BMAB Base64 Image

Most of the time you get images into a ComfyUI graph by loading a file. Sometimes that's not the shape your data comes in - you're receiving an image inline as text, from an API call, a webhook payload, or a script that never touched disk. That's the gap this node fills.

What it is and why you'd reach for it

BMAB Base64 Image decodes a base64-encoded image string directly into a usable IMAGE tensor in your graph, no file upload required. If you're building or debugging a pipeline where images arrive as text - say, a JSON payload from an external service, or a base64 blob you're pasting in while testing - this node is the entry point instead of Load Image.

That's a genuinely useful shape if you're running ComfyUI workflows through an API rather than the web UI: passing image data inline as a string sidesteps needing a separate upload step, at the cost of a bulkier payload. If your workflow is purely interactive in the browser, you probably won't reach for this over just dragging a file in.

How it works

You paste (or programmatically inject) a base64-encoded image string into the encoding field, and the node decodes it back into pixels, reporting the resulting image's width and height alongside it. Decoding a base64 string is exact and lossless with respect to whatever image format was encoded - you get back precisely what was encoded, not a re-compressed approximation.

Inputs and outputs

• encoding (required, multiline string) - the base64-encoded image data. This is the field that actually drives the node.
• image (required, dropdown of files from your ComfyUI input folder) - a standard file-picker widget that ComfyUI attaches to image-handling nodes by convention. If you're relying on this node specifically for its base64 decoding, the encoding field is what matters; don't be surprised the node also carries the usual file dropdown alongside it.

Outputs:

• image - the decoded IMAGE, ready to wire anywhere a normal loaded image would go.
• width / height - the decoded image's dimensions as plain integers, handy if a downstream node needs them explicitly (a resize node, a crop calculation, or just a sanity check that what came through matches what you expected).

Installing it

Through ComfyUI Manager: search comfyui_bmab, install, and restart. Manual install:

cd ComfyUI/custom_nodes
git clone https://github.com/portu-sim/comfyui_bmab.git
cd comfyui_bmab
pip install -r requirements.txt

(Windows portable users: swap in python_embeded\python.exe -m pip install -r requirements.txt.) This node is plain Python string/image handling - no GPU model, no extra weights, and it doesn't depend on BMAB's two optional companion packs (comfyui_controlnet_aux, ComfyUI_IPAdapter_plus). The base install covers it.

Common issues

Base64 decoding failures are almost always about the string itself, not the node:

• A data URL prefix left on the string. If your base64 came from a browser or a <img src="data:image/png;base64,..."> context, make sure you're pasting only the part after the comma - the data:image/png;base64, prefix isn't valid base64 and will break decoding if it's still attached.
• Truncated or whitespace-mangled strings. Copy-pasting a very long base64 string through some terminals or text fields can introduce line breaks or drop characters. If decoding fails outright, try re-copying the string as a single unbroken block.
• Wrong image format assumptions downstream. The node reports width/height from whatever was actually decoded - if those numbers look wrong, the issue is almost certainly the source string, not this node misreading it.