CLIP Vision missing or incompatible: it is not the prompt's text encoder

Symptoms and scope

Reference-image workflows need a vision encoder with the expected architecture. Distinguish it from the object feeding CLIPTextEncode.

Error fragments for search; IDs, values and filenames may vary:

CLIP Vision model not found
clip_vision
CLIPVisionLoader

Source-supported context

IPAdapter Plus lists different vision encoders for its variants and places them in clip_vision. ViT-H and bigG are not arbitrary substitutes for the same file. Source 1

Cases to distinguish

1. Text-encoder or diffusion weights were supplied where a vision encoder was required.

2. A real vision encoder has an architecture incompatible with the adapter variant.

3. Unified matching expects a specific name while the original download uses a generic filename.

Diagnostic sequence

The sequence below is an editorial procedure based on the cited context, not a diagnosis already confirmed for your environment.

Step 1. Trace the image-embedding path and record the adapter variant and required vision encoder.

Step 2. Match source, identity, filename and directory to the author's pairing instructions rather than comparing extensions or approximate sizes.

Step 3. Test one image with other style, FaceID and multi-image branches removed to isolate vision encoding.

Step 4. If discovery succeeds but dimensions fail, investigate pairing. Further renaming cannot make an incompatible encoder compatible.

Completion check

The encoder produces the representation expected by the adapter and the minimal image-guided workflow proceeds through sampling.

Limitations and cautions

Moving files does not refresh every running instance. Publisher links are not evidence that this site downloaded or verified the weights.

Original sources

• ComfyUI IPAdapter Plus README — original record checked 2026-09-21.

Localization prepared: 2026-09-22. The source-check dates above were inherited from the existing article; this translation does not claim they were all rechecked today. No GPU run, minimum-memory measurement or runtime guarantee is supplied.

Related troubleshooting and guides

• IPAdapter model not found: unified presets and manual loaders use different rules
• mat1 and mat2 shapes cannot be multiplied: inspect the model component chain