Well, I finally had some success with PW3.  I put all the technical setup data on my demo page.  Essentially found (again) the nodal point to be at 67mm.  Shot 3 portrait shots. Used AE Lock and set the white balance to incandescent.   Brought them into PW3.  After several trials we finally got one that matched up rather well.  But still had blending (none) and you can see small coloration differences.  Mismatch factor was 0.1949.

Center is not always center.

One thing I found is that PW does some anticipated math work to determine the CENTER POINT of Height and width   the Y and X of the black rectangle. Typically, no two Nikons with the Fisheye have the same center of the circle image.  I have seen FE 1' s  that the circle was barely cropped off one side.. meaning a center of this image would be off center of the estimated center of the box.  This requires manually tweaking and relocating the circle X width center within the width of the rectangle width.

PLEASE WHEN VIEWING THIS PAN - disregard the MESS.  We tend to get messy when testing for nodal points and stitching.

http://360texas.com/tips/rd/pw30test/officespin.htm

The quality of the image is LOW because I stitched at 1400 x 700 dimensions about 80kb .  Typically we shoot at twice that dimension to keep more pixels better detail.  For this session we were only interested in stitching and blending.

Will let you know more when we learn more.

Dave

21Dec02 - Better performance this time:

Nodal Point Changed from 67mm to 61mm

Original single portrait 1536x2048
Full Normal 300ppi

Automatic Stitch setup:  2min Stitch time using PC P4 2.5mhz 1gig ram
Images
1.  21 785 1035 711
2. 18 784 1033 713.5
3. 18 785 1029 704

Mismatch factor .228
Image dims 1400 x 700
PS 7 Save for web 1400 x 700 95kb

I added this pan to the bottom of the page.  The URL is above.

27Dec 02  PWpro 3.0 page update

Visit  http://360texas.com/tips/rd/pw30test/officespin.htm  Texas Winter.

New 3 shot image set out side in the sunshine today.  Nodal point remains at 61 mm.  Only 1 stitch err located over double windows on roof line.   Easily fixed with Photoshop.  Also removed the lens flares as a part of normal cleanup.

Had to use the image brightness adjustment

Also image #1 in the series was noticably lighter shade and did not automatically blend well during stitch.  Manually adjusted image 1 "Brightness" darken 2 tick marks to make them closer to images 2 and 3.  Yes I did have AE Lock enabled.  Again I would have thought the software would have looked at image 2 and 3 attempted to blend image 1  seam areas but did not.  Sort of color balancing.

On the above page, I am keeping the top image and updating the bottom image for my learning curve comparision.

Just thought you might like to see this.

Dave

Hi Dave,

Thanks for sharing

I follow your posts with interest, along with all other interesting pointers on panorama imaging.

It was a nice day here in my home town in Australia so I did my first outdoor shot with my newly imported Kaidan gear. Please take a look and point out any problems you see. (Two Fisheye Shots) here

Regards, Smooth

Hi Smooth,

Interesting view.  Only suggestion might be is to slow the rotation down abit.  You might consider rotation between .2 and .4.    If you are using the default rotation speed of .8  its abit fast.  Slower rotation allows the visitors to see and enjoy the scenery as you slowly move through the panorama.

We are viewing your work with a P4 2.5ghz cpu.

Good Work...

Dave

Hi Dave,

Ummm, it's already set at 0.2 on the rotation speed?

I followed your advise from a previous posting.

Thanks for looking.

Regards, Smooth

If you have 'auto' already set for .2 then look farther down in the code for strings dealing with 'autopan'.  You should find 3 values following...(.8,0,0.1)  or something like that... the first value is autorotation.

I think its the last event takes precedent over param strings shown first in the list

Hi.

I am also very impressed with Panoweaver and its capabilities. Here is a sample of what I could achieve using a 2 fisheye shot.

Any comments would be appreciated.

Regards,
CS

Interesting approach using the Shockwave projector inlieu of java.  I found the viewer to be a bit quick. Meaning that the mouse control accelarates faster than expected. 

Typically a java viewer has a bit of a drag to it.. with out much accelaration.

Interesting photo location too.. gives the visitor some detail to view instead of blank ceilings.  The Nadir (bottom) appears to have less detail.

Good work overall.

Dave

Dave, 360Texes Wrote:

If you have 'auto' already set for .2 then look farther down in the code for strings dealing with 'autopan'.  You should find 3 values following...(.8,0,0.1)  or something like that... the first value is autorotation.

I think its the last event takes precedent over param strings shown first in the list

*************************************************************

Yup, you were right! I have changed this now.

Thanks for pointing this out

Regards, Smooth