ComfyUI Node
IronLeo78 FaceHugger (pre-prod face swap)
IronLeo78 FaceHugger: optional pre-production head swap (Flux.2 Klein 9B + BFS head LoRA), one 16:9 picture of 0.6 MP per actor. Picture_1 / Picture_2 replace the Prompt Generator's Picture 1 / 2 (main female actress / main male actor; i2v: Picture 1 only). v0.15.4: stand-alone: each result is saved in the input folder; load it in the Prompt Generator yourself.
IronLeo78 FaceHugger (pre-prod face swap)
- Picture_1
- Picture_2
- status
◄Pre_prodtrue►
◄Modev2v►
◄FacesBoth►
◄LLMHuihui-Qwen3-VL-4B-Instruct-abliterated-int8_convrot.safetensors►
◄Use LLMfalse►
◄Keep LLM in RAMtrue►
◄body_1►
◄head_1►
◄body_2►
◄head_2►
CategoryIronLeo78/MTV
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| Pre_prod | BOOLEAN | true | ON: the swap runs and each result is SAVED in ComfyUI's input folder (FaceHugger_female_001.png ...): pick it in the Prompt Generator's Picture 1 / 2 yourself. OFF: nothing runs. |
| Mode | COMBO | v2v | Mirrors the Prompt Generator's Mode (both follow each other). i2v: ONE face swap, pair 1 -> Picture 1, any gender. v2v: the gendered pairs of the preset. |
| Faces | COMBO | Both | v2v: which actors are swapped. Set by the Prompt Generator's preset (01x female, 02x male, 03x both). |
| LLM | COMBO | Huihui-Qwen3-VL-4B-Instruct-abliterated-int8_convrot.safetensors | The vision LLM that reads the Body's facial expression (models/text_encoders). Mirrors the Prompt Generator's LLM both ways. |
| Use LLM | BOOLEAN | false | ON: the LLM reads the facial expression of the Body and adds it to the prompt. |
| Keep LLM in RAM | BOOLEAN | true | Mirrors the Prompt Generator's Keep LLM in RAM (both follow each other): ON keeps the LLM in RAM for both nodes (no reload from disk). |
| body_1 | STRING | — | |
| head_1 | STRING | — | |
| body_2 | STRING | — | |
| head_2 | STRING | — |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| Picture_1 | IMAGE | — |
| Picture_2 | IMAGE | — |
| status | STRING | — |