Skip to content

Layer Add

Adds one layer to a layer stack, the way you would drop a new layer into Photoshop: pick a blend mode, set opacity and fill, optionally add a mask, and optionally clip it to the layer below.

Layer Add — before / after
In ComfyUI Add Node → ComfyUI-JH-PixelPro → compositing → Compositing: Layer Add
Good for Colour grades, light leaks, dodge-and-burn layers, texture overlays, local tints through a mask

When to use it

Use it for finishing composites that should be repeatable and adjustable. Each Layer Add node is one layer. Chain several to build a stack from bottom to top: the first one you add sits directly above the background, the next one above that, and so on.

How the stack works:

  1. Layer Stack Start creates the background layer.
  2. Layer Add appends layers on top, each with its own blend mode, opacity, fill, mask and clipping.
  3. Layer Group can nest a whole sub-stack as a single layer.
  4. Layer Flatten merges everything into one image.

Quick start

  1. Load Image → Layer Stack Start.
  2. Stack output → Layer Add stack. Plug the image you want on top into layer_image.
  3. Try blend_mode = soft_light and opacity = 0.5.
  4. Add more Layer Add nodes if you need them, then finish with Layer Flatten → Preview Image.

Settings

Setting Default What it does Tip
stack — The stack to add to. From Layer Stack Start or another Layer Add.
layer_image — The image to place on top. Resized to the background size if it differs.
blend_mode normal How this layer mixes with everything below it. See the list below.
opacity 1.0 (0 to 1) Overall strength of the layer. Start at 0.3 to 0.6 for grades.
fill 1.0 (0 to 1) A second strength control. Multiplies with opacity.
clip_to_below off Limits the layer to the mask of the layer directly below. Only has an effect if that layer has a layer_mask.
layer_mask (optional) none Where the layer shows: white = fully, black = not at all. Resized to the canvas if needed.

Output: LAYER_STACK, the stack with your new layer on top.

Blend modes

All 27 modes in core/blend_modes.py, grouped like the Photoshop menu:

Group Modes
Normal normal, dissolve
Darken darken, multiply, color_burn, linear_burn, darker_color
Lighten lighten, screen, color_dodge, linear_dodge, lighter_color
Contrast overlay, soft_light, hard_light, vivid_light, linear_light, pin_light, hard_mix
Comparative difference, exclusion, subtract, divide
Colour hue, saturation, color, luminosity

dissolve works like Photoshop: each pixel becomes either fully the layer or fully the base, with a chance equal to opacity × fill × mask. It does not blend colours. The pattern uses a fixed seed, so the same settings always give the same result.

Tips from retouching

  • Soft Light or Overlay at low opacity is the quickest way to add contrast or a colour wash without clipping.
  • Multiply darkens, Screen lightens: use them with a blurred mask for dodge-and-burn style work.
  • Color applies only the hue and saturation of the layer, which is handy for tinting skin without changing brightness. Luminosity does the opposite.
  • Feed a mask from the Mask nodes (for example Luminosity Masking) into layer_mask to limit a grade to the highlights or shadows.
  • Use clip_to_below to keep a layer inside the masked area of the layer under it.
Under the hood

Layers are rendered in order. The blend result is computed from the running image and the layer image (both clamped to 0..1), then mixed back with lerp using alpha = mask × opacity × fill, so opacity and fill behave the same here. dissolve is the exception: alpha is used as the probability that a pixel is taken fully from the layer, using a fixed-seed random value per pixel. A missing mask counts as all white. With clip_to_below, the layer's mask is multiplied by the mask of the layer below; a layer below without a mask counts as white, so clipping changes nothing in that case. Layer images and masks are resized bilinearly to the background size. The hue, saturation, color and luminosity modes are computed in HSV, so they can differ slightly from Photoshop's versions. Source: nodes/layer_compositing.py and core/blend_modes.py.

Example workflow

Copy workflows/sample_portrait.jpg into ComfyUI/input/ first.