Node Reference

Comprehensive guide to all 74 nodes in Radiance v2.3.3

◎ Color Science & HDR

33 nodes

◎ White Balance ▼

RadianceWhiteBalance

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

How it Works

This node calculates the physical color of the light source and applies a chromatic adaptation transform to shift the image's color temperature (Kelvin) and green/magenta tint. It works in linear space to mathematically preserve physical skin tones and neutral grays, unlike simple color overlays.

When to use it

Use this early in your color pipeline to correct improperly shot footage, or to creatively warm up/cool down a scene without breaking realistic color relationships.

Inputs

Name	Type	Default
image	`IMAGE`
preset	`?`
temperature	`INT`
tint	`FLOAT`
source_temperature (opt)	`INT`
intensity (opt)	`FLOAT`
use_gpu (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Depth of Field ▼

RadianceDepthOfField

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

How it Works

Simulates a physical camera lens by blurring out-of-focus areas based on a depth map. Supports different bokeh shapes and highlights boosting, replicating optical aberrations of cinematic lenses.

When to use it

Apply when you have a separate depth pass (or one generated via Depth Anything) to add realistic foreground/background separation and draw the viewer's eye to the main subject.

Inputs

Name	Type	Default
image	`IMAGE`
blur_amount	`FLOAT`
depth_map (opt)	`IMAGE`
focus_distance (opt)	`FLOAT`
focus_range (opt)	`FLOAT`
bokeh_shape (opt)	`BOKEH_SHAPES`
highlight_boost (opt)	`FLOAT`
foreground_blur (opt)	`BOOLEAN`
use_gpu (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Motion Blur ▼

RadianceMotionBlur

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

How it Works

Accumulates samples along a vector (directional, radial, or zoom) to mimic the temporal blurring effect of a slow shutter speed on moving subjects or a moving camera.

When to use it

Use to add energy and speed to AI generations (which often look perfectly sharp and artificial) or to simulate aggressive camera pans and zooms.

Inputs

Name	Type	Default
image	`IMAGE`
blur_type	`BLUR_TYPES`
amount	`FLOAT`
angle (opt)	`FLOAT`
center_x (opt)	`FLOAT`
center_y (opt)	`FLOAT`
samples (opt)	`INT`
use_gpu (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Rolling Shutter ▼

RadianceRollingShutter

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

How it Works

Introduces non-simultaneous sensor readout artifacts such as skew (jello effect) during fast pans, wobble vibrations, or partial-frame flash banding normally seen in CMOS sensors.

When to use it

Perfect for achieving a 'found footage', amateur video, or handheld chaotic look that makes CGI/AI feel grounded in physical reality.

Inputs

Name	Type	Default
image	`IMAGE`
skew_amount	`FLOAT`
shutter_direction (opt)	`SHUTTER_MODES`
wobble_frequency (opt)	`FLOAT`
wobble_amplitude (opt)	`FLOAT`
flash_band_position (opt)	`FLOAT`
flash_band_width (opt)	`FLOAT`
use_gpu (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Compression Artifacts ▼

RadianceCompressionArtifacts

Add compression artifacts (JPEG blocking, color banding).

How it Works

Downsamples the image into DCT blocks and discarding high-frequency detail to simulate JPEG/MPEG compression, color banding from low bitrates, and chroma subsampling.

When to use it

Use for lo-fi aesthetics, simulating dashboard cameras, early digital video tape formats, or web streaming glitches.

Inputs

Name	Type	Default
image	`IMAGE`
artifact_type	`ARTIFACT_TYPES`
quality	`INT`
block_size (opt)	`INT`
color_subsampling (opt)	`BOOLEAN`
banding_levels (opt)	`INT`
noise_amount (opt)	`FLOAT`
seed (opt)	`INT`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ LUT Apply ▼

RadianceLUTApply

Apply a 3D LUT (.cube) with trilinear or tetrahedral interpolation. Supports log-space input and HDR (unclamped) output.

How it Works

Loads an industry-standard 3D LUT (.cube file) and interpolates the input colors using trilinear or high-quality tetrahedral math. It seamlessly supports clamping for standard SDR or pure float32 HDR when unclamped.

When to use it

Apply standard film emulation LUTs, conversion LUTs, or creative looks exported from grading software like DaVinci Resolve.

Inputs

Name	Type	Default
image	`IMAGE`
lut_file	`?`
strength	`FLOAT`
log_space	`BOOLEAN`
log_encoding (opt)	`LOG_ENCODINGS`
clamp_output (opt)	`BOOLEAN`
interpolation (opt)	`COMBO`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Color Matrix ▼

RadianceGPUColorMatrix

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.

How it Works

Multiplies each pixel's RGB values by a custom 3x3 or 4x4 matrix on the GPU, allowing complex channel mixing, color inversions, sepia tones, and matrix transformations at lightning speeds.

When to use it

Use for primary color corrections, creating custom black and white mixes (e.g., infrared emulation), or mathematically precise gamut conversions not natively supported.

Inputs

Name	Type	Default
image	`IMAGE`
preset	`?`
matrix_type	`COMBO`
r_vector	`STRING`
g_vector	`STRING`
b_vector	`STRING`
a_vector (opt)	`STRING`
offset (opt)	`STRING`
clamp_output (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ OCIO Transform ▼

RadianceOCIOColorTransform

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

How it Works

Leverages the OpenColorIO (OCIO) standard used in VFX to map colors between two defined spaces within a custom configuration file.

