Skip to content
Isekai Documentation

Image Manipulation Nodes

Isekai provides 22 professional-grade image manipulation nodes inspired by Blender’s compositing system, plus cloud upload capabilities. All image nodes use only PIL/Pillow for zero additional dependencies.

Quick Navigation

Section titled “Quick Navigation”
  • Blend Nodes - Color adjustment, blending, levels, color grading (5 nodes)
  • Effects Nodes - Blur, grain, vignette, filters, distortion (11 nodes)
  • Transform Nodes - Rotate, scale, crop, flip, translate (6 nodes)
  • Upload & Save - Cloud upload and local compression (2 nodes)

Blend Nodes

Section titled “Blend Nodes”

Blend Images

Section titled “Blend Images”

Mix two images with professional blend modes.

Location: Isekai/Image/Blend

Inputs:

  • image_a (IMAGE): Base image
  • image_b (IMAGE): Blend image (auto-resized to match base)
  • blend_mode (COMBO): Normal, Multiply, Screen, Add, Subtract, Difference, Lighten, Darken
  • opacity (FLOAT, 0.0-1.0): Blend strength (default: 0.5)

Outputs:

  • image (IMAGE): Blended result

Example Use Cases:

  • Add texture overlays (Multiply mode)
  • Create light effects (Screen mode)
  • Combine render passes (Add mode)

Color Adjust

Section titled “Color Adjust”

Adjust brightness, contrast, saturation, and sharpness.

Location: Isekai/Image/Blend

Inputs:

  • image (IMAGE): Input image
  • brightness (FLOAT, 0.0-2.0): Brightness multiplier (default: 1.0)
  • contrast (FLOAT, 0.0-2.0): Contrast multiplier (default: 1.0)
  • saturation (FLOAT, 0.0-2.0): Color saturation (default: 1.0)
  • sharpness (FLOAT, 0.0-2.0): Sharpness multiplier (default: 1.0)

Outputs:

  • image (IMAGE): Adjusted image

Tips:

  • Values below 1.0 decrease the effect
  • Values above 1.0 increase the effect
  • Stack multiple adjustments by chaining nodes

Color Ramp

Section titled “Color Ramp”

Apply gradient mapping for color grading.

Location: Isekai/Image/Blend

Inputs:

  • image (IMAGE): Input image
  • color_start (STRING): Hex color for shadows (default: “#000000”)
  • color_end (STRING): Hex color for highlights (default: “#FFFFFF”)
  • intensity (FLOAT, 0.0-1.0): Effect strength (default: 1.0)

Outputs:

  • image (IMAGE): Color-graded result

Use Cases:

  • Create cinematic color grades
  • Apply temperature shifts (warm/cool)
  • Stylistic color treatments

Levels

Section titled “Levels”

Adjust black point, white point, and midtones.

Location: Isekai/Image/Blend

Inputs:

  • image (IMAGE): Input image
  • black_point (INT, 0-255): Input black level (default: 0)
  • white_point (INT, 0-255): Input white level (default: 255)
  • midtones (FLOAT, 0.1-10.0): Gamma adjustment (default: 1.0)

Outputs:

  • image (IMAGE): Level-adjusted image

Tips:

  • Increase black_point to crush blacks
  • Decrease white_point to bring down highlights
  • Adjust midtones > 1.0 to brighten, < 1.0 to darken

Split Toning

Section titled “Split Toning”

Apply separate colors to highlights and shadows.

Location: Isekai/Image/Blend

Inputs:

  • image (IMAGE): Input image
  • highlight_color (STRING): Hex color for bright areas (default: “#FFD700”)
  • shadow_color (STRING): Hex color for dark areas (default: “#1E3A8A”)
  • balance (FLOAT, 0.0-1.0): Highlight/shadow balance (default: 0.5)
  • intensity (FLOAT, 0.0-1.0): Effect strength (default: 0.5)

Outputs:

  • image (IMAGE): Split-toned result

Popular Presets:

  • Golden hour: #FFD700 (highlights) / #1E3A8A (shadows)
  • Cinematic teal/orange: #FF8C42 / #008B8B
  • Cold/warm: #87CEEB / #FF6347

