In this work, we implement a dynamic gas/oil interface tracking algorithm for the mobilization of bubbles under intense pressure gradients in order to improve the simulation of solution gas drive for heavy oil in the framework of a pre-existing pore-scale network simulator. The model is used to characterize both the stationary capillary controlled growth of bubbles characteristic of slow depletion rates (far-wellbore region) and the flow phenomena in the near-wellbore region: in this case, it is shown how viscous forces lead to an increased persistence of small bubbles for a longer time, creating an effect similar to what described as foamy oil. The study has identified three different regimes of bubble growth, depending upon capillary number and depletion rate, and these regimes appear to cover the entire range of phenomena observed experimentally. These three regimes are (a) the conventional capillary-controlled growth pattern at low capillary numbers, (b) viscous biased growth at intermediate capillary numbers, and (c) bubble mobilization and breakup leading to foamy behaviour at the highest capillary numbers and depletion rates. A predictive methodology for the associated continuum-scale constitutive relationships, such as relative permeabilities, is also proposed for each of the three depressurization regimes.