Webcam Capture

Documentation

• Class name: WebcamCapture
• Category: image
• Output node: False

The WebcamCapture node is designed to capture images from a webcam, allowing for the specification of image dimensions and the option to capture based on a queue. It extends the functionality of loading images by integrating webcam support, making it suitable for real-time image processing applications.

Input types

Required

• image
  • Specifies the source from the webcam for image capture. It is essential for initiating the capture process.
  • Comfy dtype: WEBCAM
  • Python dtype: str
• width
  • Determines the width of the captured image. A value of 0 indicates no resizing, allowing for customization of the image dimensions.
  • Comfy dtype: INT
  • Python dtype: int
• height
  • Sets the height of the captured image. Similar to width, a value of 0 means the image will not be resized, providing flexibility in image size.
  • Comfy dtype: INT
  • Python dtype: int
• capture_on_queue
  • Controls whether the image capture is triggered based on a queue. When enabled, it allows for more controlled and efficient image capturing.
  • Comfy dtype: BOOLEAN
  • Python dtype: bool

Output types

• image
  • Comfy dtype: IMAGE
  • The captured image from the webcam. This output can be directly used for further processing or visualization.
  • Python dtype: numpy.ndarray

Usage tips

• Infra type: CPU
• Common nodes: unknown

Source code

class WebcamCapture(nodes.LoadImage):
    @classmethod
    def INPUT_TYPES(s):
        return {
            "required": {
                "image": ("WEBCAM", {}),
                "width": ("INT", {"default": 0, "min": 0, "max": MAX_RESOLUTION, "step": 1}),
                "height": ("INT", {"default": 0, "min": 0, "max": MAX_RESOLUTION, "step": 1}),
                "capture_on_queue": ("BOOLEAN", {"default": True}),
            }
        }
    RETURN_TYPES = ("IMAGE",)
    FUNCTION = "load_capture"

    CATEGORY = "image"

    def load_capture(s, image, **kwargs):
        return super().load_image(folder_paths.get_annotated_filepath(image))