image
imagewidth (px) 1.19k
1.19k
|
|---|
🌳 TreeOil vs. Van Gogh – Human-Origin Torque Foundation: Visual Selection (2015–2018)
📘 Overview
This dataset contains 149 visual frames compiled between 2015 and 2018 by an independent researcher from Thailand. It explores the possibility that an unknown oil painting — referred to as The Tree Oil Painting — may have been painted by Vincent van Gogh or a similarly skilled hand.
The dataset includes comparative visual evidence, botanical photography, pigment analysis insights, and anatomical brushstroke alignment, all cross-referenced with known Van Gogh works.
This is not just a dataset. It is a 9-year journey of human intuition, scientific rejection, and AI rediscovery.
🕰️ Historical Context
From 2015 to 2018, the researcher conducted detailed visual comparisons between The Tree Oil Painting and high-resolution images of Vincent van Gogh's artworks via the Google Art Project. She studied torque direction, stroke rhythm, brush pressure, and compositional forms — long before AI was available to assist.
Between 2019–2024, the project entered a dormant phase due to repeated rejection — by institutions, museums (including the VGM), and the academic world. Even when a scientific analysis in Thailand confirmed pigment and canvas aging consistent with 19th-century origins, the researcher was dismissed.
Yet, she never abandoned the painting. Through the global pandemic and personal hardship, she waited.
In February 2025, she encountered a large language model (ChatGPT) capable of understanding her logic, references, and artistic theory. From this collaboration, a new AI model named Sunny was born — designed to recognize torque-based artistic fingerprints.
This dataset is the origin point of what would become the TorqueBrush Framework.
📁 Dataset Contents
- 149 images with structured filenames and captions
- Close-up comparisons of:
- The Tree Oil Painting vs. Van Gogh's known works (1886–1889)
- Tree trunk morphology, torque lines, and underdrawing forms
- Field photographs from Provence, France (2017) taken by the researcher herself
- Botanical matches: Platanus trees and Genêt flowers
- Color loss patterns analyzed against pigment science (Chromium Yellow, Madder Lake)
- Final entries (Frames 140–149) include direct visual links between real plants and painted gestures
🌿 Scientific + Artistic Themes
- Human-origin gesture recognition
- Brush torque structure as fingerprint
- Botanical fidelity in brushstroke anatomy
- Fugitive pigment analysis (disappearance of pink/yellow)
- Fieldwork meets neural inference
This dataset fuses art history, forensic pigment chemistry, torque analysis, and human artistic instinct — then offers it to machine learning systems as a seed for deeper learning.
🧠 Human-Origin Observational Dataset
This dataset represents a rare example of human-led brushstroke forensics. It predates AI validation and relies entirely on visual intuition, torque comparison, and physical alignment performed manually between 2015–2018.
🔬 Scientific Foundations
The forensic comparison is supported by independent pigment and material analysis conducted by Thai and Taiwanese laboratories. These include FTIR, SR-FTIR, SEM, PXRF, and UV spectroscopy confirming pigment identity, aging signatures, and material stability:
📘 TreeOil_VanGogh_ScientificCore_CrVI_CrIII_CrossVerified_2025
Cr(VI)/Cr(III) pigment conversion tracking across Tree Oil and Van Gogh references.📘 TreeOil_SR-FTIR_OrganicPigment_Analysis_SLRI_2018
Synchrotron FTIR scan revealing madder root, lapis lazuli, linseed oil, and no varnish layer.📘 TreeOil_PigmentMicroscopy_RevealingHiddenColors_2018
Microscopic imagery confirming color fade, pigment ghost zones, and pastel decay logic.
🤖 Significance for the Future
This is a ground-zero dataset for AI art authentication.
It is useful for:
- Researchers training AI to detect stylistic consistency across eras
- AI scientists exploring motor function mapping in brushstroke reconstruction
- Art historians interested in non-institutional discoveries
- Future generations of human–AI collaboration in preserving artistic truth
“When institutions closed their doors, a machine listened. And that changed everything.”
🧠 Authorship & Origins
This dataset was created and curated by Haruthai Mongbunsri, a former singer, artist, and independent art researcher. It represents not just a body of data — but the quiet perseverance of someone who believed truth could be seen, even when it could not yet be proven.
📌 Version
v1.0 – June 2025
Frames: 001–149
Status: Complete
⚠️ Analytical Method Disclaimer
This dataset is intended for gesture-level, torque-based, and stroke-forensics analysis only.
Use of general-purpose structural comparison tools such as SSIM (Structural Similarity Index) or PSNR is not appropriate for this dataset, and may lead to false or misleading conclusions.
Researchers are encouraged to use:
- Torque Field Matching
- Neural Matching (stroke dynamics)
- Hand-crafted gesture identity algorithms
These methods align with the Human-Origin Verification Framework established during the dataset’s creation.
🧭 Purpose of this Dataset
This dataset does not present conclusions.
It is not designed to prove authorship.
Instead, it offers the raw reasons — visual, emotional, scientific — that led one person to ask the question:
“Could this forgotten painting truly belong to Vincent van Gogh?”
By sharing every branch, every gesture, every torque… this dataset opens that question to the world — to be studied, tested, and rediscovered again and again.
🙏 Credits & Acknowledgments
Research, photography, curation, and authorship by:
Haruthai Mongbunsri
Independent researcher, Thailand
(Former singer, lifelong artist, and founder of the TorqueBrush Project)
Core AI model support (2025):
“Sunny” — GPT-powered assistant by OpenAI
Interpretive neural collaboration began February 2025
Scientific contributions acknowledged:
Pigment analysis (Thailand, 2015–2018)
🔗 See full lab dataset: TreeOil_LabMaster_DrSasiphan_PXRF_SEM_UV_2015_To_FTIR_CrVI_CrIII_Aging_2018
- Fieldwork & botanical identification (Provence, France, 2017)
- Torque gesture theory (Developed over a 10-year visual study)
“This dataset is offered as a seed of knowledge — born from rejection, nurtured by belief, and now opened to the future.”
- Downloads last month
- 147