ComfyUI Node
CV Quality BRISQUE (no-reference)
Grades image quality with NO reference image (cv2.quality.QualityBRISQUE): an SVM over 36 natural-scene-statistics features, trained on the LIVE database. Score runs 0 (best) to 100 (worst) - use it to reject blurry/noisy/over-compressed frames when there is no ground truth to compare against, which is every real capture. Needs two data files in ComfyUI/models/quality (brisque_model_live.yml and brisque_range_live.yml from opencv_contrib, modules/quality/samples). Scores the whole batch.
CV Quality BRISQUE (no-reference)
- images
- worst_score
- worst_index
- scores
◄model_file►
◄range_file►
Categoryimage/CV/quality
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| images | NPARRAY,IMAGE,MASK | Image(s) to grade. Every frame of a batch is scored; frames may differ in size. Accepts a ComfyUI IMAGE/MASK directly (frame 0 of a batch) or an NPARRAY. Arithmetic ops (add, multiply, etc.) process the full IMAGE batch when both inputs have the same batch size. | |
| model_file | COMBO | The trained SVM (brisque_model_live.yml) from ComfyUI/models/quality. | |
| range_file | COMBO | The feature-range file (brisque_range_live.yml) that normalizes the features before the SVM. It MUST be the pair of the model file above. |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| worst_score | FLOAT | The highest (= WORST) BRISQUE score in the batch. Compare it against a threshold to gate a pipeline: LIVE-trained BRISQUE puts clean photos roughly under 30 and heavy degradation over 60. |
| worst_index | INT | Index of the worst-scoring frame in the batch. |
| scores | NPARRAY | float32 array of shape (N,) - one score per frame, in batch order. Chart it with 'CV Chart Series'. |