We will present results from the New Horizons close observations of Kuiper Belt Objects (KBOs) in 2018-2019. These observations were obtained in the months before and after NASA’s New Horizons spacecraft flew within 3600 km of the cold classical KBO (486958) Arrokoth (formerly 2014 MU69). The LORRI narrow-angle camera on New Horizons was used to observe five objects on two epochs at distances of between 0.09 and 0.25 AU, with spatial resolutions of 70 to 140 km/pixel. Deep imaging sensitivity was achieved by taking 250 images at each epoch, with an exposure time of 0.5 sec for each image, which is the longest available without introducing unacceptably large pointing smear. To reduce data downlink volume by roughly a factor of five, only small windows of each image were downlinked, containing the KBO and a few bright stars for guidance. We were then able to fit the locations of the stars in each of the images, and therefore co-register and stack the 250 images to produce deep stacks of the KBOs. These stacks are the deepest full-resolution images obtained with the New Horizons LORRI camera. Of the five objects, two appear to show signs of be very close binaries. 2011 JY31 appears to be binary in both epochs. By combining these resolved observations with the unresoved lightcurve of 2011 JY31 previously observed by New Horizons, were were able to fit a circular mutual orbit with a period of ~2.0 days and semimajor axis of ~200 km. In addition, 2014 OS393 appears to be binary in the first epoch, but is not visible in the second epoch. This is consistent with the lightcurve variation observed in the unresolved images. 2011 JY31 and 2014 OS393 are among the tightest know binaries in the Kuiper Belt, thus providing a fascinating constraint on both their formation and the formation of contact binaries like Arrokoth. The New Horizons project is currently observing with Hyper SuPrime Camera (HSC) on the NAOJ Subaru Telescope to discover additional KBOs that will be close enough to repeat these observations with, hopefully increasing the number of known tight binary KBOs.