Last updated: July 30, 2025
This file outlines the planned direction of the Visual Thinking Lens (VTL) project.
VTL is stable as a research framework but continues to evolve in how it is documented, demonstrated, and shared.
This repository includes:
- Conceptual briefs and whitepapers explaining core theory
- Watermarked case studies showing recursive critique in action
- Legal and usage restrictions to protect interpretive artifacts
The core Sketcher Lens engine and internal scoring logic are not released.
These items are under consideration for future publication:
-
Prompt Collapse Suite (Observation Map)
Visual diagrams tracing symbolic degradation and recursive loop behavior. -
RIDP Engine Case Studies
Reverse Image Decomposition Protocol walkthroughs showing collapse → repair cycles. -
Expanded Figure Stability Study
More long-run constraint tests beyond the "Thirty Figures" set. -
Validator Crosswalk (High-Level Overview)
Conceptual map of validators (Prompt Pressure, Compositional Predictability, etc.) without revealing implementation details. -
Marrowline Protocol Primer
Introductory document on symbolic rupture and refusal logic.
- Raw prompt strings, model-specific recipes, or scoring formulas
- Executable code for the Sketcher Lens or related engines
- Unwatermarked images or assets for external reuse
- Consequence over exposure: Publish what demonstrates recursive visual intelligence, not what replicates the engine.
- Research utility first: Focus on clarity for LLM alignment researchers and artists.
- Protective publishing: Maintain strict watermarking and licensing to prevent misuse.
Questions or research collaboration requests: