It worked!!!
For the moment instead of changing offsets values in the header, i retain absolute offset position of the data just adding a bunch of zeros between the header and the "body" (real data).
Gpsmapedit opened it without any problem!!!
I'm not kidding.
On the contrary cGPSmapper 0085 don't like it, failed, probably because it uses a different algorithm (to retrieve the data) and get confused about the zeros.
So what i did, with a pseudo-NT file, i retrieve the gmp subfile and start extracting each subfile within it (TRE, RGN, LBL, NET, NOD).
I can get them by knowing where RGN, NET, NOD starts using the 0x15 offset value in their respective header.
Then i make some assumptions and extract each header and body subfile.
Then i create individually the complete subfile stacking the header, then a bunch of zeros (to maintain the absolute position of the body data that was in the gmp), and then the body data.
Then with gmaptool join them together and get the img.
Things to do is
+forget about adding zeros and change the offset values in the header, i think it is possible just need analysis.
+analize where TRE and LBL points to the offset, as they are different to the others. To get a precise way to find all body starts position without doubts.
EDIT: Joining the subfiles using GmapTool with the "short header" option enabled do the trick and NOW cgpsmapper 0085 get it right!!!!!!!!
Bookmarks