We evaluate the direct detection of extrasolar giant planets with a two-aperture nulling infrared interferometer, working at angles \({\theta}<{\lambda}/2B\), and using a new `ratio-of-two-wavelengths' technique. Simple arguments suggest that interferometric detection and characterization should be quite possible for planets much closer than the conventional inner working angle, or angular resolution limit. We show that the peak signal from a nulling infrared interferometer of baseline (\(\lesssim 40\) meters) will often occur `inside the null', and that the signal variations from path-difference fluctuations will cancel to first order in the ratio of two wavelengths. Using a new interferometer simulation code, we evaluate the detectability of all the known extrasolar planets as observed using this two-color method with the proposed {\it Fourier Kelvin Stellar Interferometer (FKSI)}. In its minimum configuration {\it FKSI} uses two 0.5-meter apertures on a 12.5-meter baseline, and a \(\pm 20^{\circ}\) field-of-regard. We predict that \(\sim 7\) known extrasolar planets are directly detectable using {\it FKSI}, with low-resolution spectroscopy (\(R \sim 20\)) being possible in the most favorable cases. Spaceborne direct detection of extrasolar giant planets is possible with \(\sim 12\) meter baselines, and does not require the much longer baselines provided by formation flying.