AIO Model Info

AIO Image Generate deliberately hides its internals. The model, the CLIP, the conditioning, the VAE - all of that happens inside one node, and the pack bundles those values into a single opaque model_info output instead of wiring them around. That keeps the graph clean right up until you need one of those values for something the main node doesn't do. AIOModelInfo is the unpacker: feed it model_info, and it hands back the standard ComfyUI MODEL, CLIP, CONDITIONING (positive and negative), and VAE values as regular typed outputs.

What it gives you

One required input, model_info (from AIO Image Generate's model_info output). Five outputs:

• model - the loaded (and LoRA-patched) diffusion model.
• clip - the text encoder/CLIP as ComfyUI uses it.
• positive / negative - the encoded conditioning, if you need the pack's own encodings elsewhere (e.g. to feed a second sampler that should use the exact same conditioning the main run used).
• vae - the VAE used for decoding.

A neat lazy-loading detail from the source: the heavy fields (model, clip) are only populated when their matching outputs here are actually connected - so you can wire up just positive and negative without forcing the pack to hand you full models you don't need. Connect what you use; the pack materializes only that.

When you'd actually use it

This is an interoperability node, not a daily driver. Reach for it when:

• You want the main node's exact conditioning to drive a second pass or a side-branch sampler.
• You need the VAE for a VAE-side operation (encode/decode round-trip, VAE inspection) and want to be sure it's the same VAE the main run used.
• You're bridging to a custom node that takes MODEL/CLIP/CONDITIONING sockets and want it to see precisely what the facade loaded.

Note it's not the same role as AIO Load Pipeline Models. That node loads fresh models for you to patch and feed back in; this one exposes the models already inside a completed run. If you're after patching, use the loader.

Install and wiring

It ships with the pack:

cd ComfyUI/custom_nodes
git clone https://github.com/helto4real/comfyui-all-on-one-image-generation-node
cd comfyui-all-on-one-image-generation-node
python -m pip install -r requirements.txt

Or ComfyUI Manager → search "AIO Image Generate" → install → restart; it's under AIO/Image. Connect AIO Image Generate.model_info → AIOModelInfo.model_info, then wire whichever outputs you need.

Gotchas

• Old workflows break here. This pack replaced the main node's old direct model/CLIP/conditioning/VAE outputs with the model_info bundle. Load a pre-migration workflow that wired those old sockets and it will fail - reconnect through this node.
• Empty outputs are meaningful. If a field comes back empty, either you didn't connect that output (lazy population) or the graph couldn't prove which field a downstream consumer needs. Connect the specific outputs you're using.

It's the plumbing layer doing its job: one opaque bundle in, typed sockets out, nothing materialized until you ask.