Actually this is a frame by frame conversion using a good dithering (Floyd and Steinberg).
Dragon's lair used a not so good converter, but has palette changes at each frame
The encoder of my videos does much more work apart screen 2 conversion, as a single tile set is optimized across many frames thanks to vector quantization.
The problem with videos like those is that the compression probably converted some of the dithering back into colours. Making it seems like the source was much better than it actually was.
For Sure the videos may be converted to screen 2 frame format, but without taking care of the tms palette.
This is the main reason that makes pictures look better than should be