When to use it

Required when ingesting elements from different sources (like CGI renders in ACEScg vs web images in sRGB) to ensure a unified working color space.

Inputs

Name	Type	Default
image	`IMAGE`
config_file	`STRING`
input_space	`STRING`
output_space	`STRING`
direction	`COMBO`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance OCIO Display/View ▼

RadianceOCIODisplayView

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

How it Works

Applies an OCIO Display/View transform, formatting scene-referred linear data into display-referred formats with specific viewing conditions, tonemapping, and filmic looks baked in.

When to use it

Use as the final node before saving to disk to ensure the pixel values correctly match the monitor or target display (e.g. Rec.709, sRGB, P3-D65).

Inputs

Name	Type	Default
image	`IMAGE`
config_path	`STRING`
display	`STRING`
view	`STRING`
look (opt)	`STRING`
exposure_adjust (opt)	`FLOAT`
context_key (opt)	`STRING`
context_value (opt)	`STRING`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance CDL Transform ▼

RadianceOCIOCDL

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

How it Works

Applies ASC CDL (American Society of Cinematographers Color Decision List) math (Slope, Offset, Power) linearly over the image. It can read these values directly from `.cc` or `.ccc` files.

When to use it

Ideal for matching grades across different tools, applying preliminary set-looks, or fine-tuning exposure and contrast objectively.

Inputs

Name	Type	Default
image	`IMAGE`
cdl_file (opt)	`STRING`
slope_r (opt)	`FLOAT`
slope_g (opt)	`FLOAT`
slope_b (opt)	`FLOAT`
offset_r (opt)	`FLOAT`
offset_g (opt)	`FLOAT`
offset_b (opt)	`FLOAT`
power_r (opt)	`FLOAT`
power_g (opt)	`FLOAT`
power_b (opt)	`FLOAT`
saturation (opt)	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance Workflow Presets ▼

RadianceSceneLinearWorkflow

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

How it Works

A preset manager that automatically configures decoding, working space, and encoding variables for standardized pipelines (e.g., ACES, DaVinci Wide Gamut).

When to use it

Use this to quickly set up pipeline variables at the start of a flow instead of manually typing in color science parameters across multiple nodes.

Inputs

Name	Type	Default
preset	`?`

Outputs

Name	Type
STRING	`STRING`
STRING	`STRING`
STRING	`STRING`

◎ Log Curve Decode ▼

RadianceLogCurveDecode

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

How it Works

Flattens manufacturer-specific logarithmic curves (ARRI LogC, Sony S-Log3, RED Log3G10) back into linear light values while allowing precise Exposure Index (ISO) offsets.

When to use it

Place immediately after loading raw/log video clips to bring them into a scene-linear working space for mathematically accurate compositing and grading.

Inputs

Name	Type	Default
image	`IMAGE`
curve	`?`
target_gamut	`GAMUTS`
exposure_index (opt)	`EI_OPTIONS`
use_gpu (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Log Curve Encode ▼

RadianceLogCurveEncode

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

How it Works

Takes scene-linear data and compresses the dynamic range into a standard Log curve, safely packing highlight information for 10-bit/12-bit delivery or intermediate storage.

When to use it

Use before saving intermediate EXR/ProRes files that you intend to further grade in DaVinci Resolve or Nuke.

Inputs

Name	Type	Default
image	`IMAGE`
curve	`?`
source_gamut	`GAMUTS`
exposure_index (opt)	`EI_OPTIONS`
apply_gamma (opt)	`BOOLEAN`
gamma (opt)	`FLOAT`
exposure (opt)	`FLOAT`
clamp_output (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance ACES 2.0 Output ▼

RadianceACES2OutputTransform

ACES 2.0 Output Transform for professional SDR/HDR display delivery with improved gamut mapping.

How it Works

Executes the new ACES 2.0 reference rendering math, fixing legacy issues like the 'blue light artifact', and elegantly mapping high dynamic range scene data into display constraints with parametric gamut compression.

When to use it

The ultimate final step for outputting cinematic images, replacing older tone mappers for the highest quality color fidelity.

Inputs

Name	Type	Default
image	`IMAGE`
display_target	`DISPLAY_TARGETS`
peak_luminance	`FLOAT`
contrast	`FLOAT`
gamut_compress (opt)	`BOOLEAN`
gamut_threshold (opt)	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance LUT Blend ▼

RadianceLUTBlend

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

How it Works

Combines two 3D LUTs by interpolating their lattice points or blending them dynamically using modes like Add, Multiply, or Overlay before applying the result to the image.

When to use it

Use to creatively mix film stock emulations, or to ease a transition between two completely different color grades smoothly.

Inputs

Name	Type	Default
image	`IMAGE`
lut_a	`?`
lut_b	`?`
blend_factor	`FLOAT`
blend_mode	`BLEND_MODES`
strength (opt)	`FLOAT`
clamp_output (opt)	`BOOLEAN`
log_space (opt)	`BOOLEAN`
log_encoding (opt)	`LOG_ENCODINGS`
interpolation (opt)	`COMBO`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance 32-bit Denoise ▼

RadianceDenoise

Removes noise while preserving edges using a 32-bit float compatible Bilateral Filter.

How it Works

Uses an edge-preserving bilateral filter running on the GPU that works natively in 32-bit float, removing noise in flat areas while leaving sharp high-contrast edges untouched.

