ComfyUI Node
Image Normalize
Scale an image onto the resolution bucket a model was trained on (Qwen-Image 1K / 2K, SDXL 1K, SD 1.5 512 / 768, Flux 1MP / 2MP) and center-crop it to that bucket's ratio. Also reports the resulting width and height, the source aspect ratio's matched bucket name and the tier's base square side, so downstream nodes can tell how much was trimmed.
Image Normalize
- image
- image
- width
- height
- matched_ratio
- resolution_value
◄modelQwen-Image 1K►
◄resize_methodlanczos►
CategoryQG-tools/Image
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| image | IMAGE | Image to normalise. It is matched to the closest bucket by aspect ratio, scaled without distortion, then center-cropped. | |
| model | COMBO | Qwen-Image 1K | Which model's bucket table to use. Families differ in aspect ratios and total pixels; the output size always lands on one of that model's buckets. |
| resize_method | COMBO | lanczos | Resampling filter. lanczos is the sharpest (recommended), bicubic is smoother, bilinear is faster, nearest keeps hard pixel edges. |
Outputs (5)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | The image cropped to the matched bucket, one frame per input frame. |
| width | INT | Width of the matched bucket. |
| height | INT | Height of the matched bucket. |
| matched_ratio | STRING | Name of the matched bucket's ratio, e.g. 16:9. |
| resolution_value | INT | The tier's base square side (its 1:1 bucket width). |