The gemara in Baba Kama 72b has a discussion whether eidim zomimin are passul retroactively from the time they gave their false testimony, or only from here on in, from the time that they are turned into zomimin. Rava holds that they are only passul from now on because the concept is a chiddush, why believe the latter more than the former, therefore we only invalidate them from the moment of chiddush. Abaye disagrees and holds that we invalidate them retroactively, implying that it is not a chiddush at all.
The mishna in Makos says that eidim zomimin only applies when the testimony to invalidate the first group is a testimony on their time and place, but not when the testimony is on the murderer or victim. There seems to be a machlokes Rambam and Tur in understanding this din. The Rambam (Hilchos Eidus 18:2) that eidim zomimin only applies when the second group is only testifying about the first witnesses, but if they are testifying about both the witnesses AND the murderer and victim, it is considered just contradictory testimony (trei u'trei), not a case of zomimin. Rambam seems to hold that it only qualifies as a case of eidim zomimin if it models exactly the case spoken in the Torah, but if it includes a contradictory testimony, such as claiming that the murderer or victim weren't even present, it doesn't qualify as a case of zomimin. However, the Tur (c.m. 38) writes that the rationale to believe the second group more than the first is that the second group are character witnesses, they are undermining the people, not just their testimony. He compares this to testifying that they are mechalilei shabbos, or passul for eidus - they are believed more since they are testifying about the first witnesses themselves. The Ramban (Devorim 19:19) says exactly the same thing. The Tur and Ramban would seem to hold, based on their rationale that so long as the testimony of the second group includes character testimony and not just testimony contradicting their testimony, the second group is believed. The Lechem Mishna holds that the machlokes between the Rambam and Tur (ramban) is whether according to Abaye who holds that eidim zomimin are passul retroactively, agrees that it is a chiddush or holds that it is rationale. Rambam holds that it is a chiddush and therefore must directly resemble the case in the Torah (see p'nei yehoshua who makes the Rambam consistent with the Tur, just that if they testify that the eidim were in the same location as the murderer/victim, even if it is a different place, it isn't zomimin since the original group may have messed up about the place but did indeed witness the murder). The Tur (ramban) hold that it is perfectly rationale since they are giving character testimony, therefore if they add in their testimony a contradiction to the original testimony, it doesn't undermine the aspect of zomimin.
The Rambam seems to go li'shitaso that it must resemble the case of the Torah exactly where the second group is only testifying about the location of the first group and not saying anything about the eidus. In Pirsuh HaMishna the Rambam writes:
ולא באו להעיד על עצמה של עדות לא לקיימה ולא לבטלה לפי שהם אומרים אנו אין אנו יודעים אם זה הרג כמו שאתם חושבים או שלא הרג, ואין עלינו לדעת עדותם, אבל מה שאנו מעידים שאתם הייתם במקום פלוני ביום פלוני לפיכך שומעין דבריהם ויהרגו העדים
The Rambam seems to hold that the second group must focus their testimony only on the time and location of the first group, but not make any comment about the testimony of the first group. Of course, it is impossible for them to actually know anything about the testimony since they were in a different place, but the nafka mina may be where the second group were located in the middle between the murder scene and the first group. If the second group would testify that the first group could not have seen the crime - that is zomimin. But, if in addition the second group confirms or contradicts the crime, it undermines their believability and wouldn't be able to turn the first group into zomimin.