It is well known that 'hearing voices' (when nobody is there) is a common sign of the 'functional psychoses' of schizophrenia, mania and psychotic depression - when patients may appear to be alert and orientated; but that true visual hallucinations are rare except when there is delirium, acute confusional state, 'clouded consciousness' - when the patient is distractible, perplexed, either hyper-excited or drowsy, and disorientated (does not know where they are, the time or who are the people in their environment).
I suggest that the reason may be related to the necessity for 'global', multi-system brain dysfunction before visual illusions become real and convincing - and this is because humans are highly visual animals such that about half of the brain seems to be involved - in some way or another - with the visual system.
This massive input from the visual system has bee posited as a major driver in the evolution of sleep - a state when eyes are closed and the brain largely cut-off from the environment, presumably to allow internal processing - since an animal's sleep needs seem to correlate with the volume and complexity of its visual input:
This implies that for a person to become convinced that an internally-generated visual illusion is actually real, would require that a great deal of the brain be functioning abnormally, and this would imply a global brain dysfunction with multiple deficits, such as delirium or widespread dementia - which probably is itself exacerbated by delirium:
When only a part of the brain is involved in generating a visual illusion, as happens with some instances of Temporal Lobe Epilepsy, the experience is of a visual illusion which is known to be false, not really happening in the environment - i.e. a 'vision', rather than a visual hallucination which is experienced as real.
So a temporal lobe seizure may create a stereotyped 'vision' (probably an activated memory) of small human figures performing actions, rather like a video replay; but this is not experienced as really happening in the environment so it is not a true hallucination.
In other words, when the input from the various parts of the visual system is inconsistent - so that one input contradicts another - then the experience is of an illusion, not an hallucination.
By contrast, the auditory system involves, as I understand, a much smaller proportion of the brain than the visual system. We may infer that in the functional psychoses, a much more localized and partial brain abnormality, involving the totality of auditory regions, may generate auditory material such as voices that are experienced as real because the auditory information in self-consistent, un-contradicted.
The results is that people suffering only from auditory hallucination are more partially impaired, and therefore less globally impaired, than those who suffer from visual hallucinations.