I wonder which is the best method to be used to achieve millimag. photometry that is required for instance for exoplanet transits. I have read some papers that recommend defocusing stars so that the effects of seeing are strongly reduced. According to the authors defocusing stars and using integration times between 100 and 200 secs will deliver excellent resuslts, particularly if SNR is very high (e.g. at around 1,000). What do you think?