Effects Nodes

Section titled “Effects Nodes”

Blur

Section titled “Blur”

Apply Gaussian or Box blur.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image
  • blur_type (COMBO): Gaussian, Box
  • radius (FLOAT, 0.0-100.0): Blur radius (default: 5.0)

Outputs:

  • image (IMAGE): Blurred result

Use Cases:

  • Depth of field simulation
  • Soft focus portraits
  • Background defocus

Chromatic Aberration

Section titled “Chromatic Aberration”

RGB channel offset for lens distortion effect.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image
  • offset (INT, -50 to 50): Pixel shift amount (default: 5)
  • direction (COMBO): Horizontal, Vertical, Both

Outputs:

  • image (IMAGE): Aberrated result

Color Filter

Section titled “Color Filter”

Apply Sepia, Grayscale, or Black & White filters.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image
  • filter_type (COMBO): Sepia, Grayscale, Black & White, None
  • intensity (FLOAT, 0.0-1.0): Blend with original (default: 1.0)

Outputs:

  • image (IMAGE): Filtered result

Filter Modes:

  • Sepia: Vintage warm tone
  • Grayscale: Full desaturation
  • Black & White: High contrast monochrome
  • None: Pass-through (no effect)

Edge Enhance

Section titled “Edge Enhance”

Enhance or detect edges in images.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image
  • mode (COMBO): Enhance, Enhance More, Find Edges
  • intensity (FLOAT, 0.0-1.0): Effect strength (default: 1.0)

Outputs:

  • image (IMAGE): Edge-enhanced result

Use Cases:

  • Sharpen detail
  • Create line art from photos
  • Enhance texture definition

Glare

Section titled “Glare”

Add bloom/glow to bright areas.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image
  • threshold (INT, 0-255): Brightness threshold for glow (default: 200)
  • intensity (FLOAT, 0.0-2.0): Glow strength (default: 0.5)
  • radius (FLOAT, 1.0-50.0): Glow spread (default: 10.0)

Outputs:

  • image (IMAGE): Result with glare

Tips:

  • Lower threshold = more areas glow
  • Higher intensity = stronger effect
  • Larger radius = softer, wider glow

Grain

Section titled “Grain”

Add film grain or noise texture.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image
  • intensity (FLOAT, 0.0-1.0): Grain amount (default: 0.1)
  • grain_type (COMBO): Gaussian, Uniform, Salt & Pepper
  • monochrome (BOOLEAN): Same grain for all channels (default: True)

Outputs:

  • image (IMAGE): Grainy result

Grain Types:

  • Gaussian: Natural film grain
  • Uniform: Even noise distribution
  • Salt & Pepper: High contrast speckles

Invert

Section titled “Invert”

Invert all image colors.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image

Outputs:

  • image (IMAGE): Inverted result

Simple but useful for:

  • Creating negative effects
  • Inverting masks
  • Artistic treatments

Pixelate

Section titled “Pixelate”

Create mosaic/pixel art effects.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image
  • pixel_size (INT, 1-100): Size of pixels (default: 10)
  • sampling (COMBO): Nearest, Bilinear

Outputs:

  • image (IMAGE): Pixelated result

Tips:

  • Nearest: Sharp pixel edges (true pixel art)
  • Bilinear: Smooth pixel edges (soft mosaic)

Posterize

Section titled “Posterize”

Reduce color levels for poster effect.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image
  • bits (INT, 1-8): Bits per channel (default: 4)

Outputs:

  • image (IMAGE): Posterized result

Bit Levels:

  • 1 bit = 2 colors per channel (8 total colors)
  • 4 bits = 16 colors per channel (4096 total)
  • 8 bits = 256 colors per channel (16.7M total - no effect)

Sharpen

Section titled “Sharpen”

Sharpen images with Unsharp Mask.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image
  • method (COMBO): Sharpen, Unsharp Mask
  • radius (FLOAT, 0.0-10.0): Mask radius (default: 2.0)
  • percent (INT, 0-500): Sharpening strength (default: 150)
  • threshold (INT, 0-255): Edge detection threshold (default: 3)

Outputs:

  • image (IMAGE): Sharpened result

Method Differences:

  • Sharpen: Simple sharpening filter
  • Unsharp Mask: Advanced with radius, percent, threshold control

Vignette

Section titled “Vignette”

Darken edges with radial gradient.

