Why can't we specify layer numbers in most learners?

I’ve found that we can pass in layer= to a tabular learner but we can’t pass in such a parameter to an image learner? Wondering why this is?

Wouldn’t it be important fine tuning for us to play around with number of layers?