ParaView-Scientific-Visualization-AI-Asset-Lab

Portfolio project for a Scientific Visualization (ParaView) Specialist role: AI-assisted scientific visualization asset refinement for multiphysics simulation data.

This repository generates deterministic synthetic 3D simulation data for a thermal plume and airflow around an AI research cooling module, exports ParaView-compatible VTK assets, builds automated ParaView scripts, renders polished preview images, and runs asset QA checks suitable for an AI research visualization workflow.

Reviewer Snapshot

• Primary VTK dataset: data/vtk/cooling_module_primary.vtk
• Preferred XML ImageData dataset: data/vtk/cooling_module_primary.vti
• Time sequence: data/vtk/cooling_module_timeseries.pvd
• ParaView automation: paraview/render_paraview_scene.py
• QA report: outputs/reports/sciviz_asset_qa_report.md
• Executive summary: outputs/reports/executive_summary.md
• 60-second approval guide: outputs/reports/fellowship_reviewer_guide.md

Render Outputs

Animation: outputs/animations/plume_evolution.gif

Why This Fits A ParaView Specialist Role

This project demonstrates the practical work expected from a scientific visualization specialist:

• VTK-compatible scientific data generation and export.
• Structured volume fields with scalar and vector point arrays.
• Obstacle/module polydata and streamline seed polydata.
• Derived fields: velocity magnitude, vorticity, divergence, temperature gradient magnitude, segmentation masks.
• Time-series export through .pvd.
• ParaView automation for slices, contours, thresholds, streamlines, color maps, camera, and screenshots.
• Fallback batch rendering when pvpython is not installed.
• Visualization asset QA for AI research data pipelines.
• Reproducible command-line execution and tests.

Technical Stack

• Python 3.12 tested locally
• NumPy
• Matplotlib
• Pillow
• Pytest
• ParaView / pvpython optional
• VTK and PyVista optional, not required

Dataset Description

The synthetic scenario is a thermal plume and airflow around an AI research cooling module. The 3D domain contains:

• temperature: Kelvin scalar field.
• pressure: Pascal scalar field.
• density: derived scalar field.
• velocity: 3-component vector field.
• velocity_magnitude: derived scalar field.
• vorticity and vorticity_magnitude: derived vector/scalar fields.
• divergence: vector field quality diagnostic.
• temperature_gradient_magnitude: edge/transition diagnostic.
• module_mask: obstacle segmentation.
• hotspot_mask: high-temperature threshold segmentation.

All data are generated locally and are license-safe.

Visualization Pipeline

python scripts/00_generate_simulation_data.py
python scripts/01_preprocess_fields.py
python scripts/02_export_vtk_assets.py
python scripts/03_render_previews.py
python scripts/04_asset_qa.py
python scripts/05_build_paraview_state.py
python scripts/06_build_submission_manifest.py
python scripts/run_pipeline.py

On this Windows workspace, use the repository virtual environment if the global python alias is disabled:

.\.venv\Scripts\python.exe scripts\run_pipeline.py
.\.venv\Scripts\python.exe scripts\validate_outputs.py
.\.venv\Scripts\python.exe -m pytest

ParaView Opening Instructions

1. Open ParaView.
2. Load data/vtk/cooling_module_primary.vti for the clean XML ImageData reference view, or data/vtk/cooling_module_timeseries.pvd for temporal playback.
3. Load data/vtk/cooling_module_obstacle.vtp or data/vtk/cooling_module_obstacle.vtk.
4. Optionally load data/vtk/streamline_seed_points.vtk.
5. Apply a Slice on temperature, a Contour at 342 K and 355 K, a Threshold on hotspot_mask, and Stream Tracer With Custom Source using velocity.
6. Use paraview/color_presets.json and paraview/camera_presets.json for consistent presentation.

If pvpython is available, run:

pvpython paraview/render_paraview_scene.py

That script exports outputs/renders/paraview_scene_overview.png and a .pvsm state file. If ParaView is not installed, paraview/STATE_EQUIVALENT.md documents the script-based equivalent.

QA Checks

The asset QA module validates:

• Missing dataset files.
• Malformed or unreadable VTK headers.
• Missing scalar/vector arrays.
• NaN and infinite values.
• Temperature range sanity.
• Dimension consistency.
• Missing time steps.
• Metadata schema.
• Naming conventions.
• Render existence, image size, and nonblank output.

Outputs:

• outputs/reports/sciviz_asset_qa_report.md
• outputs/tables/asset_qa_findings.csv
• data/processed/normalized_metadata.json

Repository Structure

data/
  raw/                 deterministic NumPy timestep bundles
  processed/           metadata, summaries, reference fields
  vtk/                 ParaView-compatible VTK and PVD assets
paraview/              automation scripts, color presets, camera presets
scripts/               numbered pipeline and validation scripts
outputs/
  renders/             PNG preview renders
  animations/          plume GIF
  reports/             Markdown reports
  tables/              CSV QA and field summaries
docs/                  methodology and portfolio notes
tests/                 pytest validation tests

Reports

• outputs/reports/executive_summary.md
• outputs/reports/fellowship_reviewer_guide.md
• outputs/reports/methodology.md
• outputs/reports/sciviz_asset_qa_report.md
• outputs/reports/reproducibility_report.md

Submission Hygiene

This workspace may contain unrelated local projects. For a fellowship submission, create a clean GitHub repository named ParaView-Scientific-Visualization-AI-Asset-Lab and include the project files listed in outputs/tables/submission_manifest.csv. The checklist is written to outputs/reports/github_submission_checklist.md.

To materialize a clean local folder for publication:

python scripts/07_prepare_clean_github_repo.py

That creates dist/ParaView-Scientific-Visualization-AI-Asset-Lab/.

Limitations

The simulation is synthetic and analytic, not a CFD solver result. The Python renders are faithful preview assets, but ParaView remains the intended environment for production-grade volume rendering, transfer-function tuning, clipping, and interactive streamline review.

Future Improvements

• Add native XML .vti writers with appended binary payloads.
• Add ParaView Catalyst-style in situ examples.
• Add GPU volume rendering presets for production ParaView.
• Add a small active-learning loop for image quality annotations.

What This Demonstrates

For hiring reviewers, this project shows end-to-end ownership of a scientific visualization workflow: deterministic data generation, VTK pipeline design, scalar/vector derived fields, ParaView automation, camera and color design, batch rendering, reproducibility, and asset QA for AI research visualization refinement.