Acquisition Parameters: Scaling relative to image size

[Julian-Hochhaus]

Currently, many acquisition parameters are given in absolute pixels and have limits.
Are there any reasons (i. e. stability of the prediction algorithm) to not remove the upper limits on parameters such as the upper limit of 99pixels for the default radius of a new spot?

Additionally, our camera resolution is quite high and some large spots do not even fit in the current maximum of the spot radius. Because I had no time yet to look into the problem, my quick and dirty solution was to scale down my input images. I assume, scaling the parameters relative to the image size could fix various similar problems with different resolutions of the input data.

What were your reasons for the standard parameters? Are they scaled relative to the size of your input data?
Probably, I will look into the problem in the next months. If the standard parameters are optimized for your input image resolution, assuming these parameters are proven to deliver good results, I would like to take the ratio of standard parameters to your input image size as a starting point to find good ratios and to let the user use a relative scale/factor to modify the ratio.

Greetings from Dortmund

[andim]

Hi Julian,
Indeed these parameters represent defaults that worked reasonable well on the set of images we and our collaborators have analyzed. The ranges similarly reflect the range of parameters that we thought would be reasonable given the kind of resolutions used in our experiments.

There are thus no strong reasons that justify the choice of any of these values, so if you want to prepare a pull request enlarging the range of allowed values or alternatively a way of scaling them with respect to the image size that would be more than welcome!

Greetings from Princeton.