When to use it

Apply to noisy AI outputs before upscaling or grain addition to ensure artifacts aren't artificially enhanced.

Inputs

Name	Type	Default
image	`IMAGE`
d	`INT`
sigmaColor	`FLOAT`
sigmaSpace	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Film Grain ▼

RadianceFilmGrain

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.

How it Works

Procedurally generates authentic Gaussian film grain that varies in size and density based on the underlying image luminance, correctly modeling the physical behavior of silver-halide crystals.

When to use it

Add this at the very end of your pipeline to unify disparate AI or 3D elements, mask digital banding, and give an organic, physical texture to the final output.

Inputs

Name	Type	Default
image	`IMAGE`
profile	`?`
intensity	`FLOAT`
seed	`INT`
size (opt)	`FLOAT`
iso (opt)	`INT`
halation (opt)	`FLOAT`
blend_mode (opt)	`COMBO`
gate_weave (opt)	`FLOAT`
use_gpu (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance Grade ▼

RadianceGrade

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

How it Works

Provides traditional Lift, Gamma, Gain, and Offset controls tailored for 32-bit HDR images alongside cinematic preset looks. Output includes the grade math in JSON format.

When to use it

Your primary color correction tool for balancing shots, dialing in a creative look, or establishing the mood of an AI generation.

Inputs

Name	Type	Default
image	`IMAGE`
preset	`?`
preset_strength	`FLOAT`
reference_image (opt)	`IMAGE`
match_strength (opt)	`FLOAT`
preset_file (opt)	`STRING`
lift_r (opt)	`FLOAT`
lift_g (opt)	`FLOAT`
lift_b (opt)	`FLOAT`
gamma_r (opt)	`FLOAT`
gamma_g (opt)	`FLOAT`
gamma_b (opt)	`FLOAT`
gain_r (opt)	`FLOAT`
gain_g (opt)	`FLOAT`
gain_b (opt)	`FLOAT`
offset_r (opt)	`FLOAT`
offset_g (opt)	`FLOAT`
offset_b (opt)	`FLOAT`
contrast (opt)	`FLOAT`
pivot (opt)	`FLOAT`
saturation (opt)	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

◎ Apply Grade Info ▼

RadianceApplyGradeInfo

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

How it Works

Takes the JSON structural data from a `RadianceGrade` node and re-applies the exact same math to a new input image without needing to duplicate the settings or nodes.

When to use it

Use in batch processing or multi-pass setups where you want to construct a single "hero" grade and propagate it to other generations automatically.

Inputs

Name	Type	Default
image	`IMAGE`
grade_info	`STRING`
strength (opt)	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

◎ Grade Match ▼

RadianceGradeMatch

Match source image color statistics to a reference image using enhanced CIE L*a*b* mean/std matching for accurate shot-to-shot consistency.

How it Works

Analyzes the mean and standard deviation of L*a*b* color channels from a reference image and mathematically shifts the source image to match this statistical profile.

When to use it

Crucial for shot-matching across a scene, ensuring consistent color palettes from reverse-angle shots, or stealing the look of a cinematic reference frame.

Inputs

Name	Type	Default
source	`IMAGE`
reference	`IMAGE`
strength	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

◎ Radiance Image to Float32 ▼

ImageToFloat32

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

How it Works

Transforms standard 8-bit or 16-bit integer pixel data into full 32-bit floating point precision, unlocking values outside the normal 0.0 to 1.0 range.

When to use it

Use right after loading standard images (like PNGs) to prepare them for advanced math, compositing, and HDR manipulation.

Inputs

Name	Type	Default
image	`IMAGE`
normalize (opt)	`BOOLEAN`
source_gamma (opt)	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Color Correct ▼

Float32ColorCorrect

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

How it Works

Performs fundamental exposure, contrast, brightness, and saturation adjustments using strict float math without clamping highlights or crushing blacks.

When to use it

Ideal for fast technical adjustments to fix blown-out highlights or underexposed shadows before entering the main grade.

Inputs

Name	Type	Default
image	`IMAGE`
exposure	`FLOAT`
contrast	`FLOAT`
brightness	`FLOAT`
saturation	`FLOAT`
gamma (opt)	`FLOAT`
lift_r (opt)	`FLOAT`
lift_g (opt)	`FLOAT`
lift_b (opt)	`FLOAT`
gain_r (opt)	`FLOAT`
gain_g (opt)	`FLOAT`
gain_b (opt)	`FLOAT`
luma_space (opt)	`?`
clamp_output (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance HDR Expand Dynamic Range ▼

HDRExpandDynamicRange

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

How it Works

Synthetically expands the dynamic range of SDR footage by pulling up highlights based on a selected rolloff curve, restoring a pseudo-HDR latitude.

When to use it

Use when you have standard web images (like those generated by basic AI models) but you want to grade them in an HDR pipeline.

Inputs

Name	Type	Default
image	`IMAGE`
source_gamma	`FLOAT`
highlight_recovery	`FLOAT`
black_point	`FLOAT`
target_stops	`FLOAT`
highlight_rolloff	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance HDR Tone Map ▼

HDRToneMap

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

How it Works

Compresses high-dynamic-range pixel values back into a viewable SDR range using advanced operators like Reinhard, Filmic, or ACES heuristics.

When to use it

Required when outputting HDR composited work to standard 8-bit formats like JPEG without losing highlight details to pure white clipping.

