Earth's atmosphere is a bit of a drag when you're launching (HA GOT EM) but it definitely makes up for it on the return leg, gives you something stationary to slam into and scrub relative velocity without turning you into shocked minerals, and actually we've worked it out that a chemical rocket with additional dry mass for aerobraking actually gets BETTER performance to literally any place in the solar system (where both targets have atmospheres) compared to advanced nuclear thermal propulsion, orbit-only designs. That means using the Starship method a la SpaceX effectively gives you the payload capacity to and from Mars of that of a gas-core nuclear lightbulb design that can't aerobrake.
The same is true for Titan way out at Saturn, although in that case we'd definitely need to look at either nuclear thermal propulsion using methane to decrease travel time, or nose-to-nose Starship spin gravity, or both, because a transfer to Saturn from Earth even with a Jupiter assist still takes the better part of a decade. Since in that case we'd be using the NTR system only as a means of accelerating beyond the initial escape burns from either end, the engine can be quite low power, and we only need two (one to actually use, the other for redundancy). That gets around most of the disadvantages of nuclear thermal propulsion, by using a low mass low thrust but decent TWR engine, which outputs less neutron flux during operation and requires less radiator surface area to manage decay heat after every burn. Also using methane gets you >600 Isp, so better than any chemical engine possible, but with about 5x the propellant density of hydrogen, so you save on dry mass (before you even consider the vastly reduced insulation mass required for methane).
Solar electric propulsion a meme by the way, chemical with aerobraking fucks its ass in terms of every real performance metric (Isp is meaningless in that case until it gets >20x higher and you aren't keeping people alive).