How to discriminate values between spikes?

I have values that represent the temporal complexity of each proceeding video frame. I am trying to remove all the visually-duplicate ones but the trouble is they aren't exact duplicates so they need a threshold which itself is a problem because setting it high enough to kill every duplicate will also kill legitimate frames with low complexity such as slow panning or cloud movement.

Luckily each animated video has patterns. Some scenes use the full FPS and have no duplicates, some are half the framerate, some are 1/3 of it. To deal with the scenes where every second frame is a duplicate, I did =A1/A2 formula which I think effectively discriminates the bottom value between each spike:

35.9146 6.475182548
5.5465 0.179759003
30.8552 10.61629507
2.9064 0.111036401
26.1752 4.630484008
5.6528 0.216141811
26.1532 5.487452791
4.766 0.209058011
22.7975 7.026073289
3.2447 0.190551976
17.0279 5.253254766
3.2414 0.200646248
16.1548 5.420346262
2.9804 0.228211764
13.0598 11.73492677
1.1129 0.159372762
6.983 3.499373591
1.9955 0.024428432
81.6876 29.42318914
2.7763 0.121364592
22.8757 18.77519698
1.2184 0.029882227
40.7734 7.880288359
5.1741 0.136040238
38.0336 20.69743143
1.8376 0.045935177
40.0042 17.44927157
2.2926 0.039427047
58.1479 45.35717629
1.282 0.021224533
60.4018 20.98960976
2.8777 0.057400397
50.1338 11.24858085
4.4569 0.096459049
46.2051 22.12675989
2.0882 0.052623223
39.6821 20.97362579
Anything below 0.5 up here is obviously a bottom spike. But my problem is with scenes that have one unique frame in every third frame. A floor-floor-spike pattern, which my current formula will fail on. Observe:

43.4271 8.137596972
5.3366 1.021319758
5.2252 0.114447671
45.6558 12.40714169
3.6798 1.045338333
3.5202 0.075928235
46.3622 23.7681739
1.9506 0.5542422
3.5194 0.096100967
36.6219 10.38388908
3.5268 0.541460045
6.5135 0.188867174
34.4872 7.103439753
4.855 1.155017367
4.2034 0.13446405
31.2604 3.216123623
9.7199 2.13024897
4.5628 0.132141699
34.5296 9.93457433
3.4757 1.112508802
3.1242 0.087898917
35.5431 6.515692026
5.455 1.538179562
3.5464 0.184684286
19.2025 3.585700148
5.3553 1.259122543
4.2532 0.062959164
67.5549 8.260766954

I need the low values between each spike to be 0.5 or less or the threshold won't recognize them as duplicates.

Anyone have the obvious solution my feeble-minded head is not comprehending?

44 views and no replies.... I don't understand what I am seeing and what we are meant to do with it!! You have two columns of numbers. Start by telling us what they are, and take it from there....

Post a workbook instead where you include some sample data and the expected results. It will make it much easier to help you.

In B1 and drag down=if(a1<8,"",a1) will do i, but there isn't any science behind the assignation of the cut-off point. One could also do it on the basis of the mean minus 3 or 4 SD of the peaks, but we'd need to know a bit more about your data. Can you tell a priori if a dataset is going to be Hi-Lo-Hi-Lo or Hi-Lo-Lo-Hi-Lo-Lo?

Yes, I see the problem: in your first post you had a H-L-H-L sequence:

16.1548
2.9804
13.0598
1.1129
6.983
1.9955
81.6876
2.7763

You need to assign the line in bold, above, as a spike. It's night time here (I'm currently in Poland). I'll ruminate on this overnight and, if no-one else has sorted you out; I'll be back sometime during the EU's morning.

Yes, there are 3 different patterns. Ones with no fluctuation (no spikes) because the scene has no duplicate frames. Ones with a Hi-Lo pattern and ones with a Hi-Lo-Lo pattern.
I can post the whole log if it'll be helpful.
Thanks for the help.

I had posted last night, asking you to post the log... as this is either much simpler than I think, or much more complicated than I fear - but I don't know which... However, the post hasn't appeared. Thank you... free wifi in Krakow.