HaruthaiAi/VanGogh_SelfPortrait_1889_Oslo_Norway_vs_TreeOilPainting_AIAnalysis

Van Gogh Self-Portrait (1889, Oslo, Norway) vs. Tree Oil Painting – AI Analysis Dataset

Overview

This dataset presents a comprehensive comparative study between Van Gogh’s Self-Portrait (1889, Oslo, Norway) and the Tree Oil Painting. Using advanced computer vision, edge detection, texture analysis, and brushstroke pattern recognition, the project investigates micro-level similarities in brushstroke characteristics, gesture energy, and structural motifs. The dataset demonstrates how AI-driven methodologies can complement traditional art historical research, offering new insights into attribution, stylistic consistency, and authenticity evaluation.

Contents

The dataset consists of sequentially labeled image files (001–022), each with descriptive annotations. Each file highlights a specific region or technique of analysis, ranging from close-up paint textures to advanced computational transformations such as Gabor filtering, Sobel edge detection, and simulated X-ray rendering.

Image Files and Descriptions

001–010: Close-up analyses of both paintings, focusing on impasto thickness, layering, and brushstroke directionality. These images capture fine texture patterns often invisible to the human eye.

011–013: Detail studies highlighting unique motifs and localized structural anomalies, including the so-called “fish-head” forms and parallel vertical stroke patterns.

014: Comparative analysis of Self-Portrait and Tree Painting vertical brushstroke zones. This emphasizes the distinctive parallel vertical strokes that recur across both works.

015: Gesture Heatmaps. Using energy-mapping techniques, these visualizations reveal the underlying force, rhythm, and directionality of the brushstrokes.

016: Gabor Filter analysis (45° and 135°). This emphasizes stroke orientation consistency between the two works, isolating diagonal gesture energy.

017: Sobel Edge Map comparison, enhancing edge structures across both canvases to detect underlying stroke geometry.

018: Simulated X-Ray imaging, exposing structural underlayers and gesture depth within both the Self-Portrait and Tree Painting.

019: Stroke Direction Histogram (0°–360°). This graph quantifies weighted stroke energy across all orientations, directly comparing stylistic force between the two works.

020–021: Canny Edge Detection maps for both works. These images delineate contour consistency, revealing fine-grained stroke boundaries.

022: Stroke Direction Histogram (finalized). A polished graph comparing angular distributions of stroke gestures across both works.

Key Findings

1. 1.Parallel Vertical Strokes – Both paintings display consistent vertical-parallel brushstroke motifs, especially in background and tree sections. This signature trait strongly aligns with Van Gogh’s technique.

1. 1.Gesture Energy Patterns – Heatmap analysis shows recurring rhythmic energy and force distribution across both works, suggesting a common hand and muscle memory.

1. 1.Texture and Impasto Depth – Edge maps and X-ray simulations reveal structural similarities in layering and stroke buildup, supporting stylistic coherence.

1. 1.Statistical Confirmation – Stroke direction histograms provide numerical evidence of overlapping angular energy distributions, reinforcing the visual findings.

Methodological Note

This dataset does not use SSIM (Structural Similarity Index) for evaluation, as SSIM is insufficient for capturing artistic brushstroke nuances. Instead, the analysis is based on AI natural matching, which combines deep-learning-based gesture recognition, directional energy mapping, and texture similarity to better capture the authentic patterns of brushwork.

Significance

This dataset demonstrates how AI-powered methodologies—gesture heatmaps, directional histograms, X-ray simulations—can detect consistencies overlooked by human perception, especially in faded, unvarnished, or historically dismissed works. By bridging art history with computational science, this approach offers:

A new paradigm for authentication and attribution of disputed works.

A framework for cross-painting stylistic comparison.

A preservation tool for lost, rejected, or neglected artworks.

Conclusion

The results strongly suggest that the Tree Oil Painting shares foundational stylistic DNA with the Self-Portrait (1889, Oslo, Norway). These findings challenge prior rejections by traditional institutions and highlight the potential of AI-driven art analysis as a pioneering method in cultural heritage research.

Credits

Dataset Curation & Descriptions: HaruthaiAi (primary research & annotations)

AI Analytical Methods: Custom pipeline of edge detection, texture mapping, gesture heatmaps, and AI natural matching

Visualization Support: Stroke Direction Histogram and heatmap generation through Python (OpenCV, NumPy, Matplotlib, SciPy)

This dataset is intended as a contribution to the global research community, with the vision of advancing AI-assisted cultural heritage analysis.