![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
A display calibration involves a linearisation and characterization. The linearisation - ensuring the desired whitepoint and gradation respectively a good grey balance - can be accomplished via display or videocard LUT. In the latter case this information is stored in the vcgt of the display profile and loaded during system start. When loading a game the videocard LUT is usually resetted. But even if you manage to persist this data the game engine will not carry out color space transformations based on the characterization data of the profile - unlike a CMM in color aware software like Photoshop. Therefore the display should reproduce the intended (ambiguous in this case but assuming sRGB should be fine) source characteristic by itself (=> e.g. color space emulation).
Choose a whitepoint which looks good in your environment. There is - contrary to popular belief - no D65 constraint. Think of different colour matching conditions, color constancy of human perception (should be considered when processing colorimetric data - implicitly done for you when the profile is created but neglected in many TV calibration solutions) and impacts of observer metamerism. But give your eyes a chance (=> time + neutral stimulus) to adapt to the respectively selected whitepoint when you experiment with different targets.
Of course you are not.
Which is exactly why you brought it into the conversation. Now get back on topic color, accuracy troll.
.As I said the whitepoint is a quite flexible factor - especially when only judging only one source at a time. During profile generation (ICC workflow) visual adaption regarding display white is assmumed and the characterization data is adapted to the reference white of the PCS (D50). However: Color constancy has its limits and environments with different sources must be aligned to one another (although that always means a visual but not not necessarily colorimetric identical whitepoint - modern ISO recommendations like the upcomming 14861 account for that). Moreover, color critical work requires defined and constant color matching conditions.
Again I don't like your diction - never "bashed" anyone. Just mentioned basic principles of human perception and colorimetry (reaching back to Graßmann and von Kries). This may collide with some best practises in the video sector (in contrast to fields of application in the graphic industry) but is valid.
A normative reference white doesn't fixate the reproduction white. But even if I don't account for color constancy in the human visual system (which you experience each day) observer metamerism is a strong factor. The neutral tone reproduction of two screens calibrated to the same whitepoint can vary considerably (triggered by the sharp spectra of self-luminous devices) due to differences between normative and actual observer. To visually achieve the "grading-whitepoint" you would have to bring grading screen and TV together and adjust the latter ones whitepoint till there is a match for your individual perception. Just replicating the whitepoint colorimetrically will in most cases lead to deviations during parallel viewing.
Adaption processes (think of a "RGB gain control" of the visual system) still take place (also not contradicted by Fairchild). But you rely solely on arithmetic calculations (via transfering and scaling the tristimulus data in a cone response domain) when determining the corresponding colors* under the new "illumination". In the ICC workflow this process is transparent for the user. However: D50 is a quite "borderline" calibration target and won't lead to a visual match with the media whitepoint of standard paper types under D50 normlight during proofsimulation. 5800K (daylight or blackbody locus) is a good starting point in this case.
I'll try again in a bit.The visual system will chromatically adapt to the neutral tone(s) - unlike a camera where the white balance must be defined during RAW development (digital workflow). Though D50 (respectively the colour stimulus - not the "synthetic" spd which comes into play for illumination) can be quite borderline as I said. Please keep in mind that you have to adapt the reference data when performing an "illumination change" to account for the automatic white balance of the visual system.