’s also a few characters that have been repurposed from previous projects and modified to suit this film.
Between all these new characters, plus the original cast, that puts the total at 20 custom characters for the film! If we had to make all these characters from scratch, that could have easily taken over half a year, but using this streamlined process, we’re able to create a bunch of characters in a week or two. Now, I’ll be the first to admit they are not as polished as the work of some of the other Blender pros at work out there, but they will more than do for the purposes of this project, and it was worth the tradeoff for the speed of creation. Perhaps in the next project we’ll be able to make use of some of those pros out there!
In addition to making characters, I’ve also proceeded in setting up the layouts for the film, starting with the restaurant scene. “Layout” refers to process of setting up the files in Blender with all the sets, props, characters, lights, and special effects that will be needed for the scene, and then determining the basic posing of characters and finding camera angles for each of the shots. I also use this process to create test renders from a few key shots to decide on the look of scene, especially the lighting.
Here’s what that looks like:
Layout interface in BlenderSet modelCharacter modelsCombined modelsAnimation interface in BlenderPreview of the shot in the cameraFinal composited test render.
It’s been a few weeks in the making, but we can finally announce we’ve officially signed our voice cast!
First off, I’m excited to announce we’ve landed Dayeanne Hutton to play Kate!
Dayeanne’s a veteran video game voice actor from the Life is Strange series, and starred in the Emmy Award-winning show, Emma Approved. Her range is deep and broad, we’re really happy to have her on board!
Next, we’re delighted to announce that we’ve landed the incredibly talented Zach Valenti to be Mike Stuart’s voice!
For our supporting cast, we’ve landed Peter Hu to be the voice of Kuan, and a member of the White Dragon gang. Peter is also a podcast fiction veteran. And Gloria Tsai will be lending her voice to our Hostess. Both will also provide some additional background voices for film.
Welcome to the team, everyone! We’re so lucky to have you!
Last week I talked about creating the street scenes for the film by modeling some of the key buildings by hand. However, doing that for an entire neighborhood wasn’t going to work.
I still needed to create dozens of other buildings, and if the ones I’ve bought don’t fit the character of the real Chicago Chinatown, what do I do? I have to find a way to quickly create lots of them. The answer came in the form of Blender’s “Geometry Nodes” which with the help of this incredible tutorial series by KammerBild, showed me how to create “procedural buildings”:
I can’t recommend this tutorial enough to you Blender heads out there!
I used this technique to create all the buildings for the skyline in Restless Sleep, and repurposed it for Chinatown using elements from the various models in my asset library.
Basically, the building (left) is created using a visual program (center) and controlled using the properties in the viewer on the lower right. Once the program was set up in Geometry Nodes, I don’t have to mess with it anymore. I then can feed new building models into it to create new buildings of any size and customize them with different sets of doors and windows, based on the model sources.
The models can look like anything, so long as they fit into a cube. In this case, I used pieces of a building model I purchased from DAZ3D by an artist called StoneMason. It felt a little like playing with LEGOs!
This building… …is made from these parts!
Over about a week, I adapted about a dozen building models like this to work with this so I could populate the entire Chinatown neighborhood. Then all I needed to add was some signs—both customized and stock—to give the street a little more character. A number of these signs were made in Photoshop based on some Creative Commons photos of Chinatown from about 15 years ago, which the era I knew the neighborhood from. It’s changed quite a bit since I moved away, but I wanted to preserve the character of the Chicago I once knew and loved.
While these buildings aren’t an exact match to the real ones in Chicago’s Chinatown, they definitely do the job. If I had infinite amounts of time or budget to hire a team of skilled BG artists, maybe we could get closer, but I’m pretty happy with the results I’ve been able to achieve in a short period of time using this method.
One of the most intimidating parts of tackling Paradigm Shift has always been the urban environments. When I first started to draw the comic, it was easily the thing that I was scared of doing most, which is ironic considering that the highly detailed drawings of Chicago neighborhoods became one of the things I’m most known for. The only reason that was possible was I had to take that fear of drawing those shots head on. I drew some test cityscapes, and started with them on page one, and by the time I’d done a few, they weren’t so scary anymore.
Original Chinatown opening shot from 1999.Test render of the new Chinatown street set in 2024Ref photo from 1999Pool hall exterior in the comic.Pool hall test render.
Fast forward 25 years, and the same problem hit when trying to bring Paradigm Shift to animation—these bloody cityscapes! Now, video games have been tackling this problem for decades now—managing the huge amount of assets that will be required to make a believable city street: buildings, cars, people, signs, etc. Modeling every little detail by hand the way I drew every line for the comic is a daunting proposition, and not exactly the best use of my time today.
The Chinatown street set in Blender.
Thankfully, I don’t have to because there are so many talented artists out there in the 3D world already who have made their work available for a very reasonable price. I’ve spent the last couple of year accumulating assets through DAZ3D’s store with an eye towards making Paradigm Shift: Chinatown—buildings, cars, weapons, sets, and various asian-themed environments. Now I can finally put them to good use! However, I’m not deploying them as-is, but rather I use them like LEGOs, like I did with the sets I got as a kid. I disassemble them and use them for parts. In short—I kit bash. This way, I can save my modeling time to create the very specific buildings and props I need to make Chinatown look like the neighborhood I knew back during my time in Chicago. Buildings like the Pui Tak Center:
