We wouldn't praise Homer for Alexander Pope's rhymes (in a translation of Homer). Nor the balance and symmetry of his couplets. What about Chapman's fourteeners? Those seem closer to dactylic hexameters in syllable count, at least, which max out at 17. But we couldn't credit the Greek author with a renaissance guy's handling of that meter either. Suppose a translator wrote in Homer's meter, dactylic hexameters? That would still be the skill of the translator, though. We would give him credit for trying (if that's the kind of thing we approve of!) but it would a category mistake to think that those were Homer's dactyls and spondees. They are no more Homer's than Pope's rhyme was Homer's rhyme, when you really think about it.
The entire relationship between the signifier and the signified is what is at issue, not simply the technical requirements of the meter. Of course, linguistic considerations come into play as well, like the predominance of stress over vowel quantity in English, but the main issue is not even linguistic, except insofar as translation involves two languages.
I've had a hard time explaining this to people in the past, and have struggled to come up with an easy to understand explanation. One way of getting people to see this is by quality: suppose the meter is the "same," but the translator has a "tin ear" and the original poet an exquisite one. But that implies that a good enough translator would produce a good result. Well, yes, the result could be good, but it will not have anything to do with the metrical prowess of the original. In fact, the translator needs his or her own skill to get up to a level that we think is good, so that proves that this effort comes from the translator alone and is independent of the original.
If we think of translation as the transfer of semantic meaning, we can see that it takes no skill at all to translate the word pájaro as bird. We think that that simply is the translation.