Load64image☀

Sometimes your image doesn't live on disk - it lives in a base64 string. Maybe it came out of an API response, maybe you copied it from a browser's dev tools, maybe your whole pipeline is string-based. ComfyUI has no native way to get from "a long base64 blob" to "an IMAGE tensor," and that's the gap Load64image fills. It's a tiny node with a single input and a single output, and there's genuinely nothing else to it.

How it works

You paste the base64 text into the base64_image field (multiline, so no paste-size limits) and it decodes, opens the bytes as an image with Pillow, converts to RGB, and returns a standard ComfyUI IMAGE tensor. Two niceties are worth knowing:

• It strips a data:image/...;base64, prefix if present, so raw data URIs from APIs work as-is - no manual editing.
• It's tolerant of whitespace and newlines in the blob, which you'll appreciate the first time you paste a wrapped base64 string that some tool line-wrapped.

Inputs and outputs

• base64_image - the string. That's the only input.
• image - the decoded IMAGE tensor, ready to wire into anything.

When you'd actually reach for it

Honestly, this is a niche node, and that's fine. The use cases are real but specific:

• You're working with an API that returns images as base64 (which is exactly the shape the API nodes in this pack deal in).
• You're building a string-driven workflow where the image enters through a text field.
• You want to test a decode path without touching the file system.

Don't reach for it to load normal images - LoadImage and this pack's LoadImageList are better for files. This one exists for the string-shaped cases.

Install

It's part of DashuaiTools:

cd ComfyUI/custom_nodes
git clone https://github.com/Hasasasa/ComfyUI_DashuaiTools

then restart ComfyUI, or grab it through ComfyUI Manager ("DashuaiTools"). Dependencies: just Pillow, numpy, and torch, all already in ComfyUI. If the field is empty or the data is garbage, you get a clear base64_image is empty / invalid image data error rather than a silent failure - which is more than some nodes manage.