prism-viewer 1.1.0

OCIO still-image viewer for A/B comparison

Prism

(Pipeline Review and Image State Monitor)

Prism is a lightweight OCIO image viewer built for fast visual inspection, transform validation, and A/B comparison workflows.

It was created from a very practical need: quickly loading images, testing colorspace transforms, comparing outputs, and debugging OCIO behavior without having to launch a full compositing or grading application.

Prism focuses on still-image workflows and immediate visual feedback:

• compare two images side-by-side
• wipe between transforms
• inspect differences
• validate looks and context variables
• probe pixel values
• and quickly iterate on OCIO decisions

The goal is not to replace tools like Nuke, RV, or Resolve.

The goal is to provide a focused, lightweight environment for color pipeline inspection and experimentation, especially during look development, pipeline debugging, config validation, and day-to-day production support.

Built with:

• PySide6
• OpenColorIO
• OpenImageIO
• NumPy

---

Why Prism Exists

Many OCIO workflows still rely on launching large DCC applications just to validate a transform, compare two outputs, or debug context-dependent behavior.

Prism was built to reduce that friction.

It provides a fast standalone environment for inspecting color behavior directly, with minimal setup and immediate visual feedback.

Whether you're:

• validating an OCIO config
• checking a display transform
• comparing renders
• debugging context variables
• verifying looks
• or simply inspecting image data

Prism aims to make that process faster and more direct.

---

Features

• Load and compare two images simultaneously
• Apply independent OCIO transforms to A and B
• Compare images using Split, Wipe, Full, and Diff modes
• Inspect pixel values directly in the viewer HUD
• Adjust exposure and luminance interactively
• Edit OCIO context variables live
• Inspect LUT transfer curves in a dedicated modeless LUT window
• Navigate synchronized A/B views
• Toggle transform bypass per image
• Use EXR, DPX, JPG, PNG, TIFF, and other common formats
• Built-in diagnostics window for environment validation

---

Requirements

• Python 3.11+

• Runtime packages:

  • PySide6
  • OpenImageIO
  • OpenColorIO
  • numpy

---

Install

Install from PyPI

python -m pip install prism-viewer

Install for Development:

python -m pip install -e .

---

Launch

prism

If the command is not found, activate your virtual environment first and retry.

---

Quick Start

1. Open image A from File -> Open Image A...
2. Open image B from File -> Open Image B...
3. Load an OCIO config from File -> Open OCIO Config...

4. Select Input and Output colorspaces independently for A and B
5. Optionally select a Look
6. Switch compare modes (Split, Wipe, Full (A), Full (B) Diff)

7. Hover over the viewer to inspect pixel values in the footer HUD

Prism also supports drag-and-drop for images and OCIO configs.

---

Compare Modes

Split

Shows A and B side-by-side.

Useful for:

• framing comparisons
• look validation
• general A/B review

---

Wipe

Displays A/B across an interactive divider.

Use the wipe handle to inspect transitions and subtle image differences interactively.

---

Full

Displays only the currently selected side (A or B).

Useful for focused single-image inspection while preserving the compare setup.

---

Diff

Displays grayscale absolute difference between overlapping regions.

Brighter values indicate stronger differences.

Useful for:

• transform validation
• render verification
• subtle change detection

---

OCIO Context Variables

Prism can detect and expose OCIO context variables directly from the loaded config.

Open from:

• View -> OCIO Context Variables
• Hotkey: E

Behavior:

• Hidden by default
• Dynamically rebuilds from the loaded config
• Updates processing live as values change
• Can be docked or undocked
• Persists current values while visible or hidden

This is especially useful for:

• shot-based transforms
• context-driven looks
• sequence-dependent processing
• debugging OCIO environment behavior

---

Viewer Controls

Exposure and Luminance

Interactive controls allow quick image inspection without modifying source data.

