ComfyUI Node
CV Concat Arrays
Joins arrays along an EXISTING axis - the inverse of splitting one. axis 0 appends rows (two detection tables, two point sets, two frames of samples), axis 1 puts them side by side, axis -1 merges channels back together (the return leg of 'CV Split Channels' / cv2.extractChannel). Unlike 'CV Stack Batch' it adds no new axis and the inputs need only agree on the OTHER axes. Mixed dtypes are promoted the numpy way (uint8 + float32 -> float32); cast first with 'CV Cast Array' if you need to control that.
CV Concat Arrays
- array1
- array2
- array3
- array4
- nparray
- length
◄axis0►
Categoryimage/CV/low-level
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| array1 | NPARRAY | First array. Its number of dimensions and its sizes on every axis EXCEPT the concat axis must match the others. | |
| array2 | NPARRAY | Second array, appended after the first. | |
| axis | INT | 0-8–8 | Which axis to join along. 0 = rows (append), 1 = columns (side by side), -1 = the last axis (merge arrays that already have a channel axis). Negative counts from the end. ONE PAST the end - axis 2 for (H,W) inputs - stacks a NEW last axis instead, which is cv2.merge: three (H,W) channels become one (H,W,3) image. Splitting with cv2.extractChannel drops that axis, so that is the setting the round trip needs. |
| array3opt | NPARRAY | Optional third array, appended after the second. | |
| array4opt | NPARRAY | Optional fourth array, appended after the third. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| nparray | NPARRAY | The joined array. Its size on the concat axis is the sum of the inputs'; every other axis is unchanged. |
| length | INT | Size of the result along the concat axis - handy for splitting it back apart with 'CV Take By Index'. |