I looked at the Jython script published in Nature Method (doi:10.1038/nmeth.2019) on the figure 2 (Scripting and ImgLib ) and tried to do the same on some dataset.
The DoGPeaks output 4584 peaks that I later try to plot on a Image3DUniverse.

I did several test and realized I was not able to plot more than 3495 points, mo matter what diameter is used to plot the icosphere.
# diameter in microns
cell=4.0
# diameter Cell visualization
cellIcosophere=3.0
# Show the peaks as spheres in 3D, along with orthoslice
minPeak=40
# The minimum intensity for a peak to be considered
#imp=IJ.openImage('http://fiji.sc/samples/first-instar-brain.zip')
#16 bits (Do not work with 32 bit)
imp = IJ.openImage(pathStack16Bits)
imp8Bits = IJ.openImage(pathStack8Bits)
cal=imp.getCalibration()
sigmaLarge=[cell/cal.pixelWidth,cell/cal.pixelHeight,cell/cal.pixelDepth]
sigmaSmall=[a/2 for a in sigmaLarge]
#Difference of Gaussians
peaks=DoGPeaks(ImgLib.wrap(imp),sigmaLarge,sigmaSmall,minPeak,1)
univ=Image3DUniverse(712,712)
#Plot first 3499 points, looks to not plot over 3499 points
univ.addIcospheres(points[0:3495],Color3f(1,0,0),2,cellIcosophere/2, 'Cells').setLocked(True)
#Add the orthoslice of the stack
univ.addOrthoslice(imp8Bits).setLocked(True)
univ.show()

So if I try to plot more than 3495 (e.g. points[0:3496]), the plot is not done.
There is no error on the output