Studio42 Layer Composer

Studio42 Layer Composer takes a background and a foreground image and composites them: blend modes, pixel-precise positioning, scaling, rotation around an anchor, flipping, feathering, and a mask input - the standard layer stack you'd expect from an image editor, as a node. It's the natural landing spot after the pack's Background Remover: cut your subject out with a mask, composite it here, and you've got a replacement-background workflow with no Python.

It's not flashy and it's not AI - and that's the point. Sometimes you need to put a foreground on a background exactly, and this node gives you the controls to do it without leaving the graph.

The inputs that matter

• background and foreground - the two images.
• blend_mode - 17 modes including normal, multiply, screen, overlay, soft_light, hard_light, color_dodge, color_burn, difference, and exclusion. soft_light/overlay are the ones for realistic lighting integration.
• opacity - 0–1 global alpha for the foreground.
• foreground_mask - optional MASK; this is how you bring in a cutout from a background-removal node.
• position_mode - custom (uses x/y) or one of nine presets (center, corners, edges).
• x_position / y_position - pixel coordinates from the top-left, ±4000.
• scale_x / scale_y - independent scaling, with maintain_aspect (on by default) to lock them together.
• fit_mode - none, fit_width, fit_height, fit_both, fill, stretch for auto-sizing the foreground.
• rotation - ±360°, around anchor_x/anchor_y (0–1, center 0.5/0.5).
• flip_horizontal / flip_vertical, feather_edges (0–50), anti_aliasing, clip_to_background.

The proxy and batch stuff (actually useful)

• use_proxy_workflow - scales large images down for fast processing, then upscales the result (via upscale_method: bilinear, bicubic, lanczos). On by default, and it's why big composites don't crawl.
• batch_processing_mode - auto, full_batch, or chunked, with max_chunk_size for memory-safety on batches. ComfyUI eats a lot of VRAM on image batches; this is the pack handling that for you.

What comes out

composite_image (IMAGE) and composite_mask (MASK) - the latter being the effective placement mask, handy if you want to inspect or reuse where the foreground landed.

Installing it

Same pack, 24oiduts-ComfyUI:

• ComfyUI Manager: search 24oiduts, install, restart.
• Manual:

  cd ComfyUI/custom_nodes/
  git clone https://github.com/GeekyGhost/24oiduts-ComfyUI
  pip install -r 24oiduts-ComfyUI/requirements.txt

  then restart.

Pillow, numpy, and opencv-python cover it - no model downloads. The GitHub README is a stale Studio42 template; ignore the "WIP" banner.

Troubleshooting

• Composite looks off-center - check position_mode; presets override x/y. Set it to custom for direct coordinate control.
• Foreground scaled weirdly - toggle maintain_aspect and check fit_mode is none if you want raw x/y scaling.
• Slow on huge images - confirm use_proxy_workflow is on; it exists precisely for this.
• Foreground edges visible - feed a feather (feather_edges 3–8) or start from a real mask.

The one you'd actually reach for when a background replacement needs to look intentional rather than pasted.