One is no more "correct" than the other but it is important to know that (in my humble opinion and not knowing the Meiji set up at all) the knob default position is probably with polars perfectly crossed, while the image on the far right is with the polars slightly uncrossed by a few degrees. The latter setting will enable better observation of the anisotropic rotation tints.

ln my opinian the knob default under cross polarized give you better image for the coarse as well as the fine grains. Try to check the better image by examining some standard minerals and this will give you the answer

Before you get a picture, calibrate the analyzer and polarizer. They must be mutually perpendicular. To be sure this without plate (without sample) go turning the knob until the field is completely dark, when this happens you will have the right position of the analyzer. With this procedure done, all images with crossed polarizer is correct.

some ore minerals has highlighted its anisotropy when we rotate the polarizer slightly

I also agree with Cristopher. The first position is most probably the "default" position, where both polarizer and analyzer are perfectly crossed. Moving the knob means that you're not crossing the nicols completely. This position is fine just to enhance some minerals that present a very faint anisotropism. No problem with the microscope at all!!.

As mentionned by Stanley and Silva you got a fully answer and procedure to determine the coorect result. Factory parameter especially in the case of Japanese one are generally good and analyzer polariser position optimal. The extreme right picture corresponds to a non perfect A-P position which induces the difference observed. Before microscope session as recommanded by Silva check the A-P position and proceed to the corrections if necessary

Thanks to all for the exhaustive answers. I really appreciate it!

@Ioan Pintea: is effective for minerals larger than ~1-2mm.

Yes, all minerals with covalent bonds are very effective for Raman analyzes, and in the case of the minerals with ionic bonds (NaCl) this is not effective, even in the case of metallic bonds (native element Au). Another detail consist in the laser excitation wavelength used for the analyzes. Also, the spectral domain is very important (in my case where the spectral domain is between 210-3400 cm-1, for the most of sulphosalts the Raman peaks arise bellow ~380 cm-1. In this case, a spectrograph with a larger spectral domain is more efficiently for the sulphosalts, sulfides and so on. Nevertheless, the main peak ~380 cm-1 can be enough to make the difference between tetrahedrite and tennantite.

I too think it's a good idea to adjust your analyser-polarizer set. Then the best settings are the ones which give you the most useful images. In the photos you provide it seems that the crystals have a different orientation which explains their anisotropy. All three photos look useful to reveal the minerals you're looking for.