Context : Using observations to deduce dust properties , grain size distribution , and physical conditions in molecular clouds is a highly degenerate problem .

Aims : The coreshine phenomenon , a scattering process at 3.6 and 4.5 \mu m that dominates absorption , has revealed its ability to explore the densest parts of clouds .
We want to use this effect to constrain the dust parameters .
The goal is to investigate to what extent grain growth ( at constant dust mass ) inside molecular clouds is able to explain the coreshine observations .
We aim to find dust models that can explain a sample of Spitzer coreshine data .
We also look at the consistency with near-infrared data we obtained for a few clouds .

Methods : We selected four regions with a very high occurrence of coreshine cases : Taurus–Perseus , Cepheus , Chameleon and L183/L134 .
We built a grid of dust models and investigated the key parameters to reproduce the general trend of surface brightnesses and intensity ratios of both coreshine and near-infrared observations with the help of a 3D Monte-Carlo radiative transfer code .
The grid parameters allow to investigate the effect of coagulation upon spherical grains up to 5 \mu m in size derived from the DustEm diffuse interstellar medium grains .
Fluffiness ( porosity or fractal degree ) , ices , and a handful of classical grain size distributions were also tested .
We used the near– and mostly mid–infrared intensity ratios as strong discriminants between dust models .

Results : The determination of the background field intensity at each wavelength is a key issue .
In particular , an especially strong background field explains why we do not see coreshine in the Galactic plane at 3.6 and 4.5 \mu m. For starless cores , where detected , the observed 4.5 \mu m / 3.6 \mu m coreshine intensity ratio is always lower than \sim 0.5 which is also what we find in the models for the Taurus–Perseus and L183 directions .
Embedded sources can lead to higher fluxes ( up to four times greater than the strongest starless core fluxes ) and higher coreshine ratios ( from 0.5 to 1.1 in our selected sample ) . Normal interstellar radiation field conditions are sufficient to find suitable grain models at all wavelengths for starless cores .
The standard interstellar grains are not able to reproduce observations and , due to the multi-wavelength approach , only a few grain types meet the criteria set by the data .
Porosity does not affect the flux ratios while the fractal dimension helps to explain coreshine ratios but does not seem able to reproduce near–infrared observations without a mix of other grain types .

Conclusions : Combined near– and mid–infrared wavelengths confirm the potential to reveal the nature and size distribution of dust grains .
Careful assessment of the environmental parameters ( interstellar and background fields , embedded or nearby reddened sources ) is required to validate this new diagnostic .