Infotext To Image Info

Every old-school Civitai image is a recipe waiting to be decoded. The A1111 infotext string - that wall of Prompt: ... Steps: 28, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 12345, Model: ... - is the entire generation settings in plain text. IPT-InfotextToImageInfo takes exactly that string and converts it into this pack's image_info bundle, so pasted metadata becomes data your workflow can actually use.

It's the text-version companion to Image Reader, which does the same job from a file on disk. Use this one when the recipe lives outside a file - you copied the infotext from a Civitai page, someone pasted it in a Discord channel, or you have a raw string from a prompt database and want to load it into your graph without saving an image first.

How it works

It parses the infotext using the pack's A1111 infotext utility, which splits out the positive and negative prompts (including <lora:...> tags), the model reference, sampler, scheduler, steps, CFG, seed, and dimensions. Then it normalizes the result against ComfyUI's current options - so a sampler name that's valid in A1111 gets mapped to the matching ComfyUI sampler, and the LoRA references are normalized to the pack's lora_stack format. The single output, image_info, is exactly the bundle Image Saver writes into metadata and Image Reader reconstructs, so everything stays on the same format.

Inputs and outputs

• infotext (required, STRING) - the A1111 infotext string. It's a forced-input field, so it can be typed, but it really wants to be wired from something - a string node, a captured output, or a bit of your own plumbing.
• image_info (output, IPT-IMAGEINFO) - the reconstructed settings bundle. Wire it into Image Info Context, Image Info Defaults, a saver, or an Image Info Fallback to fill gaps in an existing setup.

Installing it

Same install as the rest of the pack:

cd ComfyUI/custom_nodes
git clone https://github.com/kinorax/comfyui-info-prompt-toolkit.git
cd comfyui-info-prompt-toolkit
pip install -r requirements.txt

Restart ComfyUI, or search "ComfyUI-Info-Prompt-Toolkit" in ComfyUI Manager. Needs ComfyUI 0.17.0+. No models, no heavy dependencies for this node.

Common issues

The infotext formats are not perfectly standardized across tools. A1111 writes its own flavor; some Civitai uploads have extra fields or slightly different ordering; Forge and newer tools add their own keys. The parser handles the common case well but a heavily custom infotext can leave fields empty - check the resulting image_info instead of assuming everything mapped. Also, this expects the A1111 text-parameter format, not ComfyUI's embedded-workflow PNG metadata and not a bare prompt. Paste the wrong thing and you'll get an image_info with the prompt filled in and the sampler settings missing, which is still better than nothing but won't reproduce the generation faithfully. When in doubt, pair it with an Image Info Fallback so missing fields get filled from your current defaults instead of staying blank.