Open code, practice, and notes →
Download source with English comments · for reading
Gaussian Splatting · 19 Reducing Gaussian Count
Series contents · Technology map
Stage: Visibility and Quality Experiments · Editable example · Storage ID: gaussian-splatting-step-19
Goals and learning outcomes
- Create a reproducible subset.
- Evaluate the visual impact of reduction strategies.
Prerequisites
Continue using the output and records from Unit 18. Before the Reducing Gaussian Count experiment, preserve the previous inputs and reference outputs.
Key concepts and explanations
Subset selection
Reducing item count lowers storage and sorting work. Random removal may disproportionately damage thin structures or low-opacity regions.
Reproducibility
Using different random values each time makes reduction strategies hard to compare. Save the seed and selected ids to use the same subset.
Practice setup
First check the execution scope guidance on the example screen. The following describes the environment needed to extend the assignment in a separate project. This is a document-guided exercise in a separate local project. Use only small data you create or local files whose usage permission you have checked. The existing preview is a CPU teaching viewer for .splat/ASCII GS PLY files of at most 1 MiB and 512 items; it is not a Spark GPU renderer.
The linked example is a starting point. It does not implement every task below. Complete work beyond its supported scope in a separate project.
Practice tasks
- In subset.js, implement seeded selection and selection ordered by opacity.
- From a synthetic fixture of 256 items, select 128 and save the id list.
- Compare empty regions, sorting time, and file bytes at the same three viewpoints.
Variables to observe
| Variable | Baseline | Change | Expected change / hypothesis |
|---|---|---|---|
| Retained count | 256 | 128 | Expect less sorting work but more holes in some regions. |
Predictions in this table are not measured results. Change one condition at a time and record why the actual result differs.
Completion criteria
- Using the same seed twice produces identical id lists.
- Compare images at three viewpoints for both selection strategies.
Self-check
- Are low-opacity items always less important?
- How does pruning during training differ from simply reducing a file?
Official references
Linked example and execution scope
Example source · Runtime environments and limits
CPU Canvas2D synthetic fixtures and contract checks. This does not provide Spark GPU rendering, training data, or claims about reconstruction quality or GPU performance. The provided example keeps the first N entries using data.slice(0,count). Seed-based selection and opacity selection policies are separate assignments; screen-contribution pruning is not implemented.
Source code and records are not overwritten automatically. Resetting an existing example also preserves your notes.
Next learning
Previous: Resolution and Pixel Workload · Next: Color Space and Background