impact of zero-padding on spatial normalization

Greetings everyone,

Firstly, thanks everyone for asking questions and providing answers. I am new to the fMRI, and this is my first time asking a question in this forum. Please excuse the wording of my questions. Would you like to provide information about the following questions?

My questions are about @auto_tlrc command. In the past, default -pad_base was 40. Recently, the default -pad_base is 15 (Version AFNI_19.3.16 ‘Nero’).
a) What is the purpose of pad_base?
b) What was the reason that the default pad_base was changed?
c) What is the impact using big or small pad_base input?
d) What are the pros-cons of using bigger pad_base?

Thanks and have a nice day

The default zeropadding was reduced to save memory. For high resolution datasets, the former default of 40mm on all sides could make some datasets exceed the memory capacity of even very large servers. Alignment would also use vast amounts more (~18x). The zeropadding helps to allow for tilting of the brain inside the dataset grid but may only be needed for very oblique datasets.

Thanks a lot for your response.