The VG Resource

Theory :

The principle is simple : a video of a 2D game have in each frame an part of the precedent frame.
Therefore : The frames are continuous.
Conclusion : we can reconstitute the map by assembling frames.

Now moving from theory to practice :

1 - Find a video of exploration of the 2D or 2,5D map (Youtube,capture...) in MP4 (best quality possible!)
2 - Download and install Microsoft Image Composite Editor(copyright to Microsoft)
3 - Import it and set Planar projection and wait.
4 - Fix manually problems with an Image Editor

Example (Affordable Space Adventures Wii U)(from 1080P video) :
As you can see, there are some problems in the map, but it can be fixed manually.

Hoping I could help.

don't videos generally use a lossy jpeg type of compression that creates artifacts, though?

It depends of the recorder. I search always the better quality video (1080P).
Example added.

No matter the resolution, if a video is compressed there will be some loss of quality. Just like JPG'd images this is not an acceptable rip, so unless you record it yourself without compression, this method doesn't work.

I agree with you, but not all videos have big compression (even if it is rare to find one).
But if you record yourself, it's easy to apply this method, it saves time even if it is necessary to make concessions on the quality.

The easy and faster way isn't always the best way. What's the point of ripping something if you lose even a little bit of the original quality in the process? I mean, I can see this method being useful for looking at the entirety of a level/map in general, but not for ripping.