Thanks for the files Liao
I'll be back to once I have had a good look, Hugh.
Thanks for the files, which I have looked at, and I think that the problem is to do with Control Points because of the message in the "Control Point Assistant".
If you open PTGui with "Liao.pts" there are some 182 Control Points, which are mainly around the centre of the images with very few in the Zenith and Nadir parts, which I think is "pulling" the images too much.
If you open PTGui with "Hugh.pts" you will see only 52 Control points much more evenly distributed.
To investigate this I opened PTGui with "Liao.pts" then used "Delete All Control Points"
then "Generate Control Points"
then I ran the "Optimizer"
then used "3. Create panorama..."
which gave me a much better Equirectangular and SWF, but still with a small displacement at the Zenith.
So, the question is "How are you generating all the Control Points?"
The physical set up of your kit is, in my opinion, good, so PTGui can use "2. Align images..." to generate the required Control Points without any manual intervention.
My installation is set up as:
Generate at most 15 control points per image pair
Generate at most 2500 control points per project
and this works fine with you images.
I noticed that in the "Liao.pts" file the "Optimizer" is set at "Heavy".
I would recommend using it at "Heavy + lens shift".
You can see that in the "Lens Settings" the values in "Image shift:" are 0 with "Liao.pts" and in "Hugh.pts" they have values, which make the a. b and c values much smaller.
When I set up the R1 I found it difficult to get a perfect orientation for the camera and find that using "Heavy + lens shift". compensates for any slight rotation off the the perfect and any minor errors in leveling when taking the images.
Personally I would recommend 6 shots round with your kit as this gives more overlap between images giving a better spread of Control Points, but four does work OK as in this example.
If you are still getting problems it may be worth you asking for help on the PTGui forum:
If you do post on this forum please attach your PTS file as it is probably what Joost will ask for.
I don't wish to dissemble my R1 to investigate as my set up is working just fine as it is, but suggest that is is possible that the differences are down to where exactly you have the lens in the clamp when you tighten it.
Maybe Bill or Nick can clarify this, but the important thing is to find the NNP (Nodal Ninja Point) about which the unit rotates to minimise parallax and although there is a theoretical correct setting the general opinion that appears to be expressed on this forum is to find your own setting that suits you rather than rely on numbers as there are a few factors which can influence the final position.
One example, is where exactly is the Entrance Pupil (No Parallax Point) for your lens.
Some lenses do actually have a single point but fisheye lenses do not and choosing the right setting is what works best for your set up.
In theory you should use the point where the rays at the "join" between your images meet the principal ray and this will depend on the number of images round.
e.g. With a Sigma 8mm lens and 4 shots around the "join" ray is 45° either side of the principle ray which meets it approximately 4mm in front of the gold ring, but for 6 shots round the "join" ray is 30° either side of the principle ray which meets it approximately the gold ring.