3 Months to Compete. 3 Weeks to Grind. 

Life doesn’t check your calendar

Life happens: circumstances outside our control, the few things we can control, and deadlines that can't be bargained with.  What was initially a strict and well laid out pipeline turned into 3 weeks of sleepless nights caused by events that shaped the way I was able to direct a film that pushed me to my creative limits. 

Color scripts got cut in order to solve complex problems involving weight painting and vertex groups exploding. I switched from Unreal to EEVEE in Blender 5.2. Storyboards with 13 frames in mind were slashed into 6. All of it—triage. I had to decide as I worked, what I’d be keeping, removing, redoing and ways to make the short film as painless as possible. But 14 hour work shifts made that very difficult and, ironically, rewarding.

Deadlines don't care what's happening behind the scenes. Instead of fighting against setbacks, I decided to say, “Okay, this is happening. How can I make something I'm proud of within these time constraints? Let's see what can be done.”

Who Lycoris is

Lycoris began as a character I designed in Fantasy Life i: The Girl Who Steals Time by Level-5. I’ve always had a deep fondness for characters that are misunderstood. Hanyō (半妖) or half-demon, are often the bridge between the supernatural and the mundane. Where life shouldn’t exist, there is the hanyō. A half-demon priestess with purification powers? It’s a contradiction that makes Lycoris so appealing—even her powers don’t match with the expectation of who, or what, someone is. 

A bow that regenerates arrows from how much spiritual energy she has? Talismans that purify demons? Spider lilies that often announce her presence and place between life and death, Lycoris makes the viewer ask: is darkness inherently evil? Or can purity exist in obscurity? 

Pre-production

I started off with a mood board: priestesses wearing Miko clothing (巫女装束 , miko shōzoku),Kagura suzu (神楽鈴 ) bells that announce her presence, fangs, incense, solemn expressions attached to crimson eyes. Lycoris has a very specific aesthetic that ties both to her sacred and underworld nature. 

Then came the sketches, storyboarding, color scripting, which I later had to cut—all done in Clip Studio Paint. All of these took a lot of time, but made creating her in Character Creator 5 easier than creating her from imagination alone. 

Building Lycoris

The base started in Character Creator 5. Everything from the eye color which was edited in Affinity, to her makeup, hair, nails, and eventually, her clothing being successfully attached through Reallusion’s skin transfer weights. This made clothing physics both a breeze and a nightmare. I’ll explain why later.   

The bow was modeled and rigged in Blender. Her outfit was custom made in Marvelous Designer. Small steps with complex logic were required to create a dynamic character ready for animating. Victor Soupday’s CC/iC Blender Tools  made transferring her back and forth between CC5 and Blender incredibly easy. His hard work is what allowed me to enter this competition with confidence that her rig and base mesh wouldn’t break.

From Unreal to EEVEE Next

The toughest decision was making the switch from Unreal Engine to EEVEE. It was last minute and forced me to pivot how the rest of the short film would be produced. I had built the entire film around Unreal, so losing that pipeline meant restructuring the project from the inside out.

Time was running out, and in Unreal, clothing physics had to be built through a node-based setup. Blender let me bake the simulation directly, and with the deadline closing in, that difference decided it. 

It was like plumbing a house with copper, only to realize midway through that you had to switch to PEX. Both work beautifully, but nothing connects the same way.  

Since Niagara only exists in Unreal, the spider lily bloom animation (a vertex animation texture, or VAT, which bakes a mesh's animation into an image Unreal can play back) didn't survive the move. The irony? A scale bug had already cost me time on it. The spider lily model was importing as an FBX at a scale of 0.004, which broke the shape keys I was using for the blooming effect with OpenVAT. I eventually created a new file, reset the scale to 1, applied it, and rebuilt the shape keys from there, only for the whole effect to be cut.

The blood moon was another snag—and worth every second I spent on it. Rather than build another complex shader from scratch, I reverse-engineered one I had already made for Lycoris's talisman. The original material creates a burning-from-the-inside-out effect. By adjusting several nodes and recalibrating the math driving them, I transformed the transparent, fire-eaten effect into the blood-red corruption that consumes the moon after Lycoris's arrow pierces it: her declaration of war. 

