Load Image¶
Loads an image file from the workspace input directory or via upload and outputs it as an image tensor along with a corresponding mask. If the source is multi-frame (e.g., animated GIF/PNG), frames of identical size are combined into a batch. The mask is derived from the image’s alpha channel when available; otherwise a blank mask is provided.
Usage¶
Use this node at the start of image-processing workflows to bring external images into your pipeline. Select or upload a file from the input directory. The node outputs an image suitable for downstream processing and a mask that reflects image transparency.
Inputs¶
| Field | Required | Type | Description | Example |
|---|---|---|---|---|
| image | True | STRING | The filename of the image to load from the input directory. Supports standard image formats and can accept uploads via the UI. | example.png |
Outputs¶
| Field | Type | Description | Example |
|---|---|---|---|
| IMAGE | IMAGE | The loaded image (or batch of frames) normalized for processing. | An IMAGE tensor representing a single RGB image or a stacked batch of frames. |
| MASK | MASK | A mask derived from the image alpha channel; if no alpha channel exists, a blank mask is returned. | A MASK tensor aligned with the image (per-frame if animated). |
Important Notes¶
- Multi-frame images (e.g., animated GIF/PNG) are returned as a batch when all frames share the same dimensions.
- Frames with mismatched dimensions are skipped to maintain a consistent batch size.
- If an alpha channel exists, the mask represents transparent areas of the image; if not, a blank mask is generated.
- Some multi-image formats (e.g., MPO) are treated as single-frame and will not produce a batch.
- Ensure the file is present in the input directory or use the UI upload option to make it selectable.
Troubleshooting¶
- File not listed: Place the image in the input directory and refresh, or use the upload option in the UI.
- Unexpected batch output: The source file may be animated; if you need a single frame, choose a static image or extract a specific frame upstream.
- Missing transparency in mask: The source image may not have an alpha channel; supply an image with transparency if a non-blank mask is required.
- Inconsistent frame count: Frames with differing dimensions are ignored; ensure all frames in the source have identical width and height.
Example Pipelines¶
