In Harrison's 10 commentary, he has an appendix that focuses on some of these, attempting to categorize them.
“(i) Aen. 1. 561 ‘vultum demissa’, 228 ‘lacrimis oculos suffusa’. In this, the most common type, the direct object of the active form of the sentence (vultum demittere, lacrimis oculos suffundere) becomes the indirect object in a passive form of the sentence, but keeps its accusative case: cf. (iii) below.
(ii) Aen. 2. 273 ‘perque pedes traiectus lora’, G. 3. 307 ‘vellera . . . Tyrios incocta rubores’. In this rare type the accusative after the passive verb replaces an instrumental ablative in the active form of the sentence: traicere eum loris, incoquere vellera ruboribus (for the latter cf. G. 4. 279).
(iii) Aen. 10. 156-7 ‘puppis / . . . Phrygios subiuncta leones’, Ecl. 3. 106-7 ‘inscripti nomina regum / . . . flores’ (the only examples of this type). Here the subject of the passive form is surprisingly the indirect object in the active form: subiungere puppi leones, inscribere nomina floribus; the direct object of the active form of the sentence is now indirect, but keeps its accusative case, cf. (i).
(iv) Aen. 2. 275 ‘exuvias indutus Achilli’. Superficially analogous to (i), such examples are in fact a special case. induor and similar verbs of clothing and unclothing contain a strong reflexive element and imitate the middle voice and form of their Greek counterparts…”
He goes on to say that (ii) and (iii) are most likely developments of (i). These three occur most of the time after a past participle (only exceptions in the Aeneid are 1.713, 8.265, and 9.646), with (i) commonly being used of parts of the body. This, he says, suggests an analogy between the accusative of respect involving an adjective (e.g., saucius pectus), a Greek construction which the poets introduced to Latin.
“...no doubt the similarity of the participial form to a common construction after an adjective motivated the original use of the 'Greek' accusative after verbs in Latin.”
(i) first occurred probably in Ennius and definitely in Lucretius; (iv) is found in Plautus, and (ii) and (iii) are first found in Virgil.
“These facts would seem to indicate that the construction is originally a poetic Graecism [he previously cited examples from Homer, Aristophanes, and Sophocles], and it is not found in the prose of Caesar and Cicero (though Thucydides has it, 1. 126).”