Inputs

Name	Type	Default
image	`IMAGE`
preset (opt)	`LOOK_PRESETS`
operator (opt)	`TONEMAP_OPERATORS`
exposure (opt)	`FLOAT`
gamma (opt)	`FLOAT`
white_point (opt)	`FLOAT`
contrast (opt)	`FLOAT`
saturation (opt)	`FLOAT`
highlight_compression (opt)	`FLOAT`
shadow_lift (opt)	`FLOAT`
use_gpu (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance Color Space Convert ▼

ColorSpaceConvert

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

How it Works

Applies standard ICC matrix math to convert pixel data between common color spaces (e.g., sRGB to Rec.2020) without the overhead of OCIO configurations.

When to use it

Use for fast, lightweight color conversions on the fly, especially when targeting different web or mobile display standards.

Inputs

Name	Type	Default
image	`IMAGE`
source_space	`COLOR_SPACES`
target_space	`COLOR_SPACES`
exposure (opt)	`FLOAT`
gamma_adjust (opt)	`FLOAT`
chromatic_adaptation (opt)	`CHROMATIC_ADAPTATIONS`
use_gpu (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance HDR Exposure Blend ▼

HDRExposureBlend

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

How it Works

Fuses multiple bracketed exposures into a single 32-bit HDR frame by taking highlight details from the darkest exposure and shadow details from the brightest.

When to use it

Combine standard Lora bracket generations (over/under exposed prompts) into a master file with unclipped latitude.

Inputs

Name	Type	Default
low_exposure	`IMAGE`
high_exposure	`IMAGE`
blend_method	`BLEND_METHODS`
mid_exposure (opt)	`IMAGE`
shadow_weight (opt)	`FLOAT`
highlight_weight (opt)	`FLOAT`
transition_smoothness (opt)	`FLOAT`
exposure_offset_low (opt)	`FLOAT`
exposure_offset_high (opt)	`FLOAT`
exposure_offset_mid (opt)	`FLOAT`
ghost_removal (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`
IMAGE	`IMAGE`
STRING	`STRING`

◎ Shadows & Highlights ▼

HDRShadowHighlightRecovery

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

How it Works

Applies a local contrast preservation algorithm to specifically lift shadows and compress highlights while maintaining mid-tone integrity.

When to use it

Perfect for rescuing detail in high-contrast outdoor scenes, similar to Lightroom's shadow/highlight sliders.

Inputs

Name	Type	Default
image	`IMAGE`
shadow_amount	`FLOAT`
highlight_amount	`FLOAT`
shadow_tone (opt)	`FLOAT`
highlight_tone (opt)	`FLOAT`
color_correction (opt)	`FLOAT`
local_contrast (opt)	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance OCIO List Colorspaces ▼

OCIOListColorspaces

List all colorspaces available in an OCIO configuration file.

How it Works

A utility that parses an attached OCIO config file and returns an array of all available color spaces and families.

When to use it

Use to populate custom scripts or dynamically route color transforms based on the available config rather than hardcoding names.

Inputs

Name	Type	Default
ocio_config_path (opt)	`STRING`
filter_family (opt)	`STRING`

Outputs

Name	Type
STRING	`STRING`

◎ Radiance GPU Tensor Ops ▼

GPUTensorOps

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

How it Works

Bypasses standard CPU arrays to perform basic arithmetic (Multiply, Add, Power) directly on the tensor using PyTorch GPU operations.

When to use it

Use for lightning-fast image math specifically optimized for massive 32-bit float batches where CPU processing bottlenecks.

Inputs

Name	Type	Default
image	`IMAGE`
operation	`COMBO`
value (opt)	`FLOAT`
force_gpu (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

◎ Radiance ACES 2.0 Output Transform ▼

ACES2OutputTransform

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

How it Works

Applies the ACES 2.0 Output Transform with proper gamut mapping for SDR, HDR, or Cinema output, ensuring out-of-bounds colors roll off smoothly.

When to use it

Use exclusively for professional delivery, substituting standard tone mapping for industry-regulated cinema/broadcast targets.

Inputs

Name	Type	Default
image	`IMAGE`
input_colorspace	`COMBO`
output_transform	`OUTPUT_TRANSFORMS`
peak_luminance (opt)	`FLOAT`
surround (opt)	`COMBO`
creative_white_scale (opt)	`FLOAT`
exposure_adjust (opt)	`FLOAT`
gamut_compress (opt)	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

◎ Radiance ACES Config Manager ▼

ACESConfigManager

Detect, download, and manage ACES OCIO configurations.

How it Works

Automatically downloads, verifies, and manages the large generic OCIO configuration files from the ACES Academy repository.

When to use it

Run once to set up your environment, or use in automated installations to ensure the correct color science files are available for other nodes.

Inputs

Name	Type	Default
action	`COMBO`
custom_config_path (opt)	`STRING`
install_path (opt)	`STRING`

Outputs

Name	Type
STRING	`STRING`
STRING	`STRING`
STRING	`STRING`

◎ Radiance DaVinci Wide Gamut ▼

DaVinciWideGamut

Convert to/from DaVinci Wide Gamut and DaVinci Intermediate.

How it Works

Converts RGB data into or out of DaVinci Wide Gamut/Intermediate space using the official Blackmagic Design transform matrices.

When to use it

Your bridge point when exporting EXRs from ComfyUI to be professionally color-graded inside DaVinci Resolve.

