Radiance

Bake a 7-zone false-color exposure overlay as an IMAGE for pipeline/batch use.

Analyze HDR image histogram with dynamic range statistics, clipping indicators, and stops visualization.

Production QC with comprehensive checks and structured reporting

Visualize color saturation and hue distribution (U/V plot).

Visualize image exposure and channel balance using a Waveform monitor.

Apply ACES 2.0 Output Transform with proper gamut mapping for SDR, HDR, or Cinema output.

Apply a saved grade_info JSON to any image. Replicates a RadianceGrade exactly.

Convert to/from ARRI Wide Gamut 4 (AWG4) for Alexa 35.

ASC CDL transform for shot matching. Supports .cc/.ccc files or manual SOP values.

Apply a 3×3 or 4×4 color matrix to an image. Includes presets (Identity, Sepia, B&W, Invert) and full custom vector entry with optional offset.

GPU-accelerated color space conversion (sRGB, ACEScg, ACEScct, Rec.2020, DCI-P3). Uses industry-standard matrices.

Convert to/from DaVinci Wide Gamut and DaVinci Intermediate.

Expand SDR images to HDR dynamic range by recovering highlights and extending stops of exposure latitude.

Professional 32-bit color correction with exposure, contrast, saturation, gamma, and per-channel lift/gain controls.

GPU-accelerated HDR operations. Fast exposure, gamma, and normalization.

Professional color grading with per-channel LGG/Offset, cinematic presets, LAB match grading, and JSON grade export.

Bakes Radiance grading math into the image tensor permanently.

Match source image color statistics to a reference image using LAB mean/std matching.

Blend multiple exposures (bracketing) for extended dynamic range. Takes highlights from low exposure and shadows from high exposure for optimal color grading.

Recover shadow and highlight detail from a single HDR image for better color grading flexibility.

Professional HDR tone mapping with presets and advanced controls. GPU-accelerated.

Synthesize high dynamic range details in clipped highlights to simulate film scan quality.

Convert images to 32-bit float precision for HDR processing. Preserves full dynamic range without clamping.

Decode log footage (LogC3/4, S-Log3, Log3G10, etc.) to linear with EI control and gamut transform.

Encode linear footage to log curves (LogC3/4, S-Log3, Log3G10, etc.) with EI control.

Blend two LUTs with various blend modes for creative color grading.

Apply an OpenColorIO color space transform between any two spaces defined in the config.

OCIO Display/View transform for professional output. Supports ACES configs with Look and context variables.

Adjust white balance using color temperature (Kelvin) and tint.

Scene-linear workflow presets. Returns decode/working_space/encode settings for color pipeline.

Depth Anything V2 monocular depth estimation. Video-safe — standardizes each frame with spatial-temporal alignment preventing flickering. Outputs 3-channel grayscale depth map. Connect to Depth of Field node for realistic defocus blur.

Reads Radiance Digital DNA signature from all frames in a batch.

QC gate — validates DNA signature (optionally blocks unsigned images).

Signs images with Radiance DNA metadata (preserves HDR special values).

Composite two images using Normal, Add, Screen, Multiply, Overlay, Soft Light, Difference, or Divide blend modes.

Burn-in metadata, timecode, and reports onto images (Slate/Burn-in).

32-bit float bilateral denoising. Preserves edges while smoothing noise. Works on SDR [0,1] and HDR linear-light images equally. Enable hdr_auto_sigma (default ON) for correct HDR operation.

Add compression artifacts (JPEG blocking, color banding).

Apply cinematic depth of field blur with optional depth map input.

Unified photographic grain simulation with camera and film stock profiles. Full float32/HDR pipeline — zero output clamping. Physically-based grain with silver halide clumping, exposure-dependent density, and HDR-safe compositing.

Apply motion blur (directional, radial, or zoom) to simulate camera movement.

Simulate rolling shutter artifacts (skew, wobble, flash banding).

v2.3.3 — Universal cinematic encoder. Auto-selects prose format for Flux/T5/Kolors and structured format for SD1.5/SDXL. Supports scene mood, A/B prompts, subject weighting, real token counts, art direction and architecture-aware negatives.

