Fellow Pros,

I currently am using a CP5K-W68, Kaidan CS - 12 step, Manfrotto Leveler, all on great Bogen tripod.  I actually have two exact setups, one as a full backup.  I use a different camera for everything else.  The main pan rig is never removed.

Now the frustration.  12 shots takes too much time.  Too much time to shoot - too much time to process.  However I have incredible control over the quality of my pans.  I have been very very interested in the 3 shot process.  I should be able to get the same result.  Well....so I thought!

I switched the Kaidan bracket for a Manfrotto 302Plus with a small mod to handle the CP5K-FE1.  I have successfully stitched a few pans that I have taken great care in setting up.  I can even run the X, Y, and Radius numbers in manually and get perfect results vs. the numbers EP comes up with.   EP says 993,1260,766 / 993,1259,765 / 992,1260,767 - It has errors.  I tell it  988, 1275, 757 on all three and get perfect results.  However if the lighting is different in a different room but everything else on the camera is the same (WB, EXP, etc) those numbers no longer work perfectly.

Is the repeatability of the algorithm in the program an issue?  Should the numbers vary for each 3 shot set?  I am looking for a process to set up shoot and know it's 99.9% there all the time.  Am I looking for the Holy Grail?  I know I haven't posted a link for the pans.  Right now I'm looking for some helpful insight - besides my wife would kill me if I posted the pans of our less than picked up house.  And I thought the extra clutter was helping the program find the blending areas!

Anything right now would help me keep my sanity intact.

Thanks!

Bo

I know how you feel as I use a similar standard xyz technique which generally works well but does occasionally fail. Sometimes EP can produce a better stitch using it's own xyz values even if they seem wildly different to my standard ones. I put it down to slight errors mounting the camera on a rotator or slight movement of the camera on the rotator. I may be wrong!

Thanks Simon - I really hate rambling on but I thought the background info may help in some way. 

By definition the x, y and radius is only defining the center of your image and the radius of the circle image information. 

Each same brand digital camera's CCD recording chip is located close to the center but not precisely.  Hence the exact position of the circle image data is in a different center point on the black back ground rectangle image.  Even the diameter changes with the resolution used.  VGA, XGA and Full (Nikon flavor) will be quite different.   Panoweaver attempts quite well sometimes to acquire the the image circle for each of the 3 images.

When I first started to understand that all it was trying to do was determine the center of the circle image data inside the 1536 x 2048 rectangle image - I did some testing and found that shooting Nikon's "full fine" the image circle framed with a box was 1424 x 1424.  In photoshop I cropped the circle image data down to 1424 x 1424. All I had to do was to tell Panoweaver that each image was X 1424  Y 1424 and the radius was 1/2 of 1424 or 712.  Given these x,y,r values Panoweaver would certainly find and use the exact center of the square box and automatically generate a great image stitch.

Following the stitch.. I notice several mismatched areas caused by a 3 other values.  So I manually stitched the image making individual corrections for Roll, Pitch, and yaw.  Now that usually FIXED the panorama. 

Roll, Pitch and Yaw issues were caused by each of the single images not being taken properly even using the Pan Head, click stop and leveling.  I pay a great deal of attention to taking properly aligned images to begin with.  This tends to minimize or eliminate post processing clean up work. 

Color varations are minimized by using Automatic Exposure Lock while taking the single images.

Did I mention testing for, locating and using the proper Nodal point will go toward significant stitch improvements. 

But then you probably already knew this.

Dave 

Dave,

As always your responses are very informative and detailed.  I have used your bettery test several times to do initial setups.  Even though many of the brackets are designed for precise nodal point placement - we all know it only gets you close.  Sensor rotation and or skew is harder to get an exact fix on.  I am curious if you have developed a method to define these aspects of a digicam.

Thanks!

Bo