> Hello again, 
>
> I've read the excellent tutorial of Ben Kreunen and I admit that nearly 75% 
> of what I've read was complete chinese to me. Do I need to learn something 
> first and then only read that tutorial ? 
>
> Fabio. 

Hahaha,

PanoTools is chinese.
It's all a hoax to make people believe there's something great and that they're missing out because they don't understand it at all.

No seriously, we probably all pulled hairs when we first started using them. It's a very steep learning curve. It took me a week to stitch my first three pictures. The good news is: once you do understand them, you really feel you've gotten insight in how things work.

I once searched for a good step by step tutorial, but never found it.

Your mailer tells me you're using windows. The easiest thing for you to do would be to download ptgui and have a look at the scripts it creates.

The basic concept of stitching panotools is:

input some pictures using i lines
        using those i lines you can tell the optimizer anything relevant about your image.
        http://www.path.unimelb.edu.au/~bernardk/tutorials/360/readme/i_line.html lists all options.
        basically you input the name of the image, the hfov (field of view), the type of lens and an estimated yaw, pitch and roll

        a common thing to do is to define parameters that are the same in all images like this:
                say image 0 and 1 have the same field of view. You would say in your script:
                i n"image0.jpg" f0 y0 r0 p0 v120
                i n"image1.jpg" f0 y30 r0 p0 v=0
        this tells the pano tools that the second image has the same field of view as the first one.
define what pano you want to have
        using a p line
        perhaps rectilinear (like an ordinary photo), or a psphere or a cylindrical view. Also specify the width and height and output format (like jpeg or photoshop).

        http://www.path.unimelb.edu.au/~bernardk/tutorials/360/readme/p_line.html lists all options
        add an m line to define whether you want to immediately create the pano after optimizing (m p1) or you want to just optimize (m p0).

        http://www.path.unimelb.edu.au/~bernardk/tutorials/360/readme/m_line.html, you've guessed it, lists all options for the m parameter.

define control points
        using the c lines.
        http://www.path.unimelb.edu.au/~bernardk/tutorials/360/readme/c_line.html lists all options
        ptgui, ptpicker or pixelpicker can help you create the c lines
optimize
        you tell the stitcher to optimize using a v line
        http://www.path.unimelb.edu.au/~bernardk/tutorials/360/readme/v_line.html lists all options.
        normally you optimize the yaw, pitch and roll of all images, except for the image you consider the baseline image.

        if you're not sure about the correctness of the field of view you entered for your lens, optimize the v parameter.

        optimizing the v-parameter is best done in 360 degrees shots. optimizing it in partial pano's often does not work too well.

        yaw pitch and roll are defined by the parameters y,p,r
        lens distortion can be optimized by optimizing the a,b and c parameters.
        if you moved the camera between shots or you haven't rotated around the nodal point of your camera, you can also optimize the d and e parameter.

       
The optimizer will optimize all parameters you asked it too. in your script file the i lines will be converted to o lines. those will be read by the stitcher later on. The optimizer also will put a table in your script file displaying the average control point distance. Something like this:

# Control Point No 0:  1.4862
# Control Point No 1:  0.487042

This would be ideal if all distances would be 0. Those distances that are really big (over 5) probably need to reviewed. Perhaps a wrongly placed point or the optimizer failing at optimizing one of your parameters correctly.

When you're done optimizing and all control point distances are to your satisfactory go stitch. When your images suffer from differences in colour or brightness fiddle with the k parameter of the p line.

Serge.