In both cases the important quantity for reliable texture measurements is the number of grains sampled. You need a sufficiently large number to obtain reproducible results. I think this number should be at least 1000 and preferably 1500. Usually X rays can sample larger areas and therefore more grains than EBSD scans. However, if the EBSD scan goes over at least 1000 grains then you would get very very similar textures. We have shown this in the case of large grained aeronautical alloys:

Hot rolling textures of Al–Cu–Li and Al–Zn–Mg–Cu aeronautical alloys:

Experiments and simulations to high strains.

Q. Contrepois, C. Maurice, J.H. Driver∗ Mater Sci&Eng (2010) A 527 7305–7312

Good question! I'm looking forward to the answers.

I'd think the key difference would be the angular resolution. 

The difference may also arise because of interaction volume (depth) esp. if we have smaller grain sizes.  i.e. volume of sampling.

In both cases the important quantity for reliable texture measurements is the number of grains sampled. You need a sufficiently large number to obtain reproducible results. I think this number should be at least 1000 and preferably 1500. Usually X rays can sample larger areas and therefore more grains than EBSD scans. However, if the EBSD scan goes over at least 1000 grains then you would get very very similar textures. We have shown this in the case of large grained aeronautical alloys:

Hot rolling textures of Al–Cu–Li and Al–Zn–Mg–Cu aeronautical alloys:

Experiments and simulations to high strains.

Q. Contrepois, C. Maurice, J.H. Driver∗ Mater Sci&Eng (2010) A 527 7305–7312

This depends on the number of grains you think you should measure for orientations and whether you suspect the presence of some strong local preferred orientations or texture gradients (and whether you need them measured, or you are interested in the bulk). There are several manuscripts on the comparison of these tecniques, e.g.,

doi:10.1016/j.tecto.2013.12.016

doi:10.1016/S0191-8141(00)00103-6

Remember, that in most general case (triclinic sample/crystal symmetry) to get good orientation distribution function ODF with 5x5x5 degree resolution you would need about 10^6 grains, thus for some coarse-grained materials you should probably aim for synchrotron or even neutron diffraction.

Another thing is the ODF resolution, but as you wish for better resolution that one easily gets with EBSD, you should probe more and more and more grains that is not so easy with EBSD. In principle, ODFs with resolution of about 1 degree could be obtained from synchrotron/neutron diffraction measurement (depends on the detectors set-up), and for this methods you usually have hundreds of thousands or even millions grains measured. 

Well, the area covered by EBSD can by clearly bigger than with conventional XRD techniques, but I guess you are asking for the volume. Since the information depth of EBSD and XRD differ by two-three magnitudes it should become clear that statistically XRD has certain advantages...if the grains are smaller than 1-5µm (information depth of XRD, depending on the radiation used and the investigated material. Regarding the number of grains required for a reliable texture interpretation I do only roughly agree with Julian. The number you need depends on the texture itself. If you have a strong texture the number can be smaller. For a very sharp texture you can certainly live with a few hundred crystals. For a week texture - and this is pointed out by Roman - you need to describe the ODF fairly and not only by smoothed data - you will need at least several 10.000 grains. Regarding the orientation precision, EBSD is already now better than standard XRD measurements (as far as I know). I have in mind that for texture measurements you do not use a strong collimation of the X-ray beam in order to get a smooth signal (theoretically the angular resolution can be higher but then you have to smooth the data). In EBSD measurements the data will hardly smoothed, per default with a FWHM of 10 degrees, since the number of grains are commonly comparatively small. They would not need it if they would collect more data (but time is money). Maybe, this is the reason of Julians estimations of 1000 grains.

You have a certain advantage if you can measure in volume using neutron diffraction or synchrotron. But this is out of discussion here.

Summarizing I would say that XRD as well as EBSD could produce comparable results. If graind become too big, booth techniques are lost. If they are too small, XRD will have statistically benefits since you can only map such microstructure with a certain step width which matches the effective beam size. The beam size becomes effectively bigger with lower magnification (for electronic reasons) since the positioning is much worse (although the real beam size is still in the scale of 10nm.

As additional factor one should keep in mind, that texture as orientation distribution is often less important than the misorientation between the grains, since the grain and phase boundaries have an essential impact of the deformation processes. And for this application XRD is out of the business. 

Concerning the number of grains required to give a good estimate of the ODF I remember some work done by R Penelle's group :

 "Estimate of the grain number for true ODF determination by individual orientation detecting method " Materials Science Forum, Volume 157-6, Issue pt 1, 1994, Pages 375-380

Proceedings of the 10th International Conference on Textures of Materials. Clausthal.

We have used these numbers successfully for several texture studies.

@ Julian: could you give a short summary of the conclusion? I guess, the use of about 1000 grains for any texture is not the content, or is it?

Thanks, Gert

Gert

I remember this number from a Conference presentation (Claustal Texture Conf) many years ago,but the proceedings are difficult to find. However, the same group published a paper in J Appl Crystallography, vol 28 (1995) pp582-89  "Estimation of the minimum number of grains for the ODF calculation from individual orientation measurements..." by T Baudin, J Jura, R Penelle and J Pospeich. Despite the title they do not give a number in the Abstract nor in the Conclusions. The paper is very theoretical. But in the text one finds a sentence to the effect that a good estimate of an ODF requires 1600 grains.

At the time this was important since EBSD measurements were much slower; now you can do a few thousand in a minute so it doesn't really matter.