Ino Load Images From Folder

Native ComfyUI can load one image at a time. If you want to batch-process a whole folder - img2img over 400 source shots, upscale an entire batch, feed references into a control pipeline - you either install something like this or you write a Python script. InoLoadImagesFromFolder is the batch loader: point it at a folder, and it hands you the images and their masks as lists, plus the paths, plus a count. The skip_from_first and load_cap inputs are what make it genuinely useful: you can process the first ten, or everything after the first twenty, without touching the folder.

It sits at the front of the batch patterns the Ino pack is built around. The images output is a list, which wires straight into Ino List Get Item or Ino On Image List Completed; the masks come along for free, which matters when your batch pipeline needs alpha or segmentation masks.

How it works

Required inputs:

• parent_folder - a dropdown: input, output, or temp. This is the base directory (ComfyUI's standard folders), and folder is a sub-path under it. If you drop images into ComfyUI's input folder, you pick input and give the subfolder name.
• folder - the subfolder path under parent_folder.
• skip_from_first - how many images to skip from the start (0 = start at the first).
• load_cap - maximum number to load (0 = no cap).

Outputs: success, message (e.g. "Loaded 23 images"), rel_path / abs_path, then the good stuff: images (a list of IMAGE tensors), masks (a list of MASK tensors), and number of images (INT). Loading honors EXIF rotation the same way native ComfyUI does, so phone photos come in upright.

One behavior worth knowing: if the folder is empty or nothing loads, it doesn't fail loudly - it returns success: False, a "No images found" message, and a single empty 512×512 placeholder image with a 64×64 placeholder mask. That's deliberate, so downstream nodes keep running and you can branch on success instead of the whole graph erroring.

Installing it

Part of the ComfyUI-InoNodes pack by nobandegani:

cd ComfyUI/custom_nodes
git clone https://github.com/nobandegani/comfyui_ino_nodes
cd comfyui_ino_nodes
pip install -r requirements.txt

ComfyUI Manager users: search "ComfyUI Ino Nodes", install, restart. The pack uses the V3 schema, so a reasonably current ComfyUI is required.

Common issues

The most common mistake is putting the images in the wrong place relative to parent_folder. input maps to ComfyUI/input/, and folder is relative to that - input + batch/ means ComfyUI/input/batch/. If success comes back False with "No images found," that's the first thing to check. Also remember the cap is a load limit, not a filter: with load_cap: 0 you get everything, and skip_from_first and load_cap combine (skip first, then cap). And don't feed the placeholder output into a sampler expecting a real image - gate your downstream on success.