From Stone Block to Masterpiece: Replicating Sculpture in 3D Interactive Narratives

1. Origin: Making the “Subtraction of Sculpting” a Perceivable Interaction

Many have seen Michelangelo’s David in museums, but few truly understand how it emerged from a block of stone. Static images only present the final result, failing to convey the most critical characteristic of the sculpting process: subtraction. Unlike painting, where layers can be endlessly covered, or 3D modeling, where geometry can be freely added or deleted, sculpting is a continuous subtraction from a solid block. Once chiseled away, the stone cannot be restored.

Therefore, if we want to digitally restore this work, the core is not just displaying a high-fidelity model, but letting the audience experience the “irreversibility of marble” through interaction. This leads to a more general technical question: How can we use interactive means to express a physically irreversible process?

The practice discussed in this article revolves around this exact goal: utilizing public 3D scan data to restore the process of a block of stone gradually revealing a human figure. This is not a product review or a tool promotion, but an implementation breakdown aimed at the intersection of culture and technology.

2. Data Prerequisites: Credibility Brought by Real Scans

To convince the public that “this is a piece of marble being chiseled,” the fidelity of the model itself is crucial. If stylized or low-precision models are used, it is difficult for the audience to perceive it as a real sculpture. Therefore, in practice, we prioritize publicly available real 3D scan results. Such data usually comes from museums, cultural institutions, or open scanning projects, released as standard mesh files with relatively permissive licenses.

In actual processing, common 3D mesh files can be directly imported into a frontend graphics environment, but several issues arise:

1. Large File Size: High-precision sculpture scans may contain hundreds of thousands or even millions of triangles, putting pressure on browser loading and rendering capabilities.
2. Uneven Topology: Scan data often contains noise, holes, and irregular triangle distributions, making it unsuitable for direct procedural reduction.
3. Inconsistent Scale: The original mesh might retain the true scale of the scan or might have been scaled by the export tool, requiring calibration to a viewport-friendly size.

A robust processing approach involves:

• Performing mesh simplification first to strike a balance between visual precision and performance.
• Sealing the mesh to ensure that internal voids are not exposed during cutting or disappearing effects.
• Recalculating normals and unifying their orientations to avoid lighting and rendering anomalies.

Once these basic processing steps are complete, the model can serve as the base asset for the subsequent “layered reveal.”

3. Core Interaction Modes: Making “Stone Peeling” Credible

To help the audience understand the sculpting process, simple text explanations are far from enough. A more effective approach is to let the user trigger the “chiseling” action themselves.

1. Press-and-Hold Chiseling

One approach is to place a seemingly intact block of stone in the scene. When the user holds down a button or continuously clicks, the surface peels off layer by layer. This process can be combined with particle effects, falling debris, and slight camera shake to enhance the perception of “stone being crushed.”

Rhythm control is crucial here. Peeling is not a one-time disappearance, but a gradual progression. Programmatically, the model can be divided into multiple levels based on spatial height or sculpting stages, where each user trigger only removes one layer. This maintains the sense of operation while avoiding information overload.

2. Slider Timeline

Another approach is more like “browsing the sculpting process.” The user drags a slider, and as it moves from left to right, the stone gradually reduces from its initial block to the final figure. This method is better suited for exhibitions or reading-oriented pages, as it lowers the barrier to operation and allows the audience to focus on observing morphological changes.

In implementation, the slider approach is similar to adjusting a scalar parameter that controls the boundary between “retained” and “removed” areas. A feasible concept is:

• Placing the model within an initial bounding volume.
• Calculating the retention area based on the slider’s value.
• Visually hiding or geometrically culling the parts outside this area.

The difficulty here is that a simple cut is not enough. In real sculpting, stone is not sliced off in one clean cut, but rather shows fracturing, chipping, and irregular transitions. Therefore, procedural methods need to introduce noise to create a random, shattered boundary that mimics the effect of hammering.

3. The “Counter-Intuitive” Design of the Undo Button

There is a detail worth noting in interaction design: the system can provide an “Undo” button, but when the user clicks it, it does not actually undo the sculpting action. Instead, it prompts: “Marble has no undo key.”

