I'm not a specialist of Zfish model, I mainly use Xenopus as model organism. However, both xenopus and zfish share a kind of development that should allow you to know the origin of a specific cell type as sson as 8/16 cell stage.
I advice you to inject your cappedRNA, if it is possible, in a specific blastomere (at 8/16cell stage?) using low amount of water (5 or 3nl, or less). If you inject at 2cell, your RNA will be distributed in an half of the embryo (teoretically), and a lineage tracer may halp you to know the distribution of your RNA. However, if you know the expression profile of your gene of interest, you could inject your cappedRNA (or your MO) in the blastomere that will give rise to the cell type of interest. This kind of injection should help you to target your gene/RNA of interest for Morpholino-mediated loss of function, whereas if you inject RNA you should obtain a gain of function. For mis-expression, you can inject your cappedRNA in blastomeres where your gene is not expressed. This "blastomere specific" approach should avoid injection of high amount of RNA or MOs, potential off-target effects. I like a lot this kind of set-up, it is really clean and it avoid indirect effects (I mean, if you wish to study the activity of your gene in a specific cell type, in ectoderm for exemple, this set-up avoid interferences on the activity of your gene in the mesoderm if it is expressed here).
Of course, people rutinely using zfish will give you better advices