I have 2 way speaker system which powered by bi-amp. This speaker is cheap but imaging/staging was good, the vocal is forward.
I have 3 way speaker also. It used passive crossover. The staging is flat.
What are influenced the staging? The crossover design, amplifier, or the shape of box?
Mainly the crossover (including driver characteristics). A little from amplifier, box and recording.
Imaging needs speaker ability to produce low level detail at midrange and higher frequencies. Forward focal is the sign of too much midrange. Tho it may improve staging, it is not always favored.
Low level details can be killed by complex filter, especially notch filter, especially notch filter working at the mid frequency (1kHz-5kHz). Second order (electrical) is usually optimum (depends on driver difficulty).
High frequencies is easily localized by ears, so any breakup (which is common with tweeters) will ruin the image. Tweeter quality and crossover design to handle the issue with tweeter affects imaging. It is better to roll-off the top frequencies than produce them but wrongly. I guess many cheap paper tweeters don't even have the capability to produce 8kHz at sufficient SPL, so it may improve imaging. Tweeters (and associated design) with limited directivity will also improve imaging (such as Vifa XT25). I've seen (not heard) many rising response at top ends. This is not good. Notching is even riskier.
Any crossover is assumed to blend well at crossover frequency. Woofer and tweeter phase response should be overlapped at crossover frequency (and roughly overlapped, probably one or two octave above and below crossover frequency). Achieving this in a 3-way is harder because the midrange should blend with the tweeter and also with the woofer. The woofer roll-off at higher end is usually not steep so the out-of-phase signal from the woofer may still interfere or ruin a little portion of the midrange.
Bigger box (as in a 3-way) cannot disappear as easily as smaller ones because of diffraction, audible box resonance, etc. You don't want to hear any mechanical sound from the box and the driver (and any overlapped mid frequency must be in phase, that's all). Low frequency mechanical sound (such as box and port resonances) has less effect than higher frequency. Piramid shape or any method to absorb tweeter diffraction will help. Tweeter baffle must also free of vibration.
Crossover components waste power, and big woofer (as in 3-way) needs extra power to come alive (big woofers are rarely sensitive). Multi amping, active crossover and high power amplifiers will help here. And high frequency reproduction is also an issue with amplifiers. Many tube amps roll-off the treble early, which may mask the HF issues (with recording or with speaker). Current feedback amplifiers usually have better performance at high frequency.