Multi-image-loader

Multi Image Loader — ComfyUI Custom Node

Drop this folder into ComfyUI/custom_nodes/multi_image_loader/ and restart ComfyUI.

---

Nodes Included

1. Multi Image Loader (2–8)

Load 2 to 8 images at once.

Inputs:

• image_1, image_2 — required
• image_3 … image_8 — optional (set to __none__ to skip)

Outputs:

• image_1 … image_8 — individual tensors (empty slots fall back to image_1)
• image_batch — all loaded images concatenated into one batch tensor
• loaded_count — INT, how many images were actually loaded

Usage with your Qwen workflow: Connect image_1 → image1 slot and image_2 → image2 slot on TextEncodeQwenImageEditPlus.

---

2. Image Select From Batch

Pick a single image out of a batch by index.

Inputs: image_batch (IMAGE), index (INT)
Output: image — the selected frame

---

3. Image Batch Builder

Chain multiple images into a growing batch progressively.

Inputs: image_new (required), image_batch (optional — if absent, starts new batch)
Outputs: image_batch, batch_size

Use this to chain: LoadImage → BatchBuilder → LoadImage → BatchBuilder → ...

---

4. Qwen Multi-Ref Picker

From a batch, pick two images by index to wire into image1 / image2 on TextEncodeQwenImageEditPlus.

Inputs: prompt, clip, vae, image_batch, ref_index_1, ref_index_2
Outputs: selected_image1, selected_image2

---

Recommended Setup for Qwen Story Workflow

[Multi Image Loader (2–8)]
  image_1 ──────────────────────────────► TextEncodeQwenImageEditPlus.image1
  image_2 ──────────────────────────────► TextEncodeQwenImageEditPlus.image2
  image_3 → [Image Select From Batch]  (route as needed per scene)

Or use Image Batch Builder to build a batch, then Qwen Multi-Ref Picker to swap which two references are active per scene.

---

Installation

ComfyUI/
└── custom_nodes/
    └── multi_image_loader/        ← this folder
        ├── __init__.py
        ├── multi_image_loader.py
        └── README.md

No extra pip dependencies required — uses only torch, numpy, PIL (all standard in ComfyUI).

Multi-image-loader