proj-imim:
Re: Questions
Philo 2000-Apr-20 11:03:16
Hi Mike,
I think I got your problem, I will answer by including comments in your
message... BTW, I began writing a tutorial on stitching method, I will post
a message in this list when I upload it to my homepage.
-----Message d'origine-----
De : The Swamp Gang <#removed#>
À : Immersive Imaging <#removed#>
Date : jeudi 20 avril 2000 07:08
Objet : ignore the last one - this is the one to see
>Hi Philippe & everyone,
>
>I've been reading the posts for a few days but felt too imitated to ask
>anything. First, I gotta say, Philippe, you must be a genius the way you
>developed a virtual tripod. What a great idea.
## thanks, but give full credit to the right person : Helmut without who all
this pano stuff would have been impossible ! ##
>
>I've been struggling with little to no success using Panorama Tools
>(WINDOWS) for a few weeks and have learned alot of terms (I even understand
>some of them now) and lost a significant portion of my life trying to
figure
>out what I was doing wrong. I tried your approach below and even got the
>script to run. But the pano never developed and the output was
>horrendous. If you don't mind, I'll try to show what I did inside your
email
>using (** **)
>in hopes of learning what and where I went wrong.
>
>> -----------------------------
>> My procedure is as follow :
>> - First I shoot the panorama, 3 fisheye snapshots 120° apart - for more
>> details on my method you can visit my website http://philohome.free.fr.
.I
>> join here "script véranda" which is the script I used to generate the
>sample
>> panorama, you can download the three fisheye images and try it yourself
>(the
>> result may be not so good because the images were heavily compressed for
>the
>> web)
>> - Using Photoshop I then crop the three photos to a square just on the
>limit
>> of the fisheye circle
>
>(*** I did this using a Nikon Coolpix 950 and 8mm fisheye lens
andPhotoShop
>5.5 ***)
## that's my setup too except I use Photoshop 5 - no difference here ##
>
>> The following step is facultative, but useful if you have to verify your
>> data...
>> - Add a new layer on each photo and place a colored arrow near 4
>> corresponding spots between photo 1 and 2, then photo 2 and 3, then photo
>3
>> and 1. Try to disperse the 4 arrows, but don't chose points too close of
>the
>> fisheye border, the distorsion is heavier here. These arrows are only a
>> visual aid, and have nothing to do with the coded arrows provided by
>Helmut.
>> - Using the Photoshop info, enter the points coordinates in the script
>file
>> opened in Notepad (I join an empty script file with the parameters I
>> optimize). Tedious, but easy...
>
>(*** Here's the 3 images I used after cropping and adding points to get the
>pixel coordinates ***)
>(*** http://www.swampcoast.com/help/index.html ***)
##Ok, seems all right that far... ##
>
>> - Run panorama tools optimiser. For a 2400x1200 panorama, the error
should
>> not exceed 5 pixels, 2 pixels for a 800 x 400 (small but fast to render).
>If
>
>(-- what error? Where is that at? **)
## I mean the figure given while the optimizer is running : "Average
distance between control points", which reflects the error between a perfect
fit (error = 0) and no fit (error > 1% of panorama dimension). Here your
distance is around 200 pixels when it should be around 2 to 5 pixels...
obviously something wrong, reflected by the value estimated by the value of
the field of view of the fisheye Hfov = 298° when it should be around 183°.
I then looked at your script and understood your problem :
( coordinates B & C ) <--- that should be coordinates C and C !
c n0 N1 x182 y362 X973 Y326
------------ | --------------
here you must have | here you must have
the coord of point C1 | the coord of point C1 (same point)
in the 1st image | in the 2nd image.
Try like this, should work much better !!!
Bye,
Philo
http://philohome.free.fr