I have some video's that were obviously captured, or encoded incorrectly.  For example it shows a frame rate of 120 already.  However frame advancing through the video shows the majority of the frames are duplicate, identical frames to the one before.

After 'step 1 split into frame', it seems to correctly extract the unique frames.  For example instead of 10,000 frames of the original, I get 900 extracted in step 1.  This tells me the effective frame rate is more like 10fps, not 120.

What's the best way to 'convert' this into a more reasonable 30fps video?  Being that the only options are 2x or more, I can only get 240fps or higher output.  But the unique frames really only gives 10fps or so.  Do I just have to run with a 240fps goal, and frame decimate later?  That seems like a lot of extra processing.  Which mode (1-4) would be best for video that have an artificially boosted frame rate?  I don't know if frame decimating the original would be best as it might end up dropping unique frames that happen to be next to each other.

use ffmpeg with the mpdecimate filter: https://ffmpeg.org/ffmpeg-filters.html#toc-mpdecimate  it says 

"Drop frames that do not differ greatly from the previous frame in order to reduce frame rate."

Have you tried the "remove duplicate frames" feature? You sort of have to do a weird work around: extract frames with remove duplicate frames, go to your frames and see if you can remove any more, and then import the PNG sequence to DAIN then process.