Create custom mask

[SensorFusion]

I am just starting to use the python API and have some problems.

I wanted to test if i can add a custom masks to an image.

Code: [Select]

    doc = Metashape.Document()
    doc.open(pathToProject)

    for camera in doc.chunks[0].cameras:
        # create mask out of rectangle       
        newMask = Metashape.Mask()
        camera.mask = newMask

How can i create a custom mask? For example a rectangle. How would i do it?

If i want to see the signature of Mask() the following appears in my IDE:

Code: [Select]

def __init__(self, *args, **kwargs): # real signature unknown

[Alexey Pasumansky]

Hello SensorFusion,

Do you need to import the mask from the file or to "draw" it?

[SensorFusion]

I need to "draw" it. I want to specify a rectangle on pixel base. Like this

Code: [Select]

Rect(left = 0, top = 0, right = 200, bottom = 200).and then add it to an image as mask.

[SensorFusion]

Does anyone have an idea? Or do I need to create my masks as PNG-files and then load it? Is this the only way to create a mask?

[Alexey Pasumansky]

Hello SensorFusion,

I'll post an example soon.

[SensorFusion]

I found a workaround: Creating an image file as mask and set it to the camera. But perhaps it is possible to set the rectangle directly?

Code: [Select]

from PIL import Image, ImageDraw

def createFullMaskFor(camera):
    width = int(camera.photo.meta["File/ImageWidth"])
    height = int(camera.photo.meta["File/ImageHeight"])
    img = Image.new("RGB", (width, height))
    draw = ImageDraw.Draw(img)
    draw.rectangle(((0, 0), (width, height)), fill="black")
    name = "MASK.JPEG"
    img.save(name, "JPEG")
    newMask = Metashape.Mask()
    newMask.load(name)
    camera.mask = newMask


[smiller]

Is there a way to draw the rectangle directly, like in the GUI? Then apply that selection as a mask?

[Alexey Pasumansky]

Hello smiller,

You can create an empty single-band image (Metashape.Image) of the required size and fill it with black and white values pixel by pixel, then assign this image to a Mask class variable, and assign that mask to the camera.

To speed up the process, external modules, like numpy, can be used.