The devil is in the details. I think Abrams will gain some of these voters back by default, namely because of less influence from third parties and the lack of President Trump on the ballot. The lion share that remains is tricky. It's a matter of getting her message to resonant with Obama/Trump and Obama/Third Party voters. I have a feeling that a good majority of these voters (or those in this group that vote) will back Abrams on November 6th.
If these voters supported Obama back in 2012, that's a pretty good indication that they're not necessarily supporters of local Republicans. I'd have to look at the correlation between Obama '12 + Carter '14, but my gut prediction is that it's a fairly high correlation. I know it sounds rather cheap to posit that Clinton was the sole cause for the defections. After all, that would make the rather hefty assumption that these voters will vote for local and statewide Democrats. But it is something to consider. It's hard to predict the exact impact that Clinton's underperformance in certain areas will have on Georgia Democrats in 2018...at least, that's what I think. I could be completely wrong.
That's something to consider, too. That actually makes me quite curious. What were some of the differences you've seen?