We have conducted a deep photometric survey of a 0.5 deg x 0.5 deg area of the Galactic Plane using the WFI instrument on the 2.2-m ESO telescope on La Silla, Chile. The dataset comprises a total of 267 R-band images, 204 from a 16 day observation run in 2005, supplemented by 63 images from a six week period in 2002. Our reduction employed the new numerical kernel difference image analysis method as implemented in the PYSIS3 code and resulted in more than 500,000 lightcurves of stars down to a magnitude limit of R ~ 24.5. A search for variable stars resulted in the detection of 1318 variables of different types. 1011 of these are eclipsing or contact binary stars. A number of the contact binaries have low mass-ratios and several of the detached binaries appear to have low-mass components. Three candidate contact binaries have periods at the known cut off including two with periods lower than any previously published. Also identified are 3 possible pre-main sequence detached eclipsing binaries.

The numerical kernel approach to difference imaging has been implemented and applied to gravitational microlensing events observed by the PLANET collaboration. The effect of an error in the source-star coordinates is explored and a new algorithm is presented for determining the precise coordinates of the microlens in blended events, essential for accurate photometry of difference images. It is shown how the photometric reference flux need not be measured directly from the reference image but can be obtained from measurements of the difference images combined with knowledge of the statistical flux uncertainties. The improved performance of the new algorithm, relative to ISIS2, is demonstrated.

We present an analysis of OGLE 2004-BLG-254, a high-magnification and relatively short duration microlensing event in which the source star, a Bulge K3-giant, has been spatially resolved by a point-like lens. We have obtained dense photometric coverage of the event light curve with OGLE and PLANET telescopes, as well as a high signal-to-noise ratio spectrum taken while the source was still magnified by 20, using the UVES/VLT spectrograph. Our dense coverage of this event allows us to measure limb darkening of the source star in the I and R bands. We also compare previous measurements of linear limb-darkening coefficients involving GK-giant stars with predictions from ATLAS atmosphere models. We discuss the case of K-giants and find a disagreement between limb-darkening measurements and model predictions, which may be caused by the inadequacy of the linear limb-darkening law.