You're a bit out of my comfort zone as I just haven't taken the time to learn as much about the cameras and protocols; but in this example, I have a megapixel IP camera - specifically a Ubiquiti AirCam - it does not support mjpeg or mp4; only h.264. I use IP Cam Viewer - I choose Type of Generic RTSP over TCP, and I give it a URL of RTSP://myhome.dyndns.org:554/live/ch00_0 (which is the full resolution stream I use in VLC) and it comes up just fine and performs pretty well. I also have a matrix screen that shows 6 cameras simultaneously - and though the rest are mp4 I believe, it shows them all pretty well with no stutter or notice of performance degradation. This to me implies that it must be possible some how!
Also I'll note that I have this same program and configuration going on an original iPad and an iPhone 4S.