I was looking to calibrate a Samsung LN-S4096D and at the moment everything runs through a Denon AVR-888 and all sources are upscaled to 1080p.

Now I know that in Colormetre HFCR that you can do SDTV - REC 601 or HDTV - REC 709. Now would you use the HDTV setting even if its a SDTV disk being upscaled. Wouldn't the Color spectrum still be SD even if it's upscaled? The only reason I want to calibrate with the upscaling on is because thats how all sources are being treated in the end. Is this the correct approach?

A measurement can be represented as xyY. Rec 709 and SMPTE-C define xy color points, and like you commented the target xy varies depending on which is chosen. A target for Y can be calculated from xy.

If you were to measure SD, HD, and upscaled SD you might find that xy measures the same for all three sources. That is how my electronics measure, and I still have not seen anyone list any consumer electronics that do otherwise. It is possible to alter primaries using things like a computer or video processor, but otherwise generally xy will be the same for a given display regardless if the source is SD or HD.

To address your question, in a display with a CMS generally you could choose one standard and calibrate to that standard. Most displays do not have a CMS, and in that case you generally have two choices. Either you can choose the standard that comes closest to your measured xy in ColorHCFR, or you can calculate a target Y based on your measured xy (the Accupel calculator allows this). For example my TV doesn't have a CMS and the primaries (red, green, blue) are close to Rec 709, so I use the HDTV setting in ColorHCFR regardless of source.



http://www.avsforum.com/avs-vb/showthread.php?p=14520435#post14520435 is a further writeup about SD, HD, and upconverted SD. The attachment with the link includes ColorHCFR measurements. The files are as follows:
AVSHD.chc - An HD source.
GetGray-SD.chc - An SD source.
GetGray-HD.chc - An SD source upconverted to HD.
GetGray-HD-IncorrectDecode.chc - An upconverted SD source with the wrong decode matrix. You will notice that this measurement varies more than the others do, especially for Y and the location of the secondaries (cyan, magenta, yellow).