return an Image Class without any properties if upscale > 1

[JChonpca]

DeepTrack Version: 2.0.1
Platform Version: Ubuntu 22.04.3 LTS
Python Version: 3.9.12

If upscale > 1 in Optical Class, then return an Image Class without any properties after calling the store_properties method.

This may be a bug, since I see similar code that uses the properties returned in the Image class with the condition that upscale = 4. But in this similar code, the DeepTrack version remains unknown (<=1.7 since it use Tensorflow).

Code:

from deeptrack.optics import Fluorescence
from deeptrack.scatterers import PointParticle
from deeptrack import Arguments

upscale_para = Arguments(upscale=1)

optics = Fluorescence(
    NA=0.7,
    wavelength=660e-9,
    resolution=1e-6, 
    magnification=10,
    refractive_index_medium=1.33,
    upscale=upscale_para.upscale,
    padding=(10, 10, 10, 10),
    output_region=(0, 0, 64, 64), 
    aberration=None,
)

scatterer = PointParticle(
    intensity=100,
    position_unit="pixel",
    position=(32, 32),
)

imaged_scatterer = optics(scatterer).bind_arguments(upscale_para)

imaged_scatterer.store_properties()

p1 = imaged_scatterer(upscale=1)
print(p1.properties)

p2 = imaged_scatterer(upscale=2)
print(p2.properties)

p3 = imaged_scatterer(upscale=3)
print(p3.properties)

p4 = imaged_scatterer(upscale=4)
print(p4.properties)

Output:

[{'position': array([32, 32]), 'z': 0.0, 'value': 1.0, 'position_unit': 'pixel', 'upsample': 1, 'voxel_size': array([1.e-07, 1.e-07, 1.e-07]), 'pixel_size': 1e-07, '_position_sampler': (32, 32), 'upsample_axes': (), 'intensity': 100, 'name': 'Fluorescence'}, {'NA': 0.7, 'wavelength': 6.6e-07, 'refractive_index_medium': 1.33, 'magnification': 10, 'resolution': 1e-06, 'padding': [10, 10, 10, 10], 'output_region': [0, 0, 64, 64], 'voxel_size': array([1.e-07, 1.e-07, 1.e-07]), 'pixel_size': 1e-07, 'upscale': 1, 'limits': array([[30, 35],
       [30, 35],
       [-2,  3]], dtype=int32), 'fields': [], 'aberration': None, 'name': 'Fluorescence'}, {'upscale': 1, 'name': 'Arguments'}, {'name': 'Microscope'}]
[{'upscale': 2, 'name': 'Arguments'}, {'name': 'Microscope'}]
[{'upscale': 3, 'name': 'Arguments'}, {'name': 'Microscope'}]
[{'upscale': 4, 'name': 'Arguments'}, {'name': 'Microscope'}]

[Pwhsky]

Hi, thank you for bringing attention to this.

As the upscale argument is no longer used when instancing classes, it is always better to use the dt.Upscale class found in features and pass either your scatterer or optical device as an argument to it when generating the images.
something like:

optics = dt.Fluorescence() # Default optical parameters for fluorescence.
scatterer = dt.Sphere() # Default scatterer parameters.

image = dt.Upscale(optics(scatterer), 4) 
image.plot()

dt.Upscale alters the size of the pixels used in the simulation of the optics and scatterer, effectively increasing the resolution.