Model comparison and range prediction

[San SM]

Dear community,

I have been using the conversations as guide for some questions I have that not always necessarily become clear from the literature and I just have the first question that I do not seem to find a similar one in this group:

I am comparing models of different configurations (different sets of predictor variables and different regularization multipliers). To do this, I modelled the niches onto maxents raw output and calculated their AICs.

Now my question is, given that the raw output is so visually not-friendly, if eventhough I do my model comparisons (and AIC calculations) on this raw outputs, I can still rerun the models with clog output and use those for visualization.

Of course the question is more directed at if the results change much and if the comparisons are not valid anymore and also towards if you have seen similar practices in the literature.

Thank you for your time and coments.

Best
Santiago Soto