Location: Isekai/Image/Effects

Inputs:

  • image (IMAGE): Input image
  • intensity (FLOAT, 0.0-1.0): Darkness amount (default: 0.5)
  • radius (FLOAT, 0.0-1.0): Effect start point (default: 0.8)
  • softness (FLOAT, 0.0-1.0): Edge falloff (default: 0.5)

Outputs:

  • image (IMAGE): Vignetted result

Tips:

  • Radius 1.0 = vignette starts at edges
  • Radius 0.5 = vignette starts halfway
  • Higher softness = smoother gradient

Transform Nodes

Section titled “Transform Nodes”

Crop

Section titled “Crop”

Crop images to specified region.

Location: Isekai/Image/Transform

Inputs:

  • image (IMAGE): Input image
  • x (INT): Left edge position (default: 0)
  • y (INT): Top edge position (default: 0)
  • width (INT): Crop width (default: 512)
  • height (INT): Crop height (default: 512)
  • from_center (BOOLEAN): Use x,y as center point (default: False)

Outputs:

  • image (IMAGE): Cropped result

Flip

Section titled “Flip”

Flip images horizontally, vertically, or both.

Location: Isekai/Image/Transform

Inputs:

  • image (IMAGE): Input image
  • direction (COMBO): Horizontal, Vertical, Both

Outputs:

  • image (IMAGE): Flipped result

Rotate

Section titled “Rotate”

Rotate images by any angle.

Location: Isekai/Image/Transform

Inputs:

  • image (IMAGE): Input image
  • angle (FLOAT, -360 to 360): Rotation angle in degrees (default: 0)
  • expand (BOOLEAN): Expand canvas to fit rotated image (default: False)
  • fill_color (STRING): Background color hex code (default: “#000000”)
  • resample (COMBO): Nearest, Bilinear, Bicubic

Outputs:

  • image (IMAGE): Rotated result

Resample Quality:

  • Nearest: Fastest, pixelated
  • Bilinear: Good balance
  • Bicubic: Highest quality, slower

Scale

Section titled “Scale”

Resize images with quality control.

Location: Isekai/Image/Transform

Inputs:

  • image (IMAGE): Input image
  • scale_method (COMBO): Factor, Dimensions, Percentage
  • scale_x (FLOAT, 0.01-10.0): Horizontal scale (default: 1.0)
  • scale_y (FLOAT, 0.01-10.0): Vertical scale (default: 1.0)
  • width (INT, 1-8192): Target width if using Dimensions (default: 512)
  • height (INT, 1-8192): Target height if using Dimensions (default: 512)
  • resampling (COMBO): Nearest, Bilinear, Bicubic, Lanczos
  • maintain_aspect (BOOLEAN): Keep aspect ratio (default: True)

Outputs:

  • image (IMAGE): Scaled result

Scale Methods:

  • Factor: Multiply by scale_x and scale_y
  • Dimensions: Set exact width/height
  • Percentage: Scale_x/scale_y as percentages

Resampling:

  • Lanczos: Best quality for downscaling
  • Bicubic: Best quality for upscaling
  • Bilinear: Good general purpose
  • Nearest: Pixel art style

Transform

Section titled “Transform”

Combined rotate, scale, and translate.

Location: Isekai/Image/Transform

Inputs:

  • image (IMAGE): Input image
  • rotate (FLOAT, -360 to 360): Rotation angle (default: 0)
  • scale (FLOAT, 0.1-10.0): Uniform scale factor (default: 1.0)
  • translate_x (INT, -2000 to 2000): Horizontal shift (default: 0)
  • translate_y (INT, -2000 to 2000): Vertical shift (default: 0)
  • resample (COMBO): Nearest, Bilinear, Bicubic

Outputs:

  • image (IMAGE): Transformed result

Use Case:

  • Apply multiple transforms in one node for efficiency

Translate

Section titled “Translate”

Shift images horizontally or vertically.

