In the tutorial the guy selects quicktime, as well as DNxHD. Isn't quicktime a lower quality format though, and shouldn't it just be DNxHD only?
Quicktime is just the container that holds video and audio encoded with a codec - the codec is what determines the quality. You can put nearly any codec into quicktime - DNxHD is a codec, as is h.264.
He also says to set the movie's resolution to whatever the original resolution was. But what if I want it to be 24 fps instead of 23.976? Since 24 fps is more universally accepted, I would like to upconvert it, if that's okay, but he says not to, so is that a problem?
Leave the frame rate at 23.976.
Wikipedia describes DNxHD as lossy as well though, so if that's lossy than how good is it though really?
Most codecs are lossy - the one's that aren't don't reduce the file size very much. The goal is always to balance the amount of quality lost against the data rate to meet a particular need.
h.264 is designed as a delivery codec. That means it's optimized for producing low data rates for delivery online, broadcast transmission via satellite & cable, or on discs like Blu-ray. What it's not designed to do is maintain quality across multiple generations very well, because it achieves those low data rates by throwing out a lot of the original picture information.
Codecs like DNxHD and ProRes are designed for post production. They don't reduce the data nearly as much as h.264 - they'll typically run at 10-100x the data rates typically used with h.264. The trade-off for that higher data rate is that a lot more of the original quality is retained, and if you do have to re-encode multiple times there is very little generational loss.
Unfortunately h.264 has also become common as an acquisition format due to the capacity & write speed limitations of inexpensive solid state memory used in consumer & prosumer cameras. So now you're starting with an h.264 file, and if you master to it you'll add a second generation of quality loss. Then if you compress from your master to h.264 for online delivery, etc you'll add a third pass of h.264 - by this point you'll start seeing visible degradation of your image.
When you go from h.264 to a higher quality codec like DNxHD you essentially don't lose any appreciable quality. So now when you use that master to generate delivery files you've basically removed a generation of encoding loss from the process.
Ideally you'd shoot on a camera that recorded uncompressed video, then you'd stay uncompressed through the post process, and then only compress when it's time to deliver. Unfortunately that's not practical if you don't have a lot of money to spend - requires too much storage, and storage that's too fast (and therefore expensive), at every step of the process. So the trade-off is you work with lossy codecs, but you try to use the least lossy codecs you can at each step of the process. In your particular case that's going to be DNxHD for your master.