Re: [gentoo-user] merging or fitting images together

[Michael Mol]

On Sat, May 19, 2012 at 12:38 AM, Philip Webb <purslow@××××××××.net> wrote:
> 120518 Michael Mol wrote:
>> Remarkably simple. Probably because I was only stitching two photos.
>
> -- details snipped --
>
> Thanks : that gives me a 3rd method to pursue.
>
> NB in your result there are some badly curved lines :
> bottom right, the front of the tram is badly distorted ;
> centre top, the sides of buildings are curved outwards ;
> also, the bottom of the photo has been lost, eg the L-side man's eyes,
> & the result if smaller than the other  2  results achieved earlier.

It's inevitable that you're going to lose some of the image. That's a
function of reprojecting the stitched image.

The distortions are very probably due to an incorrect focal length
setting--something that's going to be impossible to get correct. But I
likely could have corrected by forcing Hugin to treat it like lens
aberrations, and getting it to correct for it that way. That would
indeed take a great deal of time.

>
> No complaint at all ! -- but clearly all methods require some practice.

The problem here is that there's missing source data. (Details below)

> I've added your result to my I/net 'test' examples (I hope that's ok).

np. I was going to share an ImageShack link, but I realized I wasn't
sure whether by "keep the image off-list" you meant "don't attach the
image" or "don't show the image on the list".

>
> Any other suggestions are welcome -- apparently this is of interest -- ,
> but I will turn to other priorities & investigate panoramas a bit later.
>
> BTW the location is Steelhouse Lane with Snow Hill Sta in the background
> (I stated it incorrectly before) in May 1953 just before the final trams.
> The photo was taken with a Zeiss Ikon camera, a well-reputed make :
> perhaps you can find the focal width on the I/net somewhere.

Now here's where the fun begins. According to Wikipedia, the Zeiss
Ikon is 35mm SLR...but that's about all you're going to get from it.

Really, everything else of interest is in the lens.

Being an SLR, the lens can (and will) be swapped out by the
photographer as circumstance demands. Each lens is going to have
different aberration characteristics, but that's not nearly as
important as the other difference: Without knowing the lens used, you
know next to nothing about the focal length and field of view. (The
two values can be derived from each other, as long as you know the
frame size...which we do.)

Worse, if the photographer was not using a prime lens[1], and was
instead using a lens with variable zoom, you can't easily know what
the real focal length was, as this will change depending on how far
the photographer has zoomed in. Now, I suppose that if you knew the
physical sizes of a couple fixed lines in each picture, where the two
lines were some not-insignificant distance apart, you may be able to
roughly calculate the focal length.

But, really, without knowing the focal length, getting the stitch
right is going to be guess-and-check.

Incidentally, this is one reason why digital photography is awesome.
Almost everything interesting you may need to know about the shot is
going to get stored in the EXIF data in the image files. My camera
stores the lens focal length at the time of snap; if I have a zoom
lens on, it records the exact focal length the lens happened to be on.
It's quite nice. :)

[1] This isn't "prime" as "excellent" or "high grade"..."prime" in
this context means it has a fixed focal length. It may have additional
implications, but that's the largest functional relevance: a "prime
lens" is a lens with a fixed focal length, a lens which doesn't have a
variable zoom capability.[2]

[2] I'm dribbling in a lot of semi-relevant technical stuff in here
for those who are following the thread for informational purposes.
-- 
:wq