That doesn't work for a four way split. The top of the screen could have almost zero delay, but the bottom of the screen would need half a frame of delay.
But you can keep the delay to a maximum of about half a frame (and an average of just over a quarter frame across the whole display). It requires driving the two bottom displays out of sync to the two top displays by half a screen.
You could do more with a small software modification on the n64, make it interleave the top two and bottom two images as alternating scanlines. That way while it's outputting the repeats for the top two, the n64 is scanning out the next line for the bottom two.
No, the bottom of all four output displays would have an extra ~8ms of delay on top of any native delay. It's unavoidable (unless you can use seriously non-standard timings) as each players output display takes ~16ms to scan from top to bottom, while the n64 takes ~8ms to scan from the top to bottom of each player's view.
You can avoid (additional) unequal treatment by adjusting the timings of the bottom two players so their TVs start scanning out the top line right after the n64 scans out it's middle line.
But you can keep the delay to a maximum of about half a frame (and an average of just over a quarter frame across the whole display). It requires driving the two bottom displays out of sync to the two top displays by half a screen.
You could do more with a small software modification on the n64, make it interleave the top two and bottom two images as alternating scanlines. That way while it's outputting the repeats for the top two, the n64 is scanning out the next line for the bottom two.