How it looks like?

The common file used to generate these examples:
# image support (in pixels)
$supp = 128; 
# image scale (pixels per meter)
$scale = 32;
# telescope diameter
$diam = 1.8;
# otturation diameter
$ott = 0.0;
# simulation duration (limited by layer dimensions)
$simt = 2.0;
# timestep
$dt = 0.01;
# runs in ensamble mean
$ens = 1;
# layer structure
$str = $Seeing::test_layers;
# magnitude of object
$mag = 12;
# magnitude of reference
$mag1 = 6;
# additive gaussian noise: SIGMA PER PIXEL
$add = 0.1;
# wavelength
$lambda = 555e-9;

# describe image
$im = zeroes float,$supp,$supp;
$im->set($supp/2,$supp/2,1.0);
$im->set($supp/5*3,$supp/2,2.0);
$im->set($supp/5*3,$supp/5*3,1.0);

Real (upper) and immaginary cross-spectrum in a infinite magnitude simulation:

Recovered image (zoom), phase, log of recovered spectrum, corrected modulus using reference, log of SNR ration and original image of a infinite magnitude, cross-spectrum simulation.

Real (upper) and immaginary bi-spectrum in a infinite magnitude simulation:

Recovered image (zoom), phase, log of recovered spectrum, corrected modulus using reference, log of SNR ration and original image of a infinite magnitude, bi-spectrum simulation.

Comparative recover for a triple start of magnitude 9. Bi-spectrum (upper line) and Cross-spectrum.

Comparative recover for a triple star of magnitude 12. Bi-spectrum (upper line) and Cross-spectrum.