ComfyUI Node
CV DNN Images From Blob
Unpacks a (N, C, H, W) NCHW blob back into an (N, H, W, C) batch with cv2.dnn.imagesFromBlob - the generic POSTPROCESS that undoes the CHW->HWC transpose blobFromImage did. Values are returned untouched (still the network's output range, e.g. 0-1, and in the blob's channel order). For a 3-channel image the output channels match the blob (RGB if you swapped R/B going in). Use 'CV Unstack Batch' to split into individual frames, or 'CV Index Batch' to pick one by index.
CV DNN Images From Blob
- blob
- images
Categoryimage/CV/dnn
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| blob | NPARRAY | (N, C, H, W) NCHW blob, e.g. an image model's output from 'CV DNN Forward'. Any channel count: C=1 gray, C=3 colour, C=2 a dense flow field ((H, W, 2) dx/dy, which is what the flow nodes take). |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| images | NPARRAY | (N, H, W, C) float32 batch in the network's output range. Use 'CV Unstack Batch' or 'CV Index Batch' to split into individual frames. |