The main difference is that power is estimated while accounting for the effects of untested loci and environmental factors shared between relatives that also contribute to disease risk. When families with multiple affected offspring have been ascertained, not accounting for these can lead to biased power analyses.
No input files are needed. You just need to specify a set of input parameters that characterise the disease of interest (heritability, disease model, etc) and the sample to be analysed (number of families, family structure, ascertainment strategy, etc).
In addition to the expected test statistic and power of the TDT, the output also gives the power of the parenTDT and case-control analysis for the same disease parameters and sample size. The parenTDT combines allele differences between discordant parents with the allelic transmissions to affected offspring to provide a more powerful test of association. Some descriptive statistics are also printed that may be useful when designing association studies of complex diseases (GRR, expected prevalence of a given family configuration in the population, number of families required for 80% power, etc).
Have a look at the background section for an overview of the rationale behind the TDT power calculator.