Useful for:

• shadow inspection
• highlight clipping checks
• transform debugging
• general image evaluation

---

Background Presets

Open from:

• View -> Background

Presets:

• Black
• Dark Gray
• Mid Gray
• Light Gray

Useful when image edges become difficult to read against the viewer background.

---

LUT Inspection

Open from:

• View -> LUT Inspection or Drop LUT in main_window

Behavior:

• opens as a modeless utility window
• accepts drag-and-drop LUT files
• supports .cube, .spi1d, .csp, .spi3d, .lut, and .3dl formats
• plots LUT curves on an X/Y diagram that scales with window size
• for 3D .cube, displays a neutral-axis projection (R=G=B samples)
• for .cube, DOMAIN_MIN and DOMAIN_MAX must use matching RGB components (mixed per-channel domain values are treated as unsupported)

Diagnostics

Prism includes a diagnostics window for quickly validating runtime dependencies and OCIO environment state.

Includes:

• Prism version
• Python version
• Qt / PySide version
• NumPy version
• OpenImageIO version
• OpenColorIO version
• Active OCIO config
• Platform information

Useful for:

• support,
• debugging,
• pipeline validation,
• environment troubleshooting.

---

Typical Workflows

A/B Grade Validation

1. Load image A and B
2. Load an OCIO config
3. Configure Input and Output transforms
4. Start in Split
5. Move to Wipe for edge comparison
6. Use Diff to isolate subtle changes

---

Single Image Inspection

1. Load image A
2. Configure transforms
3. Switch to Full
4. Inspect pixel values and channels
5. Adjust exposure and luminance interactively

---

Context-Based Look Debugging

1. Load image and OCIO config
2. Open View -> OCIO Context Variables
3. Modify context values
4. Observe updates live in the viewer

---

Hotkeys

• F Fit to Window

• H Show/Hide Hotkeys dialog

• E Toggle OCIO Context Variables dock

• Tab Switch active side A/B

• Right / PageDown Next frame

• Left / PageUp Previous frame

• Up / Down Change focused Input/Output item

• R Toggle Red solo / RGB

• G Toggle Green solo / RGB

• B Toggle Blue solo / RGB

• Z Restore RGB

• 1 / 2 / 3 / 4 / 5

  • Split
  • Wipe
  • Full (A)
  • Full (B)
  • Diff

• Right Click + Drag Pan

---

Sample Assets

Prism includes sample files to validate setup quickly:

• Sample images: samples/images/
• Sample OCIO config: samples/ocio_config/
• Sample LUTs: sample/LUTs

Quick smoke test:

1. Launch prism
2. Load image A and B from samples/images/
3. Load config from samples/ocio_config/
4. Switch between Split, Wipe, Full (A), Full (B), and Diff

---

Troubleshooting

prism command not found

Activate the intended virtual environment and reinstall from repository root:

python -m pip install -e .

---

OCIO config fails to load

• Confirm OpenColorIO is installed in the active environment
• Confirm the selected file is a valid .ocio config
• If loading is canceled intentionally, Prism preserves current state

---

EXR or DPX images fail to load

Prism uses OpenImageIO fallback support for formats that may not be available through Qt directly.

Verify:

• OpenImageIO is installed
• the active environment can import it successfully

---

Movie source issues

Movie playback depends on optional backend availability such as:

• opencv-python
• imageio[ffmpeg]

Still-image workflows remain fully functional without them.

---

Known Limitations

• Movie workflows remain backend/environment dependent
• Some EXR/DPX workflows rely on OpenImageIO fallback support
• Windows taskbar icon refresh may lag behind icon updates until Explorer refreshes shell cache

---

Planned Features

• Waveform Monitor
• Vectorscope
• Chromaticity Diagram and gamut visualization
• Transform breakdown visualization
• Nuke integration
• Resolve integration
• More Display/View support

---

Architecture and Developer Docs

See docs/README.md for subsystem and architecture documentation:

• Architecture
• Core
• I/O
• UI
• OCIO Context
• Packaging