Location: Isekai/Image/Transform

Inputs:

  • image (IMAGE): Input image
  • x (INT, -2000 to 2000): Horizontal shift in pixels (default: 0)
  • y (INT, -2000 to 2000): Vertical shift in pixels (default: 0)
  • fill_color (STRING): Background color hex code (default: “#000000”)

Outputs:

  • image (IMAGE): Translated result

Tips:

  • Positive X = shift right
  • Negative X = shift left
  • Positive Y = shift down
  • Negative Y = shift up

Upload and Save

Section titled “Upload and Save”

Isekai Upload

Section titled “Isekai Upload”

Upload generated images directly to Isekai platform with compression.

Location: Isekai/Upload

Inputs:

  • image (IMAGE): Image to upload
  • title (STRING): Artwork title (max 200 characters)
  • api_key (STRING, optional): Isekai API key (uses ISEKAI_API_KEY env var if empty)
  • tags (STRING, optional): Comma-separated tags
  • format (COMBO): JPEG, PNG (default: JPEG)
  • quality (INT, 1-100): Compression quality (default: 90)

Outputs:

  • image (IMAGE): Pass-through for preview

Compression:

  • JPEG quality 90: ~80-90% size reduction, excellent quality
  • JPEG quality 75-85: ~85-92% size reduction, very good quality
  • PNG: Lossless, ~20-30% compression

Compress and Save

Section titled “Compress and Save”

Save compressed images to disk with format control.

Location: Isekai/IO

Inputs:

  • images (IMAGE): Images to save
  • filename (STRING): Base filename (default: “isekai”)
  • format (COMBO): JPEG, PNG, WEBP (default: JPEG)
  • quality (INT, 1-100): Compression quality (default: 90)

Outputs: None (saves to ComfyUI/output/ folder)

Filename Pattern:

{filename}_{counter}.{extension}


Examples:
- "artwork" → artwork_00001.jpg
- "render" → render_00001.webp

Format Tips:

  • PNG: Lossless, supports transparency
  • JPEG: Lossy, smaller files, no transparency
  • WEBP: Modern format, excellent compression

S3 Upload

Section titled “S3 Upload”

Upload images to AWS S3 or Cloudflare R2.

Location: Isekai/Upload

Inputs:

  • image (IMAGE): Image to upload
  • bucket_name (STRING): S3 bucket name
  • object_key (STRING): File path in bucket (e.g., “renders/image.jpg”)
  • endpoint_url (STRING, optional): Custom endpoint for R2/MinIO
  • region (STRING): AWS region (default: “us-east-1”)
  • access_key_id (STRING, optional): AWS access key (uses env var if empty)
  • secret_access_key (STRING, optional): AWS secret key (uses env var if empty)
  • format (COMBO): JPEG, PNG, WEBP
  • quality (INT, 1-100): Compression quality (default: 90)

Outputs:

  • image (IMAGE): Pass-through
  • url (STRING): Public URL of uploaded image

Environment Variables:

  • AWS_ACCESS_KEY_ID
  • AWS_SECRET_ACCESS_KEY
  • AWS_DEFAULT_REGION

Cloudflare R2 Example:

endpoint_url: https://<account-id>.r2.cloudflarestorage.com
bucket_name: my-bucket
region: auto

Example Workflows

Section titled “Example Workflows”

Cinematic Color Grade

Section titled “Cinematic Color Grade”
Image → Color Adjust (contrast 1.2) → Split Toning (teal/orange) → Vignette → Output

Vintage Photo Effect

Section titled “Vintage Photo Effect”
Image → Color Filter (Sepia, 0.8) → Grain (Gaussian, 0.15) → Levels (crush blacks) → Output

Soft Portrait

Section titled “Soft Portrait”
Image → Blur (Gaussian, 3.0) → Blend (original, Screen, 0.3) → Vignette → Output

Stylized Render

Section titled “Stylized Render”
Image → Posterize (4 bits) → Edge Enhance → Color Ramp (creative grade) → Output