Yes there is help available.
We need the following to help you out.
Stitching software: I assume this is PW 4.0 because this is a "panoweaver" forum. LOL
Type of rotator you are using. Agnos, 360precison etc.
And a test link showing what you have stitched. We need to look at the pano to fig out what might be going on.
You have to have the "Nodal Point" set correctly. Please advise on how you did that.
You can check out the pano here:
360Precision Pano Head Test
There are a few very minor issues but these would be easily rectified by spending another 20-30 minutes calibrating the template properly. At the moment the maximum error is 4px but we noramally aim to calibrate a template to less then 0.75px error.
Hi there Matt,
whilst we all try and be nice and not blatently advertise other software, we do however like all input that will help us produce better work. I'm sure that Easypano will also be able to put up with some crtique and should listen to it so that they can improve their software.This is a very freindly forum and any good advice is much appreciatted.
PS: I can't get the pano to work from your link
If I only had an hour to chop down a tree... I'd spend 45 mins sharpening the axe.
I have tried to do a 'neutral' experiment. The first source images were loaded to PTGui - control points were placed widely throughout the frame (6-7 per image pair, and 2 per image to the up + down shots).
PTgui reported a maximum pixel error of 4px at a bolt very close to the camera on the orange beam. The average error throught the pano was nice and low - 0.25 px. Certainly this 4px error is an 'error' - the control point is accurate, but there is slight movement between shots - back of envelope maths (ghetto maths in other words :)) puts this to equivalent about 1mm of movement of the nodal point (term chosen carefully here purposefully, see later).
The panorama was ouput using the simple 'feather' output option in panotools, this closely mimics what easypano does (from what I can tell), by placing a feather along the seam between the two overalpping images (Enblend is more complicated, and uses crazy blending techniques to do its job and its output is not immediately comparable to easypanos, hence it has not been used).
I cant see any others, I am sure there are some, If you do spot any please let me know where, and ill try to work out why that have occured.
Now we reach the real crux of my test, and the whole reason for doing this, what I shall call the.... In fact, I havent come up with a name for this yet, but I will do shortly. We shall temporariliy call it the 'difference panorama'.
-The output from PTGui was output as a photoshop layered image.
-overlapping layers are merged using the photoshop 'difference' layer blend, and output as a panorama.
If you are not familiar with the 'difference' blending mode, it essentially shows up differences between layers (I rekon you could guess that from the name :)). Areas that are black indicate that there is NO difference. Areas that are lit show up differences.
Most areas (90% of the pano) show <1 pixel error - i.e. most of the differences are black, or show a thin line. These ares will always blend fine, no problems. [Sometimes lines appear thicker than they really are, as the 10.5mm lens suffers from chromatic abberation at the edge of shots, so some edges 'bleed' - this could be correct prior to stitching, but I find it minor after blending.]
Some areas show noticable difference, e.g. trees / areas with beams of sunlight. This is information that changed between shots, nothing anyone can do about this.
One area shows a true difference between shots, that is casued by a shift in the nodal point of the camera/lens - the area where the bar is closest to the camera, looking straight away from the house. This is where the 4px error I mentioned earlier is noticable. As stated this *could* be in response to a very small movement in the camera - but is more likely a different effect. The nodal point of a lens varies as a function of angle. We have made our head so that our nodal point is 'on' at teh blending seams. As can be seen, either side of the seam there is error, but on the seam line, there is not. This will always be an inherant problem when shooting panoramas with really close subject matter. This may also be casuing easypano headache.
I beleive the excerize to have shown 100% that the images themselves are mathematically and physically correct so as to enable them to be stitched perfectly. The reason for them not stitching is not shooting / camera / tripod / 360Precision head issues.
I can only reassert my opinion that the software is the casue of error.