The significance of this design lies in using a minor sense of frustration to make the audience instantly understand the irreversibility of sculpting. It is not a technical limitation, but an interaction metaphor based on material properties. For developers, this means that interaction design is not always about helping users complete tasks efficiently; it can also be used to convey certain concepts. The visual style of the button can remain consistent with a standard undo button, but the logic actively blocks the rollback, creating a cognitive contrast.

4. Visuals and Atmosphere: Focusing Attention with a Dark Room

If the model is simply placed on a blank background, the audience’s attention can easily be distracted by environmental elements. A more effective approach is to make the scene feel like a dark room: only the stone block is partially illuminated in the frame, while the rest sinks into a dark background.

This visual strategy offers several benefits:

• Highlighting the Subject: Reducing extraneous information keeps the audience focused on the morphological changes of the stone.
• Enhancing Material Perception: Directional lighting better represents the grain and bumps of the marble.
• Creating a Sense of Ritual: Similar to how a single exhibit is spotlit in a museum, it aligns with the psychological expectations of viewing a sculpture.

In implementation, you can build a lighting scheme with low ambient light, a strong key light, and minimal fill/rim light. Additionally, apply a high-roughness material property to the stone surface to prevent it from looking too smooth and plasticky.

5. Technical Approach Breakdown: From Mesh to Frontend Rendering

To implement this type of interaction in a browser, there are usually two mainstream approaches:

1. Shader-Based Visual Reduction

Instead of directly modifying the mesh geometry, a fragment shader is used to determine whether the current fragment is within the “retention area.” If not, the fragment is discarded or rendered as peeling debris. The advantage of this method is better performance, making it suitable for high-poly models. The downside is that it requires fairly intricate shader logic, especially when handling the irregularity of the edges.

2. Geometry-Based Progressive Reveal

The model is pre-segmented into multiple stages, where each stage corresponds to an independent mesh or different visible/invisible regions of the same mesh. During user interaction, these stages are switched or interpolated. The implementation logic of this approach is more intuitive, and the visual effect is closer to “gradual sculpting,” but the upfront asset preparation is heavier.

Regardless of which approach is chosen, performance budgets must be determined early in development. High-precision models combined with complex particles, multiple lights, and real-time interaction can easily cause frame drops on average mobile devices. Therefore, it is recommended to conduct performance tests at key interaction nodes, dynamically reducing particle counts or swapping to simplified models if necessary.

6. Cross-Format Delivery: From Interactive Web to Playable Video

Not all users will open an interactive page immediately. For social media or other content channels, a playable vertical video is often easier to distribute. Therefore, after completing the interactive version, it can be recorded as video footage.

During the recording process, note the following:

• Maintain Interaction Rhythm: Do not just record auto-play. Simulate user operations, such as brief pauses, holding down the chisel, and clicking undo to trigger the prompt.
• Unify Aspect Ratio: If the target is mobile distribution, set the canvas to an appropriate vertical ratio during the development stage rather than cropping it post-hoc.
• Fix Obvious Issues: Check frame by frame for model flickering, particle clipping, sudden lighting changes, or text occlusion.
• Keep Editable Projects: The video is just one output. The editable project and source asset records are the foundation for future optimization.

From a practical standpoint, the interactive version and the video version do not conflict. The former is for deep engagement, while the latter is for lightweight distribution. Both share the same visual assets and narrative logic.

7. General Takeaways: Translating Complex Cultural Content into Interactive Experiences

The biggest takeaway from this practice is not that “a certain type of model is powerful,” but rather: complex cultural content can be broken down into a data layer, an interaction layer, and a narrative layer, which are then implemented step by step.

• Data Layer: Find credible, usable 3D scan assets and complete cleaning and simplification.
• Interaction Layer: Design operation methods that align with material logic, allowing the audience to understand the process through their actions.
• Narrative Layer: Use lighting, rhythm, text prompts, and counter-intuitive design to translate technical interaction into feeling.

When these three parts work well together, technology is no longer just a tool to display models, but a medium that helps the audience understand “sculpting” as a human endeavor. For developers, this is a highly reusable concept: in non-purely-technical fields such as art, history, and science, the primary task of interaction design is to make abstract concepts tangible, operable, and perceptible.