Inputs

Name	Type	Default
image	`IMAGE`
transform	`TRANSFORMS`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance ARRI Wide Gamut 4 ▼

ARRIWideGamut4

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

How it Works

Applies the official ARRI LogC4 and AWG4 conversion math designed for the Alexa 35 sensor.

When to use it

Use when compositing AI generations over modern ARRI Alexa footage to ensure the digital noise and color spaces match perfectly.

Inputs

Name	Type	Default
image	`IMAGE`
direction	`COMBO`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance Highlight Synthesis ▼

RadianceHighlightSynthesis

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

How it Works

Analyzes clipped white areas (over 1.0) and synthetically generates high-frequency detail and glow to simulate the natural halation and bleed of actual celluloid film.

When to use it

Use to fix blown-out CGI or bright AI areas, giving them an organic, optical feeling rather than a flat digital white.

Inputs

Name	Type	Default
image	`IMAGE`
threshold	`FLOAT`
expansion	`FLOAT`
detail_amount	`FLOAT`
detail_scale	`FLOAT`
blend_mode	`COMBO`
seed	`INT`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Radiance LUT Bake ▼

RadianceLUTBake

Bake Radiance grade parameters into a .cube LUT file for Resolve / Nuke / Premiere.

How it Works

Takes all the mathematical parameters built inside a Radiance node network and iterates them over a 3D lattice, saving the result as a portable .cube file.

When to use it

Use to design a custom look inside ComfyUI and then export that exact look to use in Premiere, Nuke, or external monitors.

Inputs

Name	Type	Default
image	`IMAGE`
output_path	`STRING`
lut_title	`STRING`
lut_size	`?`
lift_r (opt)	`FLOAT`
lift_g (opt)	`FLOAT`
lift_b (opt)	`FLOAT`
gamma_r (opt)	`FLOAT`
gamma_g (opt)	`FLOAT`
gamma_b (opt)	`FLOAT`
gain_r (opt)	`FLOAT`
gain_g (opt)	`FLOAT`
gain_b (opt)	`FLOAT`
offset_r (opt)	`FLOAT`
offset_g (opt)	`FLOAT`
offset_b (opt)	`FLOAT`
contrast (opt)	`FLOAT`
pivot (opt)	`FLOAT`
saturation (opt)	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

◎ Radiance Grade Apply ▼

FXTD_RadianceGradeApply

Bakes Radiance grading math into the image tensor permanently.

How it Works

Bakes complex grading operations (exposure, lift/gamma/gain, color temp) permanently into the image tensor's pixel values instead of retaining them as metadata.

When to use it

Use when you need to flatten your node tree for external export or to ensure subsequent math strictly calculates off the graded pixels.

Inputs

Name	Type	Default
image	`IMAGE`
exposure	`FLOAT`
offset	`FLOAT`
lift	`FLOAT`
gamma	`FLOAT`
gain	`FLOAT`
contrast	`FLOAT`
pivot	`FLOAT`
shadows	`FLOAT`
highlights	`FLOAT`
saturation	`FLOAT`
temperature	`FLOAT`
hue_shift	`FLOAT`
lut_name	`?`
lut_intensity	`FLOAT`
color_science	`COMBO`

Outputs

Name	Type
IMAGE	`IMAGE`

◎ Film & Camera

10 nodes

◎ Radiance Unified Loader ▼

RadianceUnifiedLoader

Universal loader v2.3.3 — auto-detects architecture, named CLIP slots, chainable LORA_STACK, model_meta JSON, latent_format, offload mode. ZERO stale cache hits (mtime + size fingerprinting).

How it Works

A master wrapper that intelligently auto-detects UNET architecture, loads models/CLIP/VAEs, stacks up to 5 LoRAs, and handles model_meta JSON caching natively without stale hits.

When to use it

Use this as your single starting point for any generation, replacing the clutter of 5-10 standard ComfyUI loader nodes.

Inputs

Name	Type	Default
preset	`?`
unet_name	`?`
weight_dtype	`?`
model_type	`?`
vae_name	`?`
clip_l (opt)	`?`
clip_g (opt)	`?`
t5xxl (opt)	`?`
llm_encoder (opt)	`?`
clip_dtype (opt)	`?`
offload_mode (opt)	`?`
lora_stack (opt)	`LORA_STACK`
lora_1 (opt)	`?`
lora_1_model_str (opt)	`FLOAT`
lora_1_clip_str (opt)	`FLOAT`
lora_2 (opt)	`?`
lora_2_model_str (opt)	`FLOAT`
lora_2_clip_str (opt)	`FLOAT`
lora_3 (opt)	`?`
lora_3_model_str (opt)	`FLOAT`
lora_3_clip_str (opt)	`FLOAT`
controlnet_name (opt)	`?`
controlnet_strength (opt)	`FLOAT`
controlnet_start (opt)	`FLOAT`
controlnet_end (opt)	`FLOAT`
check_vram (opt)	`COMBO`
use_cache (opt)	`COMBO`
lora_on_error (opt)	`COMBO`

Outputs

Name	Type
MODEL	`MODEL`
CLIP	`CLIP`
VAE	`VAE`
CONTROL_NET	`CONTROL_NET`
LORA_STACK	`LORA_STACK`
STRING	`STRING`
STRING	`STRING`
STRING	`STRING`

◎ Radiance LoRA Stack ▼

RadianceLoraStack

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

How it Works

Accumulates multiple LoRA files and their specific strengths into a single object wrapper that can be passed sequentially.

