After seeing all these pictures in the recent publications, I was wondering if the particle color matters to the toolbox - either white particles before black background or black before white background?
Also, is there a recommendation of how many particles should be in the image? I had some good results with a very thick particle solution - although I'm not sure if the program now rather analyses the gaps between the particles than the other way around. I guess this could be misleading if there is inhomogeneous movement, right?
PIV assumes that black is the background with no movement and white are the image parts that actually move.
If you look at figure 2.14 in the PIV chapter
http://dissertations.ub.rug.nl/FILES/faculties/science/2014/w.thielicke/Chapter2.pdf then you'll see that there should be about 5-20 particles per interrogation window (this applies for particles with the "optimal diameter" =3 pixels). I think generally the more texture available, the better will be the analysis.
The lower picture should work better. The seeding is very low however. A non-uniform background illumination will be almost removed when CLAHE is enabled.
With "texture" I mean: best solution might be when the particles cover as much area as the background. That would result in the max texture.