We report the discovery of SDSSJ1049+5103, an overdensity of resolved blue stars at (\alpha_{2000}, \delta_{2000}) = (162.343, 51.051). This object appears to be an old, metal-poor stellar system at a distance of 45 +/- 10 kpc, with a half-light radius of 23$\pm 10$ pc and an absolute magnitude of M_V = -3.0^{+2.0}_{-0.7}. One star that is likely associated with this companion has an SDSS spectrum confirming it as a blue horizontal branch star at 48 kpc. The color-magnitude diagram of SDSSJ1049+5103 contains few, if any, horizontal or red giant branch stars, similar to the anomalously faint globular cluster AM 4. The size and luminosity of SDSSJ1049+5103 places it at the intersection of the size-luminosity relationships followed by known globular clusters and by Milky Way dwarf spheroidals. If SDSSJ1049+5103 is a globular cluster, then its properties are consistent with the established trend that the largest radius Galactic globular clusters are all in the outer halo. However, the five known globular clusters with similarly faint absolute magnitudes all have half-mass radii that are smaller than SDSSJ1049+5103 by a factor of $\gtrsim$ 5. If it is a dwarf spheroidal, then it is the faintest yet known by two orders of magnitude, and is the first example of the ultra-faint dwarfs predicted by some theories. The uncertain nature of this new system underscores the sometimes ambiguous distinction between globular clusters and dwarf spheroidals. A simple friends-of-friends search for similar blue, small scalesize star clusters detected all known globulars and dwarfs closer than 50 kpc in the SDSS area, but yielded no other candidates as robust as SDSSJ1049+5103.

The Spitzer Space Telescope First Look Survey Infrared Array Camera (IRAC) near and mid-infrared imaging data partially overlaps the Sloan Digital Sky Survey (SDSS), with 313 visually selected (r<17.6 mag) SDSS Main Sample galaxies in the overlap region. The 3.5 and 7.8 um properties of the galaxies are investigated in the context of their visual properties, where the IRAC [3.5] magnitude primarily measures starlight, and the [7.8] magnitude primarily measures PAH emission from the interstellar medium. As expected, we find a strong inverse correlation between [3.5]-[7.8] and visual color; galaxies red in visual colors (`red galaxies') tend to show very little dust and molecular emission (low `PAH-to-star' ratios), and galaxies blue in visual colors (`blue galaxies,' ie, star-forming galaxies) tend to show large PAH-to-star ratios. Red galaxies with high PAH-to-star ratios tend to be edge-on disks reddened by dust lanes. Simple, visually inferred attenuation corrections bring the visual colors of these galaxies in line with those of face-on disks; ie, PAH emission is closely related to attenuation-corrected, optically inferred star-formation rates. Blue galaxies with anomalously low PAH-to-star ratios are all low-luminosity star-forming galaxies. There is some weak evidence in this sample that the deficiency in PAH emission for these low-luminosity galaxies may be related to emission-line metallicity.

The Hubble Space Telescope has obtained some of its well-deserved impact by producing stunning three-color (RGB) pictures from three-band imaging data. Here we produce a new RGB representation of the $I$, $V$, and $B$ bandpass images of the HST Ultra Deep Field (HUDF). Our representation is based on principles set forth elsewhere (Lupton et al 2004, PASP, 116, 133--137). The principal difference between our RGB representation of the data and the more traditional representation provided by the Space Telescope Science Institute is that the (necessarily) nonlinear transformation between data values and RGB values is done in a color-preserving (i.e., hue- and saturation-preserving) way. For example, if one of the image pixel values saturates the dynamic range of the RGB representation, all three of the R, G, and B values in the representation are truncated such that the hue and saturation of the pixel is the same as it would have been if the pixel had had lower flux but the same astronomical color. This, in effect, makes the bright parts of the representation an informative color map, not "whited out" as they are in traditional representations. For the HUDF, this difference is seen best in the centers of bright galaxies, which show significant detail not visible in the more traditional representation.

We present a new, and we believe arguably correct, algorithm for producing Red-Green-Blue (RBG) composites from 3-band astronomical images. Our method ensures that an object with a specified astronomical color (e.g. g-r and r-i) has a unique color in the RGB image, as opposed to the burnt-out white stars to which we are accustomed. A natural consequence of this is that we can use the same colors to code color-magnitude diagrams, providing a natural `index' to our images. We also introduce the use of an asinh stretch, which allows us to show faint objects while simultaneously preserving the structure of brighter objects in the field, such as the spiral arms of large galaxies. We believe that, in addition to their aesthetic value, our images convey far more information than do the traditional ones, and provide examples from Sloan Digital Sky Survey (SDSS) imaging, the Hubble Deep Field (HDF), and Chandra to support our claims. More examples are available at http://www.astro.princeton.edu/~rhl/PrettyPictures