DP Load Image With Seed

Most image loaders give you pixels and stop there. This one goes further: it loads the image, applies up to two on-the-fly effects, resizes it, and surfaces the generation metadata baked into the file - the prompt, a negative field, the seed, and the filename. If you're pulling old generations back into a workflow to remix them, this saves you from copy-pasting settings out of the PNG by hand.

It's part of DesertPixelAi's Desert Pixel pack, which has a whole shelf of enhanced loaders. This is the one that leans on embedded metadata.

How it works

You pick a file from the image dropdown (your ComfyUI input folder), and the node does a few things at once. It reads the workflow/prompt data stored in the PNG and exposes it as text outputs, it optionally resizes to your width/height, and it runs your chosen effects over the pixels.

The effect_A and effect_B dropdowns each offer 24 options - original, grayscale, enhance, flip_h, flip_v, rotate_180, posterize, sharpen, contrast, equalize, and more - and effect_strength (0–1) controls how hard they hit. Because there are two independent effect slots with their own outputs, you get two processed versions of the same source in one node.

The inputs and outputs that matter

Realistically you set: the image itself, resize_image plus width/height if you need a specific size, and one or both effect dropdowns if you want the processed variants (leave them on original to pass the image through untouched).

Six outputs, and the back half is the point:

• image_A and image_B - the two effect-processed versions.
• filename (STRING) - the source name, handy for re-saving.
• dp_prompt and dp_negative_or_other (STRING) - the prompt and negative recovered from the image's metadata. Wire dp_prompt into a CLIP Text Encode to regenerate from the original prompt.
• seed (INT) - the seed pulled from the file, straight into your sampler to reproduce the original.

How to install it

ComfyUI Manager: search Desert Pixel, install, restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/DesertPixelAi/ComfyUI-Desert-Pixel-Nodes

Restart ComfyUI. No downloads needed. Nodes appear under the DP category.

Common issues & troubleshooting

The prompt and seed outputs come back empty. They can only surface metadata that's actually in the file. An image that wasn't generated in ComfyUI (or was stripped of its metadata by an editor, a re-save, or an upload/download round-trip through a site that flattens PNGs) has nothing to read. JPEGs and screenshots won't carry the workflow either. If you need those outputs, load an original, unedited ComfyUI PNG.

The image dropdown doesn't list your file. It reads the ComfyUI input folder - drop the PNG in ComfyUI/input (or upload it through the node) and refresh.

The effect outputs look wrong or too strong. Both image_A and image_B run through the effect pipeline; set either dropdown to original to pass the image through clean, and ease effect_strength down from 1.0 if an effect is overcooked.