Head-Swap LoRA Debug

This is one of the smallest nodes in the pack, and honestly, that's the whole point. BFSHeadSwapLoRADebug takes a model and hands back the same model plus a debug string - a pass-through with a printout attached, the same pattern as the debug outputs bolted onto nearly every other conditioning node in this pack. It has zero GSC impressions and no README coverage as of this writing, so what follows is a straight read of the schema rather than a description the author wrote - take it as a reasonable inference, not a confirmed spec.

Given the name and where it sits in the pack (right alongside BFSHeadSwapBerniniConditioning), the sensible read is that this is a diagnostic tap you insert right after loading your head-swap LoRA onto a model, before you run a full sampling pass. Every conditioning node in this identity family depends on specific weights being present in the checkpoint - role embedding keys, AdaLN projector keys, source_id-aware layers, depending on which node you're using - and a debug node in this position exists to answer one question cheaply: did the LoRA I just loaded actually contain what the downstream conditioning node expects? That's a much faster check than discovering the answer is "no" after a multi-minute LTXV render comes back with no visible identity transfer at all.

What to actually do with it

Drop it between your LoRA loader and whatever conditioning node you're feeding - BFSHeadSwapBerniniConditioning is the obvious pairing given the shared naming, but there's nothing in the schema that restricts it to that node specifically, since it only takes and returns a bare model. Read the debug output (a STRING) with a Show Text or similar node before proceeding.

Installing it

Same pack, same install path as everything else here. Through ComfyUI Manager, search "ComfyUI-BFSNodes." By hand:

cd ComfyUI/custom_nodes
git clone https://github.com/alisson-anjos/ComfyUI-BFSNodes.git
cd ComfyUI-BFSNodes
pip install -r requirements.txt

Restart ComfyUI afterward. There's nothing this node needs beyond the base pack - no extra model file, no separate dependency - which fits a lightweight diagnostic utility.

Being honest about the gaps

We don't have a tooltip, a node description, or a README section for this one, and with zero recorded impressions there's no click-through pattern to learn from either. If you're using this node and the debug output doesn't match what's described above, trust what your own console shows over this article - it was written from the schema shape (model in, model + debug out) and the pack's broader conventions, not from the source. If the debug string comes back empty or unhelpful on a LoRA you expect to work, the more reliable next step is checking the pack's GitHub issues or the file names inside your LoRA's safetensors metadata directly, since that's ground truth this node is presumably just reading off of.