Here will show the difference I experienced from V2 to V3 tracking: (I use 4 PS Eyes).
I have very few tracking issues with V2, it works well for me on the complex actions I capture, even in poor lighting recording scenarios.
I create 1 min.+ captures for dancing as a regular process as well as other actions of that length.
These tests were with a 2:00 recording using V2 recorder, so I could show same video in each test.
Recordings done with V3 recorder had no difference in tracking time than shown below.
These are rough recordings, unedited just to show tracking times basically.
V2 tracking with Moves, no retracking to clean any mistracking:
Link:
http://youtu.be/mQiJSSEfDkkV3 tracking with Moves, no retracking to clean any mistracking:
Link:
http://youtu.be/Wq5gFi1f788The videos show briefly the tracking speeds for each version, both run in hi res. and V3 takes twice as long to track same video sequence.
I will say the tracking itself was a bit better, but very slow, approx. 3 hours 45 mins in V3 compared to approx. 1 hr 58 mins V2 for this test and consistently double on
every test so far in V3.
Also, without use of Moves in V3, (which I recorded the Moves in V2 and re-opened in V3 to get to work in this test), the wrists twitch alot even after JR and smoothing
applied, which causes bone twitch on export to other editors. I know you are working on the fix now for V3.
I generally use the Moves on all recordings, as it helps in the tracking of the arms in V2.
I can run V3 in low res and get about same tracking time as hi res in V2, but it increases jitter, even after JR/Smoothing applied, with same settings I generally use.
I don't know what other ppl are experiencing, I would like to see some posts, but seems the subscription will keep a lot from even trying/using it.
Import into MB or 3DS Max requires more smoothing to be added before export from iPi (for me) to get a better output from both.
The visible jitter said in above post on import to 3DS Max was corrected, as it was I made a retrack on a small area and didn't re-run the jitter removal again, which
would have taken another 22-24 mins. to re-run on a the 1:00 recording test. V2 would have taken 10-12 mins.
I will post other videos of export into 3DS Max soon for comparison, but so far, both relatively comparable.
My computer config. is:
i5 3.70G processor (1) 4 core
MSI Radeon 7770 GPU (1) unclocked
240G OCZ SSD where programs reside.
There was no drag difference on CPU from V2 to V3, runs at 10-15% for both.
I have no trouble at all with tracking times in V2, I suspect what I am getting in V2 now is normal, or above normal for most, so should be at least comparable with V3.
Yes, V3 is slow on every tracking sequence attempted so far. (Records from V2 or records with V3).
If you want, put a tracked recording you did on V3 on a Cloud link to show me and I will re-track it on my machine and see if any differences, then again, I don't know
the power of the machine used for your test environment to get the results your saying you get, but should be no difference on my machine from V2, which works
fine.
I understand it is a new release and may have issues, all do, I am just testing and showing my personal outcomes as of now.
It is a good product for what it is, but there are many advances in this field now.