When to use it

Use to dynamically chain or swap groups of LoRAs cleanly before plugging them into the Unified Loader.

Inputs

Name	Type	Default
lora_stack (opt)	`LORA_STACK`
lora_1 (opt)	`?`
lora_1_model (opt)	`?`
lora_1_clip (opt)	`?`
lora_2 (opt)	`?`
lora_2_model (opt)	`?`
lora_2_clip (opt)	`?`
lora_3 (opt)	`?`
lora_3_model (opt)	`?`
lora_3_clip (opt)	`?`
lora_4 (opt)	`?`
lora_4_model (opt)	`?`
lora_4_clip (opt)	`?`
lora_5 (opt)	`?`
lora_5_model (opt)	`?`
lora_5_clip (opt)	`?`

Outputs

Name	Type
LORA_STACK	`LORA_STACK`

◎ Radiance Cinematic Encoder ▼

RadianceCinematicPromptEncoder

v3.0 — 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.

How it Works

Auto-selects optimal text encoding formats (prose for Flux/T5, structured tags for SD1.5/SDXL) while injecting real-world cinematic terminology (lenses, film stocks, framing).

When to use it

Your primary prompt entry box. Essential for driving AI models to output cinematic, photorealistic compositions instead of default artificial looks.

Inputs

Name	Type	Default
clip	`CLIP`
base_prompt (opt)	`STRING`
base_prompt_b (opt)	`STRING`
active_prompt (opt)	`COMBO`
style_preset (opt)	`STYLE_PRESETS`
target_arch (opt)	`COMBO`
context_window (opt)	`COMBO`
framing (opt)	`FRAMING`
camera_type (opt)	`CAMERAS`
lens_focal (opt)	`LENSES`
aperture_dof (opt)	`APERTURES`
lighting (opt)	`LIGHTING`
style_aesthetic (opt)	`STYLES`
film_stock (opt)	`FILM_STOCKS`
shutter_speed (opt)	`SHUTTER_SPEEDS`
color_grading (opt)	`COLOR_GRADING`
aspect_ratio (opt)	`ASPECT_RATIOS`
scene_mood (opt)	`?`
art_direction (opt)	`STRING`
subject_weight (opt)	`FLOAT`
prompt_weight_mode (opt)	`COMBO`
custom_details (opt)	`STRING`
year_era (opt)	`INT`
negative_strength (opt)	`COMBO`
negative_custom (opt)	`STRING`
clip_skip (opt)	`INT`
lora_keywords (opt)	`STRING`
use_break (opt)	`COMBO`
prompt_enhancer (opt)	`COMBO`
image_ref (opt)	`IMAGE`

Outputs

Name	Type
CONDITIONING	`CONDITIONING`
CONDITIONING	`CONDITIONING`
IMAGE	`IMAGE`
STRING	`STRING`
STRING	`STRING`
INT	`INT`

◎ Radiance Load Image ▼

RadianceLoadImageMask

Advanced Image Loader + Mask Editor for Radiance.

This node loads an image, but also checks for a companion `_radmask.png` file.
If the mask file exists, it overrides the default alpha channel.
The companion frontend extension provides a WebGL-based soft brush mask editor
that saves the non-destructive mask via the ComfyUI API.

How it Works

Searches for a companion `_radmask` file beside the loaded image. If found, it seamlessly binds it to the alpha channel, interacting natively with the frontend WebGL brush editor.

When to use it

Use for quick in-browser masking, inpainting, or rotoscoping without needing to leave the ComfyUI interface for Photoshop.

Inputs

Name	Type	Default
image	`?`

Outputs

Name	Type
IMAGE	`IMAGE`
MASK	`MASK`

◎ Radiance Sampler Pro ▼

RadianceSamplerPro

v4.0 — Universal sampler. External sigma override, 5 noise types, tiled high-res sampling, multi-conditioning merge, CLIP encoder routing, quality-tier presets, latent_meta JSON output, plus all v3.x features: Flux Shift, dynamic guidance, PAG, CFG++, AYS, guidance rescale.

How it Works

An advanced rendering engine unifying Flux Shift, dynamic guidance, PAG, CFG++, and AYS schedulers. It supports external sigma overrides and high-res tiled sampling internally.

When to use it

The core generation node. Replaces basic KSampler with professional controls for optimizing generation speed, coherence, and detail recovery.

Inputs

Name	Type	Default
model	`MODEL`
positive	`CONDITIONING`
negative	`CONDITIONING`
latent_image	`LATENT`
preset	`?`
steps	`INT`
start_step	`INT`
end_step	`INT`
cfg	`FLOAT`
sampler	`SAMPLERS`
sampler_mode	`ALL`
phase_split	`FLOAT`
scheduler	`SCHEDULERS`
scheduler_mode	`COMBO`
denoise	`FLOAT`
flux_shift	`FLOAT`
flux_guidance	`FLOAT`
flux_guidance_profile	`COMBO`
add_noise	`BOOLEAN`
return_with_leftover_noise	`BOOLEAN`
seed	`INT`
pag_scale	`FLOAT`
model_type	`?`
sigma_blend_steps	`INT`
ays_schedule	`BOOLEAN`
guidance_rescale_phi	`FLOAT`
preview_method	`?`
noise_type	`?`
multi_cond_mode	`?`
cond_weight_b	`FLOAT`
conditioning_clip_target	`?`
tile_mode	`BOOLEAN`
tile_size	`INT`
tile_overlap	`INT`
tile_blend	`?`
refiner_model (opt)	`MODEL`
refiner_start_step (opt)	`INT`
noise_override (opt)	`LATENT`
sigmas_override (opt)	`SIGMAS`
latent_format (opt)	`STRING`
positive_2 (opt)	`CONDITIONING`

Outputs

Name	Type
LATENT	`LATENT`
SIGMAS	`SIGMAS`
STRING	`STRING`
STRING	`STRING`

◎ Radiance Manager ▼

RadianceCinemaStudio

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

How it Works

Injects anatomically correct camera metadata (sensor size, focal length, aperture) naturally into the prompt structure using extensive real-world database values.

