Thanks for your work building and maintaining this toolbox!
I am using a different package (patternize) to extract color patterns because I need the landmark-based functionality to batch process and ensure I am working with homologous regions across photos. However, I was hoping to use the mica toolbox to normalize the lighting of these photos first. For the moment, I am not doing visual modeling in this analysis.
Do you have recommendations on how to save/export the mspecs to a standard image format so that they could be used in other packages? I have tried Image Visualization > Make Presentation Image (using default settings), but these images appear very dull in color compared with the original photos. The duller color in the photos makes subsequent color classification more difficult. Doing the same step but selecting “CIEXYZ to sRGB conversion” results in the red channel being way over-saturated.
Thank you! This works well for individual photos. As a follow up question, I was hoping to batch process making presentation images. However, when I use Batch Make Presentation Image, at the second dialog box I don’t see an option for CIEXYZ to sRGB conversion. Am I missing something?
Assuming you’re working with normalised calibrated images then you’ll want to make a cone-catch model to convert your image to CIE XYZ space (this converts from standard camera colour space to human colour space), then tick the “CIEXYZ to sRGB” button when making a presentation image. This will create “normal” looking, colourful, standardised images.
This last step is critical for getting “normal” looking colours as it expands the colour space to fit standard (sRGB) viewing equipment.
I could recommend playing with the transform value specified in the window that you use when creating your presentation image. Given that you would have linearised your image, it is normal to see the resulting RGB reconstruction somewhat drab as you have removed the non-linearities that create the crisp contrast that we like to see (and the reason the manufacturers put them there in the first place). The options in the ‘create presentation image’ should allow you to re-introduce these non-linearities to a degree, and the ‘adjust image’ functions for brightness and contrast in ImageJ can help too. Alternatively, you can take the 24-bit RGB image and re-introduce further non-linearities by using photo editing software. From what I understand, this process is undoing the calibration (linearization) that you have used the toolbox for in the first place?