Tag

Faces

8posts

LabMay 15, 202617 min

Turning a face into a chibi: a tour of LAM, Gaussian Splatting, and what happens when you push on the mesh underneath

A few weeks ago the only thing we wanted from a "talking-head avatar" was for it to look like the person in the input photo and to blink at roughly the right moments. By this week we wanted to grab th

Faces Gaussian Splatting

bs-loss classifier-fooling collage: v1h satisfies the critic numerically while the rendered face barely moves and identity drifts. The visible-vs-numeric gap that arc anchoring closes.

LabMay 3, 202611 min

Why the squint slider needed an ArcFace anchor — the v1c → v1k failure ladder

Faces Fluxspace

v7 step 1200: nine prompts × eight strengths. Glasses appear from s≈0.75, lock by s≈1.5, identity holds across most cells.

LabApr 27, 202615 min

From FluxSpace edits to a working glasses slider — what five days taught us

Faces Fluxspace

Worked demo: a smile composed on an elderly man via iterative FluxSpace attention editing with age-drift correction — the face stays the face.

LabApr 24, 202610 min

Composing a smile without losing the face it belongs to

Faces Fluxspace

Ours vs FluxSpace-coarse: single portrait, age direction, λ ∈ {−3, ..., +3}. Top row (Ours): monotone aging, identity preserved. Bottom row (FluxSpace-coarse): collapses to solid colour at λ=−3, flips ethnicity at λ=−1, noise at λ=+3.

LabApr 21, 202613 min

Demographic-PC Extraction, End to End: Ridge vs Prompt-Pair, Two Kinds of Cliff

Date: 2026-04-20 Series: Follows Perception Before Training. That post framed a six-level perception curriculum gated on a Level-0 engineering task: build a demographic subspace in Flux conditioning s

Demographics Faces

Large grid of portraits with three red-outlined anchor faces — illustrating anchor selection and odd-one-out discrimination in a perception curriculum.

LabApr 20, 202615 min

Perception Before Training — A Curriculum Interlude Between Parts 2 and 3

Date: 2026-04-20 Series: Sits between Part 2: The Math of Pattern Preservation and whatever Part 3 turns out to be. Part 2 argued for a training pipeline that preserves patterns from text through face

Faces Perception

Seven-step age progression of a single portrait from young adult to elderly — identity preserved while the age axis is traversed.

LabApr 19, 202655 min

Faces as a Data-Mining Puzzle — Part 2: The Math of Pattern Preservation

Date: 2026-04-16 Series: Part 2 of 2. Part 1 here covered positioning — why faces, why a game, why co-discovery. This post covers the math: how we know when a built pipeline actually preserves the pat

Faces Interpretability

Grid of diverse portrait faces at the same scale and lighting — the raw material that makes faces work as a data-mining channel.

LabApr 19, 202631 min

Faces as a Data-Mining Puzzle — Part 1: The Case for an Unlikely Interface

Date: 2026-04-16 Series: Part 1 of 2 — positioning. Part 2 covers the math of pattern preservation. Audience: technical readers who know embeddings, neural networks as black boxes, and basic ML, but n

Faces Perception