When to use it

Use to enforce rigid physical camera constraints on generations (e.g., forcing standard anamorphic flares or specific depth-of-field physical responses).

Inputs

Name	Type	Default
base_prompt	`STRING`
camera	`CAMERA_LIST`
lens_series	`LENS_LIST`
focal_length	`FOCAL_LENGTHS`
aperture	`?`
shutter	`?`
iso	`?`
shot_type (opt)	`COMBO`
camera_movement (opt)	`COMBO`

Outputs

Name	Type
STRING	`STRING`
STRING	`STRING`

◎ Radiance Temporal Smooth ▼

RadianceTemporalSmooth

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

How it Works

Applies a per-pixel Exponential Moving Average (EMA) across a batch of video frames, optionally governed by motion analysis to restrict smoothing on moving objects.

When to use it

Crucial for reducing typical AI high-frequency flicker and boiling textures in video-to-video or long sequence generation.

Inputs

Name	Type	Default
images	`IMAGE`
alpha	`FLOAT`
motion_aware	`BOOLEAN`
motion_threshold	`FLOAT`
warmup_frames (opt)	`INT`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

◎ Radiance Flicker Analyze ▼

RadianceFlickerAnalyze

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

How it Works

Computes the frame-to-frame localized luma delta percentage to mathematically quantify the amount of flicker present in a video batch.

When to use it

Use as a QC tool to automatically flag rejected generations if the temporal inconsistency exceeds acceptable professional thresholds.

Inputs

Name	Type	Default
images	`IMAGE`
channel	`COMBO`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

⬆ Upscaling & Restoration

4 nodes

◎ Radiance HDR 360 Generate ▼

HDR360Generate

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

How it Works

Transforms a standard 2D image or equirectangular map into a 360-degree high-dynamic-range spherical environment using spherical harmonic interpolation.

When to use it

Perfect for creating custom IBL (Image Based Lighting) maps from AI generations to light 3D scenes in Unreal Engine or Blender.

Inputs

Name	Type	Default
source_image	`IMAGE`
projection_type	`PROJECTION_TYPES`
output_width	`INT`
output_height	`INT`
horizontal_fov (opt)	`FLOAT`
vertical_fov (opt)	`FLOAT`
rotation_x (opt)	`FLOAT`
rotation_y (opt)	`FLOAT`
rotation_z (opt)	`FLOAT`
interpolation (opt)	`INTERPOLATION_MODES`
fill_mode (opt)	`FILL_MODES`
exposure_adjust (opt)	`FLOAT`

Outputs

Name	Type
IMAGE	`IMAGE`
IMAGE	`IMAGE`

◎ Radiance Resolution ▼

RadianceResolution

Professional Resolution Generator with industry presets:
• Social Media (Instagram, TikTok, YouTube, Twitter, etc.)
• Film & Cinema (4K DCI, 2K, IMAX, Anamorphic scopes)
• Television (8K/4K UHD, HD, SD)
• AI Optimized (SDXL, FLUX)
Outputs empty latent + visual preview.

How it Works

A centralized node supplying mathematically optimal resolutions (divisible by 8, 16, or 64) for AI architectures, categorized by industry standards (Social Media, Cinema 4K/8K, Anamorphic).

When to use it

Use as the root starting dimension for generating Empty Latents. Ensures you never input arbitrary resolutions that cause AI smearing or dimension mismatch errors.

Inputs

Name	Type	Default
preset	`?`
model_type (opt)	`?`
megapixels (opt)	`MEGAPIXEL_TARGETS`
custom_width (opt)	`INT`
custom_height (opt)	`INT`
custom_aspect (opt)	`?`
use_custom (opt)	`BOOLEAN`
divisible_by (opt)	`COMBO`
batch_size (opt)	`INT`
swap_dimensions (opt)	`BOOLEAN`

Outputs

Name	Type
LATENT	`LATENT`
IMAGE	`IMAGE`
INT	`INT`
INT	`INT`
STRING	`STRING`

◎ Radiance Pro Upscale ▼

RadianceProUpscale

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

How it Works

A sophisticated upscaling array optimized specifically for Flux. It breaks the image into overlapping tiles, processes them linearly in high-frequency space, and blends them seamlessly to preserve edge contrast.

When to use it

Use for taking draft/fast generations (e.g. 1024x1024) and blowing them up to 4K delivery resolutions without losing the filmic texture or sharpness.

