There are several sources that equate albus-candidus with ater-niger, which makes niger to be the opposite of candidus - a shiny black, and ater to be the opposite of albus - a dull black. As @fdb comments it is stated explicitly in L&S dictionary (for example under candidus):
opp. niger, a glistening black; while albus is a lustreless white,
opp. ater, a lustreless black; cf. Serv. ad Verg. G. 3, 82; lsid.
Orig. 12, 1, 51; Doed. Syn. III. p. 193 sq.) (class., and in the poets
very freq.; in Cic. rare).
The issue of color is dealt in Be Not Afraid of the Dark by Shelley P. Haley pages 31-33, and we can find there the same idea:
Albus and ater connote a matte-like quality, whereas candidus and
niger imply luster and brightness.
Indeed there are scriptural examples that contrast the opposites: albus with ater and candidus with niger. Examples that Haley brings:
candida me docuit nigras
odisse puellas . . . (CIL 4.1520) [from graffito]
nec scire utrum sis albus an ater homo (Cattulus 93)
Let us hope however that the above-given distinction is not based on those contrasts, since as my collocations-tool suggests, niger is contrasted both with albus and candidus in about the same rate. In fact L&S itself mentions "exceptions to the general rule" (where albus is contrasted with niger instead of ater) :
The following are examples of the opposition of albus and niger (instead of ater) as exceptions to the gen. rule; so always in Lucr. (who also uses albus and candidus or candens promiscuously), 2, 810; 822 sqq.; 731 sq.; 790; 767-771. Once in Cic.: quae alba sint, quae nigra dicere, Div. 2, 3; so Phaedr. 3, 15, 10; Ov. M. 2, 541; cf. with id. ib. 2, 534 and 535; also id. ib. 12, 403; 15, 46; id. H. 15, 37 al.: “albi et nigri velleris,” Vulg. Gen. 30, 35: “non potes unum capillum album facere aut nigrum,” ib. Matt. 5, 36.—
We also find in Ovid ater-candidus and niger-candidus contrast in the same verse:
candida de nigris et de candentibus atra facere
This "promiscuous" usage of albus and candidus by several authors might indeed suggest, I think, that likewise any distinction between ater-niger should be taken with a grain of salt. Nor we should ignore that across time modification in meaning do happen.