Nodes/comfyui-zimage-vl/CLIP Text Image Encode
ComfyUI Node

CLIP Text Image Encode

Encode text and optional images through VLM to produce conditioning for Z-Image.

By yaofeng·Created 2 months ago·Updated 2 months ago· 0
CLIP Text Image Encode
  • vl_model
  • image1
  • image1_mask
  • image2
  • image2_mask
  • image3
  • image3_mask
  • positive
  • negative
text
Categoryconditioning

Inputs (8)

NameTypeDefaultDescription
vl_modelZIMAGE_VL_MODELOutput from Load CLIP with Visual
textSTRINGText prompt
image1optIMAGEReference image 1 (optional). If not provided, text-only encoding is used.
image1_maskoptMASKMask for reference image 1 (optional). Masked-out regions (value 0) are zeroed before VLM encoding.
image2optIMAGEReference image 2 (optional).
image2_maskoptMASKMask for reference image 2 (optional).
image3optIMAGEReference image 3 (optional).
image3_maskoptMASKMask for reference image 3 (optional).

Outputs (2)

NameTypeDescription
positiveCONDITIONINGPositive conditioning with VLM embeddings
negativeCONDITIONINGNegative conditioning (empty text, text-only VLM encode)