Finishing in Blender 5.2

With the switch made, Blender became home base for everything that was left: rigging, animation, physics, and the final render. It's also where I spent the most miserable stretch of the entire project.

The cloth simulation kept exploding. Her sleeves would tear apart, stretch into strange shapes, or snap in directions that made no sense, and I couldn't animate her at all until the clothing behaved. In weight paint mode, everything looked fine. The sleeves were solid blue, which I assumed meant they were set up correctly.

They weren't. After two days of barely sleeping and clicking through every button I could find, I discovered Blender's Zero Weights overlay, which shows vertices that aren't assigned to any group at all. In normal weight paint mode, a vertex with a weight of zero and a vertex that isn't in the group at all both look exactly the same: blue. Some of the sleeve vertices had no weights, so when her arms moved, they stayed frozen in place, and the simulation tore itself apart trying to reconcile the difference. Once I could see the problem, I could finally fix it. If there's one thing I want other Character Creator and Blender users to take from this post, it's this: blue doesn't always mean what you think it means.

For rigging, Victor Soupday's CC/iC Blender Tools did the heavy lifting again. It let me rig her with Rigify while keeping her Character Creator setup intact, which gave me a full control rig without rebuilding her from scratch.

For animation, I pulled FBX animations from Mixamo and retargeted them onto her rig with Auto-Rig Pro, using a custom bone map to tell it which bones on the Mixamo skeleton matched hers. Retargeting turned out to be so much fun that I now know remapping like the back of my hand.

One tip if you're using Auto-Rig Pro and your fingers aren't behaving: when you copy the rest pose from the source armature onto the target, select the hands first, Apply. then the first bone of each finger. Apply again. For the thumbs, select the entire chain. Apply then apply the rest pose. Adjust your IK-FK settings in ARP, and your fingers should work beautifully.

Finally, the particles. Those were easy. Tuning the wind was a different story. Getting real creative control over both her sleeves and the particle effects took a lot of adjusting of field weights and wind strength, and at one point I had three wind fields in a single scene: one driving the particle simulation, and one for each sleeve.

By the end, the pipeline looked nothing like what I'd planned three months earlier. But she moved, her sleeves flowed, and nothing exploded. After everything, that felt like a victory.

Post-Production

Most of the color work happened before the footage ever reached DaVinci Resolve. I graded inside EEVEE, guided by the color script I'd started in pre-production. It never got finished, but even half-done, it gave me a direction to follow. In Resolve, I mainly added bloom: a soft glow around the moon and around her face, so the eye always lands on the two things that matter most.

Sound is where the film really came to life. Before the first frame fades in, you hear kagura bells, announcing Lycoris's presence before you ever see her. The footsteps of the soldiers hunting her come from a crowd-running track layered with reverb, so they feel distant and close at the same time. I wanted the film to feel full and lived-in, as if the world kept going just outside the frame. That's also why the soldiers are the only part of the film with motion blur. Everything else stays crisp, so their chaos stands apart from her stillness.

The score is "Silent Arrow," a somber, contemplative piece that's still rich and ornamental. I cut and rearranged it so one of its swells lands exactly as the moon blooms red, so her declaration of war hits right on the beat.

What I'd Do Differently

I don't believe this project struggled from poor planning or procrastination. Three months was enough time to make the film I envisioned, and I still believe that. But life had other plans, and when three months became three weeks, I learned the most important lesson of the whole project: be ruthless with your scope before the deadline forces you to be.

If I did it again, I'd decide what was essential sooner. I'd commit to one engine early, test cloth physics before building anything else around them, and cut features while cutting them was still a choice instead of an emergency. I'd also protect my sleep. Pushing through with almost none got the film finished, but it isn't something I'd recommend or repeat.

What I'm Proudest Of

The finished film isn't everything I wanted it to be. But I'm proud that I entered anyway.

It would have been easy to quietly let the deadline pass. I could have told myself the work wasn't ready, that I'd do it properly next time, and stayed invisible. Instead, I treated it as a professional project from start to finish, made the hard calls, and put my name on it publicly, even though being seen is one of the hardest things for me.

Lycoris got finished. She's out in the world. And next time, she'll have the three months she deserves.