There is a small difference in the timing, as the PAL flicker frequency is 25Hz, with NTSC having 30Hz, but in other things the flicker is identical, in placement, light intensity, etc.Normally it shows no difference if any NTSC or PAL signal is used as input.
I see, and I can guess at a reason for it. This is possibly one of the cases where the two 'rectangle circuits' of DISPLAY1 and DISPLAY2 registers are used for different purposes. The main one generates the main picture, while the other one generates the shadow. But GSM always assumes that the rectangle coordinates represent the full screen, which might not be the case. We also always assume that the field/frame method must conform with the main vmode usage and smode2, but that is also not necessarily the case. Not for a shadow generated as a partial screen overlay.There is something that i think is not normal if i use GSM to set NTSC mode for FF 10-2 the shadows of the characters in-battle are only half shown.
One line shadow next one no shadow and so on. My TV has a feature of pausing the picture with this i can see that the shadow is almost normal. One time full shadow then no shadow and half shadow.
Since v0.23d and v0.23e came out later, and use a different rule for DISPLAYx register patching than some previous versions, the results could well differ. If not, then please try with v0.23b as well, since that uses a different rule from v0.23e.It is a few weeks ago that i tested that so i will test that again with the latest available version of GSM.
v0.23e patches both registers when calculating a value for either one,
while v0.23b patches only the register for which a value is calculated.
Neither one will work correctly for the shadows, if the reason for the problem is as I suspect, but there should be a difference in the symptoms for those two versions, and with the pause function of your TV you have a good chance of seeing that difference. But I also suspect that this pause function may average screen values from successive half-frames, so it is not fully reliable as indicating exactly what one half-frame generated.
Best regards: dlanor