Compose up to 5 LoRAs into an accumulating LORA_STACK. Chain multiple stacks together. Feed into Radiance Unified Loader.

Generate authentic cinematic prompts using real-world camera and lens profiles.

v4.2 — Universal diffusion sampler. Auto-detects model type (Flux, SD3, SDXL, WAN, LTX, HunyuanVideo, Lumina2, Chroma). Phase-shift sampling, AYS schedules, PAG, dynamic guidance, tiled sampling, multi-conditioning, noise types, refiner chain.

Universal loader v3.1 — auto-detects architecture (Flux, SD3, SDXL, Wan, LTX 2.3, HunyuanVideo, PixArt, Lumina2, etc.), named CLIP slots, chainable LORA_STACK, model_meta JSON, latent_format, offload mode. ZERO stale cache hits (mtime + size fingerprinting).

v2.3 — Universal 4K+ VAE decode. True HDR Compress(Log) with soft-shoulder and highlight denoise for clean, clamp-free HDR output. Auto VAE-factor, 11 color spaces, video 5D support, auto-crop from radiance_meta.

v2.3 — Universal 4K+ VAE encode. Auto VAE-factor detection, video 5D latent support, 11 color spaces, mean/sample/mode latent sampling, batched tile processing, quality report, latent_format output. Compress (Log) HDR mode now passes full scene-linear range through log curve without any post-encode clamp — decode side handles noise cleanly.

AI-powered upscaling using neural network models. Supports tiled processing for large images.

Convert between bit depths with professional dithering to reduce banding.

GPU-accelerated 32-bit downscaling with anti-aliasing.

Generate 360° equirectangular panoramas for HDRI environment mapping in 3D applications.

Professional 32-bit upscaler optimized for Flux with HDR support.

Professional resolution selector with internal preview card. Outputs empty LATENT for Flux/SDXL/SD with correct channel count. 30+ cinema/social/AI presets. Manual latent_channels override for custom architectures.

Upscale to exact dimensions with aspect ratio control.

Non-destructive debug probe. Logs tensor shape/range/dtype, string values, list/dict lengths. Data is passed through unchanged.

Boolean gate — passes data when open=True, returns fallback (or None) when open=False. Use to enable/disable branches without rewiring.

A/B/C/D input multiplexer — select which of up to 4 inputs to pass through using an index (0=A, 1=B, 2=C, 3=D). Perfect for rapid comparisons without rewiring.

Sticky note for workflow documentation. No data connections — pure display. Supports a title, body text, and color selection.

Lightweight named reroute — label any wire without visual overhead. Backward compatible: the label is optional and purely informational. Use Reroute+ for color, icons, styles and group tags.

Professional labeled reroute — color, icon, style, group tag, description tooltip, and live type introspection. Renders as a styled pill via radiance_layout.js.

Load an image with optional non-destructive mask override. If a companion '_radmask.png' file exists alongside the source image, its alpha channel is used as the mask instead of the image's own alpha. The Radiance mask editor frontend saves masks in this companion format.

Displays any input data as text directly on the node UI.

Compute per-frame flicker index (luma delta %) for video QC. Passthrough image, outputs JSON report.

Reduce inter-frame flicker in video batches via per-pixel EMA with optional motion-aware masking.

VFX Industry-Standard Viewer v2.3.3 — Pro Evolution: • GPU Waveform / RGB Parade / Vectorscope / Histogram scopes • Power Windows masking (Radial + Box, feather, rotation) • Comparison Bridge — mouse-draggable wipe + reference shelf (8 stills) • Anamorphic lens streaks + Brown-Conrady k1/k2 distortion • Edge-preserving Bilateral Filter denoising (7×7 GPU kernel) • Channel viewing (RGB/R/G/B/Alpha/Luma), False Color, Zebra • 16-bit PNG + .rhdr HDR sidecar + .exr export • IMAGE passthrough — no longer a dead-end node