Originally posted by ArMaP
I do not know if this image was one of the ordered TIFFs, but you can compare the bigger TIFF with the smaller GIF.

Looking to these two images what you think happened, the larger, blurrier TIFF (not visible inside the browser unless you use a suitable plug-in) was the original used to make the smaller, shaper GIF or was it the other way?

Or maybe both images were made from a third, unknown image.

Introduction: The Lunar Orbiter program of the mid-1960’s successfully put five spacecraft into close orbits about the Moon. The thousands of photographs returned were used in support of future manned lunar landings (i.e., the Apollo program), and served as the basis of much scientific research. For more than three decades, lunar scientists have found great use in the generally high spatial resolution prints resulting from these missions [1, 2]. Over the past five years, the USGS Astrogeology Team has endeavored to bring a subset of these photographs, in a digital format, to the desktop of scientist and amateur alike through the Lunar Orbiter Digitization Project.

Project History: Upon completion of a pilot project during 2001 and 2002 [3], we started in earnest on the systematic undertaking of generating a Lunar Orbiter (LO) global mosaic of the Moon [4, 5]. This involved scanning over 30,000 LO filmstrips at 25-micron resolution, predominantly from LO-IV, and a small number from missions III and V for farside coverage. The scanning effort alone took nearly two years to complete. Once scanned and reviewed for quality, filmstrips were reassembled into the final photographic frame as viewed by the LO spacecraft. Nearly 200 reconstructed frames resulted from this effort. Ground resolutions range from 60 to 120 m for high-resolution (HR) frames (captured through the 610-mm lens) and 500 to 1000 m for the medium-resolution (MR) frames (captured with the 80-mm lens). The result of this initial effort will be a moderate resolution, near-global, cartographically controlled digital mosaic of the Moon. In late 2003 and early 2004, as scanning for the global effort scaled back (with frame construction well in progress), another phase of the LO digitization effort began [6]. At low altitude, LO-III and -V collected hundreds of very high-resolution (VHR) frames of the lunar nearside equatorial region. Ground resolution of these data range from 1 to 5 meters for the HR frames and 10 to 40 m for the MR frames. Concurrent with the global venture, the VHR project [7] processed and delivered a portion of this very detailed imagery to the science community. To date, 166 frames (approximately 20% of the available data) have been scanned and assembled.

[snip]

Data Processing: A majority of the work during 2006 focused on refining the camera models for LOIII, -IV and -V, and geometrically controlling the global and VHR frames. We briefly discuss these advancements below, but defer detailed analyses to future documentation through a USGS Open File Report.

HR Camera Distortion Modeling. Unlike the MR camera, LO mission documentation [8, 9] does not describe a distortion model for the HR cameras. A camera distortion effect in the LO-IV data was revealed during photogrammetric processing [10], encouraging development of a model to correct the behavior. We modeled the optical distortion of LO-III, -IV and -V high-resolution cameras by measuring tiepoints between HR and simultaneously acquired MR frames. Feature coordinates in the MR frames were corrected for the known geometric distortion [8, 9]. In addition to determining (for the first time) the true focal length of each HR camera and measuring a radial distortion pattern, we found evidence of a “keystone” distortion in each camera. This results from a slight (<1°) misalignment of the fold mirror in the HR optics; we implemented the correction in our camera model as a simple transformation of the focal plane coordinates.

Boresight calculation. A fundamental component of all LO camera models is the calculation of boresight so that a true point of origin can be located for each frame. When available, fiducial marks are used to calculate the boresight. The quality and behavior of the fiducial measurements were examined while modeling the HR camera distortion. This revealed a common displacement between the outer two and center subframes of an HR frame. It was concluded that this resulted from slightly different boresights calculated from local sub-frame fiducial measurements. The LO HR frames are constructed in three separate sections or sub-frames consistent with their presentation in historic literature [e.g., 2]. Fiducial measurements and offset values for each sub-frame are now mapped into ‘full-frame’ space and used to calculate a single boresight for HR frames. This improved reconstruction of sub-frames into ‘full frames’ and generated a more accurate HR camera product.

Also amazing looking for Smart1 that there is only about 50 low res pics available, like Zorgon mentioned. I somehow doubt anything interesting will come out from Japan, sadly enough...