Inputs

Name	Type	Default
image	`IMAGE`
scale_factor	`FLOAT`
preset	`?`
method (opt)	`?`
sharpening (opt)	`FLOAT`
sharpen_radius (opt)	`FLOAT`
detail_enhancement (opt)	`FLOAT`
antialiasing (opt)	`FLOAT`
input_color_space (opt)	`COMBO`
process_in_linear (opt)	`BOOLEAN`
use_tiles (opt)	`BOOLEAN`
tile_size (opt)	`INT`
tile_overlap (opt)	`INT`
output_bit_depth (opt)	`COMBO`

Outputs

Name	Type
IMAGE	`IMAGE`
INT	`INT`
INT	`INT`
STRING	`STRING`

◎ Radiance Upscale By Size ▼

RadianceUpscaleBySize

Upscale to exact dimensions with aspect ratio control.

How it Works

Standard Lanczos/Bicubic mathematical resizing of the tensor. It offers aspect ratio constraint options ('crop', 'fill', 'fit') for predictable scaling.

When to use it

Use when conforming standard plates or reference images to match the exact mathematical bounds required by an AI model or a specific video timeline.

Inputs

Name	Type	Default
image	`IMAGE`
width	`INT`
height	`INT`
method	`?`
maintain_aspect (opt)	`BOOLEAN`
aspect_mode (opt)	`COMBO`
sharpening (opt)	`FLOAT`
process_in_linear (opt)	`BOOLEAN`
input_color_space (opt)	`COMBO`

Outputs

Name	Type
IMAGE	`IMAGE`
INT	`INT`
INT	`INT`

◎ Radiance Downscale 32-bit ▼

RadianceDownscale32bit

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

How it Works

Employs an anti-aliasing pre-computation filter running natively in linear Float32 space before decimating pixels, preventing moiré and jagged edges.

When to use it

Essential when shrinking extremely high-res outputs (like an 8K render) down to 1080p proxy or delivery files while maintaining soft, natural edges.

Inputs

Name	Type	Default
image	`IMAGE`
scale_factor	`FLOAT`
method	`?`
antialiasing (opt)	`FLOAT`
pre_blur (opt)	`FLOAT`
process_in_linear (opt)	`BOOLEAN`
use_gpu (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`
INT	`INT`
INT	`INT`

◎ Radiance Bit Depth Convert ▼

RadianceBitDepthConvert

Convert between bit depths with professional dithering to reduce banding.

How it Works

Converts between 32-bit float, 16-bit int, and 8-bit int spaces using specialized dithering algorithms (Floyd-Steinberg, Ordered) to prevent the stair-stepping banding that usually haunts gradients.

When to use it

Required step before displaying HDR work in viewers or saving to final compressed 8-bit web formats (JPEG/WebP) to ensure smooth skies and dark shadows.

Inputs

Name	Type	Default
image	`IMAGE`
output_depth	`COMBO`
dithering (opt)	`COMBO`
dither_strength (opt)	`FLOAT`
seed (opt)	`INT`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

◎ Radiance AI Upscale ▼

RadianceAIUpscale

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

How it Works

Leverages neural network ONNX models (e.g. RealESRGAN, 4x-UltraSharp) to hallucinate lost details and synthesize sharp edges where none exist, vastly improving perceived resolution.

When to use it

Use for restoring degraded reference plates, upscaling very low resolution AI outputs, or fixing soft focus problems before final compositing.

Inputs

Name	Type	Default
image	`IMAGE`
model_name	`AI_MODELS`
mode	`COMBO`
tile_size	`INT`
tile_overlap	`INT`
auto_download	`BOOLEAN`
unload_model (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

◎ Viewer & QC Tools

9 nodes

◎ Radiance HDR Histogram ▼

HDRHistogram

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

How it Works

Analyzes the statistical distribution of luminance across the image and maps it to a graphical interface, highlighting any data clipping outside the 0.0-1.0 SDR range in red/blue.

When to use it

Keep this permanently docked on your canvas to monitor exposure levels analytically, ensuring you are not clipping critical shadow or highlight detail.

Inputs

Name	Type	Default
image	`IMAGE`
mode	`MODES`
show_clipping	`BOOLEAN`
stops_range	`INT`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`

◎ Radiance QC Pro ▼

RadianceQC

Production QC with comprehensive checks and structured reporting

How it Works

An automated Quality Control tool that scans the tensor for illegal color values, harsh banding, clipping, and excessive noise based on strict threshold numbers (like Netflix QC standards).

When to use it

Insert into batch video pipelines to automatically flag or reject bad frames from an AI batch run before they get written to the final output.

Inputs

Name	Type	Default
image	`IMAGE`
black_threshold (opt)	`FLOAT`
white_threshold (opt)	`FLOAT`
overlay_opacity (opt)	`FLOAT`
banding_threshold (opt)	`FLOAT`
enable_focus_check (opt)	`BOOLEAN`
enable_artifacts_check (opt)	`BOOLEAN`
enable_noise_check (opt)	`BOOLEAN`
fail_on_errors (opt)	`BOOLEAN`

Outputs

Name	Type
IMAGE	`IMAGE`
STRING	`STRING`
STRING	`STRING`
STRING	`STRING`

◎ Export QC Report ▼

RadianceQCExport

Export QC reports to files for pipeline integration.
Supports JSON, CSV, and HTML formats.

How it Works

Packages the JSON data from `RadianceQC` and writes it as an HTML/CSV/JSON report to the disk containing timestamps, file names, and frame fail statuses.

When to use it

Use for delivering professional reports to clients or producers showing the technical health of the AI video pass.

Inputs

Name	Type	Default
qc_report_json	`STRING`
output_path	`STRING`
filename_prefix	`STRING`
export_format	`COMBO`

Outputs

Name	Type
STRING