blog.init | Welcome to My Blog

Hello

The first version of this page said only initial blog post pending. That was accurate, but not especially useful. So here is a better beginning.

I’m Danzel Serrano. A current CS doctoral candidate at the New Jersey Institute of Technology (NJIT). My research sits at the boundary between audio and geometry - speech-driven 3D facial animation, and specifically how to make its controls something an animator can understand and steer, rather than a wall of opaque knobs.

Why this blog exists

A portfolio tends to show conclusions. I want this space to preserve more of the path. Writing forces me to name assumptions, separate evidence from intuition, and notice when a tidy story does not match the system in front of me. Publishing the notes adds a useful constraint: another person should be able to follow the argument without having lived inside the project.

The original placeholder carried one mathematical object with it, an acoustic wave equation:

$$\frac{\partial^{2} p}{\partial t^{2}} = c_0^{2}\,\nabla^{2} p + \frac{1}{\rho_0}\!\left(\tfrac{4}{3}\mu + \mu_B + \kappa\Big(\tfrac{1}{c_V}-\tfrac{1}{c_P}\Big)\right) \frac{\partial}{\partial t}\big(\nabla^{2} p\big)$$

It was not meant as a manifesto. I simply liked having an equation waiting here before there were words around it. I am keeping it as a small fossil from the scaffolding-version of this blog (and reminding myself to write posts related to acoustic simulation)

My Projects

Current Research

My doctoral work is speech-driven 3D facial animation (SD3DFA): turn an audio track into a moving face, but through a latent space an animator can actually operate. The problem I care about is disentangling what a face is saying (visemes, and the way neighboring sounds bleed into one another as coarticulation) and how it is expressing (visual prosody; emphasis, mood, the lift of a brow), which are usually tangled together, which evidently results in realistic motion but the representation is uninterpretable or explicitly editable. I'm after representations where those axes come apart, each steerable on its own along continuous controls rather than a fixed menu of poses. It lives at the seam between audio and geometry—sound shaping shape, and shape implying sound—which is the seam I keep returning to.

hippo-campus.io

In the age of agentic coders, the frontier of new discovery is expanding quite quickly.

hippo-campus.io is a push against this speedy discovery of new information by utilizing agentic coders to make flash cards decks for the human to memorize and study to stay in the loop of their own fields of interests.

It is a bring-your-own-repository (BYOR) spaced-repetition environment for Markdown decks.

Intro post coming soon (TBD). If you wish to try it out as it is: hippo-campus.io

art.

comtaken

A WebGL shelf holding hundreds of thousands of dictionary words as tiny books. It began as a playful way to browse language and domain availability, and became an exercise in rendering, search, caching, and making a large digital object feel surprisingly physical.

In conclusion: a small rendering test

This new blog interface can carry live figures rather than only screenshots. For a first 3D test, here is the Stanford Bunny: drag it, zoom in, pause the rotation, or switch to its wireframe.

{"id":"stanford-bunny","autoRotate":true}

This is the full reconstruction from the Stanford Computer Graphics Laboratory’s 3D Scanning Repository.¹ It's relatively small enough to load quickly to test the interactive modules.

What comes next

Expect posts about sound and 3D motion, the design and implementation of the projects above, experiments with geometry and interactive graphics, and whatever else becomes clearer by being written down. I want this to feel less like a publication schedule and more like a durable working notebook: quiet, useful, and occasionally strange

For now, consider blog.init properly initialized.

Footnotes

1. Stanford University Computer Graphics Laboratory, The Stanford 3D Scanning Repository. ↩