SHP LabScience, Humanity & Philosophy LabQuestions become research, projects and public knowledge.All Projects ↗
SHP PROJECT CASE · 02CULTURE & MEMORY

Porcelain Database

When restoration becomes an act of interpretation

Porcelain Database asks what an intelligent system should do with a fragment. Combining object history, image classification and restoration design, the project shifts attention from producing one seamless answer to making evidence, uncertainty and alternative reconstructions visible.

Jinggong Niu · 2024–25Research + PrototypeAI / Heritage / Restoration
SELECTED MATERIAL · SHP EDIT
02

The Question Behind the Work

A fragment does not contain a single complete past.

A curve, a glaze and a partial motif can point toward a period or production tradition, but none can recover a missing object without interpretation. Every act of restoration therefore moves between material evidence and informed speculation.

The project asks how AI might support that movement without concealing it. The problem is not only whether a machine can complete an image, but whose categories, archives and historical assumptions shape the completion.

evidence / inferencerepair / reconstructionautomation / judgment
03

Following the Research

Reading history through form, pattern and absence

The research treats porcelain as a historical record whose shapes, motifs, materials and trade routes must be read together before any reconstruction can be proposed.

01

Form as Evidence

Silhouettes, rims, bases and glazes provide clues, but their meaning changes across workshops, periods and routes of exchange.

Historical typologies of form and decoration
02

The Politics of Classification

A database does not simply store history. Its fields determine which similarities become searchable and which differences disappear.

Building a comparative visual database
03

More Than One Reconstruction

Keeping several hypotheses visible prevents probability from being mistaken for certainty and invites the researcher back into the decision.

Alternative readings of incomplete evidence
04

From an Idea to a Form

What should an intelligent restoration system refuse to hide?

The proposed workflow separates scanning, comparison, visualization and human judgment. Rather than selecting one finished answer, it shows the evidence supporting each possible reconstruction.

A robotic restoration concept then connects digital inference to physical repair, while keeping the human researcher responsible for interpretation. The interface is designed to expose its reasoning rather than perform certainty.

A responsible reconstruction should make its uncertainty visible.
PROCESS · 01
EXPERIMENT · 02
05

The Work

Porcelain Database, 2024–25Image database, restoration workflow and prototype system.
06

Why It Matters

Why this project matters

Porcelain Database matters because it reframes AI restoration as an ethics of interpretation. A useful system does not erase the break, the missing evidence or the disagreement. It gives researchers and publics a clearer view of how a past is being reconstructed.