Open code, practice, and notes →
Download source with English comments · for reading
GLSL · 27 Quality Criteria and Regression Records
Series contents · Technology map
Stage: Interaction, Postprocessing, and Cost Evaluation · Editable example · Storage ID: glsl-step-27
Goals and learning outcomes
- Create reproducible visual references.
- Record local quality losses.
Prerequisites
Continue using the output and records from Unit 26. Before experimenting with Quality Criteria and Regression Records, preserve the previous inputs and reference outputs.
Key concepts and explanations
Fixed Inputs
Fixing time and pointer inputs makes image comparisons reproducible. Different rendering environments can produce pixel differences, so record tolerances.
Regions of Interest
A whole-image average difference can hide lost small highlights or thin boundaries. Compare both the full screen and selected regions.
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 document-guided exercise requires your own implementation using WebGL2 and GLSL ES 3.00 in a separate localhost project. First check the GLSL version of existing examples and avoid mixing in 1.00 syntax.
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
- Save time, pointer, resolution, and effect variables in presets.json.
- Capture reference and half-resolution images for three presets.
- Record boundary, blur, and highlight ROI differences and the environment in report.md.
Variables to observe
| Variable | Baseline | Change | Expected change / hypothesis |
|---|---|---|---|
| Quality setting | full resolution and 4 octaves | half resolution and the same octave count | Expect lower pixel cost but greater loss of boundaries and high-frequency detail. |
Predictions in this table are not measured results. Change one condition at a time and record why the actual result differs.
Completion criteria
- Save input values and environment details for all three presets.
- Describe losses in three ROIs using numbers or pixel locations.
Self-check
- Should every pixel difference on another GPU count as a regression?
- When should reference images be updated?
Official references
Linked example and execution scope
Example source · Runtime environments and limits
Runs WebGL ShaderMaterial in the browser. Comparisons use 640×360/320×180 sizes; frame intervals are different from GPU time. The provided selections change both full/half resolution and the noise interpolation expression while retaining 4 octaves. This is not a controlled comparison of resolution alone. Comparing with fixed interpolation is a separate assignment.
Source code and records are not overwritten automatically. Resetting an existing example also preserves your notes.
Next learning
Previous: Editing Failures and Context Recovery · Next: Interaction and Cost