[KPhotoAlbum] a working photo interface

[Shawn Willden]

On Friday 09 February 2007 00:30, Heinz Kohl wrote:
> What about this man machine interaction?
> How many windows flowing around do you see?

You've referred a few times to a profusion of windows, and I keep going back 
to KPA and trying to find what you're talking about.   KPA really never has 
more than two windows open at a time, as far as I can tell.

> Using linux I'm handicapped by the many incompatibilities given by
> imperfect and nonstandard graphic file encodings given by Sane, Gimp and
> other graphical linux tools, even under known names like tif, gif and jpg,
> which is making it complicated to interchange graphical data with the
> programs of the international graphical software leaders (e.g. Photoshop!).
> My children, graphical professionals, are reporting about .jpg's and .gif's
> from me, which could be viewed only in firefox and .tiff's, which are
> overstraining even market leader graphical analyzing programs, hardened for
> extreme defective and even partial graphical data.
> On a long term this can make my photos unreadable.

Really?  I've never heard of anything like that.  It's off-topic for this 
list, but I'm also a sometime GIMP developer, and I'd be *very* interested in 
seeing, for example, a JPEG or TIFF that GIMP produces which Photoshop or 
other Windows image viewing/manipulation programs can't open.

> Would it really be natural to make a category entry, which will be used
> exactly once, just to make possible another order than a time based order?

What non time-based categorization are you thinking of that would have such 
single-use categories?

> Is it naturally to use a (time-)slider holding a prominent place having no
> equivalent tool, because the part of an order key criterium is strictly
> bound to time, even in absence of usable time information?

No, but there usually is usable time-based information.  In your case, for 
example, even if the time the photo is taken is known only inexactly, it 
seems quite important to record whatever time information can be deduced -- 
even if it's only a year, or a decade.  KPA supports ranges of times 
specifically for addressing such inexact times.

Your other example -- a collection of photos depicting various species, also  
has useful time-based information in many cases.  Most researchers do care 
about where and when such photos were captured.

> Is it naturally to do totally with unordered data, when there is no time
> order but e.g. a natural number ordering?

Like what?  Stop talking in abstractions and give specific examples.  I think 
KPA handles very nicely the specific case that you started with -- scan batch 
numbers.

If you can describe actual, useful scenarios that KPA can't presently handle, 
then you'll see more interest in addressing those limitations.

> And, there's a further lasting problem: how to navigate in such a number
> heap?

By searching.

Answer me a question, Heinz:  In the case of your batches of scanned 
historical photographs, what *useful* information do you derive from seeing a 
numerical ordering of the batches?  Is there some case where you would 
actually want to quickly find the batches that are between batch 243 and 
batch 260 for example?  And would such a case be needed often enough that 
scrolling down a (perhaps long) list of batch numbers would be inadequate?

> My example using KPA:
> up to now only 10 Categories (following to the proposals I had to implement
> a 10 more).

What are these 10 categories?  I think it's likely that you're not using the 
tool correctly, which is making it more complicated than it needs to be.

> To make an entry is posing the following problem, description only of the
> very first phase:
>    upfolding of all categories.
>        result: 10 windows, one over the other
>         all at address (0,0), upper left corner
>     appopriate positioning of all windows (min. 30 seconds).
> **saving of the settings (worthless - X is ignoring global window
> positioning, and that's so in most implementations since the founding of X
> in the 1970ies).
>   seeking (positioned windows are disappearing).
>
> No, KPA can't change X - but it could make use of working parts of the GUI.

Where are all these windows coming from?  KPA doesn't do that!  Are they from 
some application that you're building using the KPA index file?

As an aside:  It sounds like you need to get a better window manager.  Mine 
handles window positioning very nicely and very flexibly.  X isn't 
responsible for window placement, the WM is, and there are many nice WMs that 
do the right thing in nearly all cases and allow you to override the other 
cases.

> ... so, after making the first entry (a 3 minutes only for positioning
> struggling), next search, beginning often, rather typically with a
> mismatch, because it's so costly and error prone to do all markers back,
> and there's no quick overview over the state.

You're obviously not talking about KPA here.  What are you talking about?

> > ... categories you enumerate a new sub-category simply by naming it when
> > you apply it to a photo.
>
> Yes, and even doing that is proved to be unnecessary clumsy, and the
> (pre-)ordering according to time is inappopriate.

What's clumsy about it?   Can you describe the process you're using (in 
detail, please)?  I'm sure that we can tell you how to do it quickly and 
efficiently in KPA -- fast and easy categorization is one of the main 
development goals, and it does work very well.

> Let's explain.
> After some years of work - up to over 1000 fotos a day, up to 100
> classifications per day, some of them from accounted experts - some guy has
> today 14141 insect photos of minimally 1764 species.
>
> O.k., it's fair for him to order them, e.g. one of them as "hexapoda,
> insecta, pterygota, neoptera, odonata, zygoptera, coenagridae, pyrrhosoma,
> nymphula"

You keep using this word "ordering" -- but using it as though you 
mean "categorization".  I think we have a communication problem.

> - but "pyrrhosoma nymphula, imago, 30.6.2002, #23" should say 
> enough and give a good ordering.

Sure, so he should create KPA categories for "Classification" and "Scan Batch" 
(assuming the 23 refers to that).  Then, for that photo, he should 
enter "pyrrhosoma nymphula" in the Classification field and "0023" in the 
Scan Batch field.

If he wants to, he can also create sub/super categories 
within "Classification" that describe the entire taxonomy, which would allow 
him to, for example, display all images of odonata, including pyrrhosoma 
nymphula, plus whatever others he has.

When categorizing a specific photo, though, he would not have to specify the 
complete tree, just the leaf (pyrrhosoma nymphula).

How elegant is KPA's categorization scheme!

> It would be complete nonsense to look at these pictures in chronological
> order. But to have a slider to find the

Nonsense?  Then why is there a date in the example file name you mention?

> But even names like "pyrrhosoma nymphula, imago, 30.6.2002, #23" alone
> would give an acceptable alphabetical ordering. Sometimes I am also
> photographing insects, but to store "pyrrhosoma nymphula, imago, 30.6.2002,
> #23" will be much better for me than to store "hexapoda, insecta,
> pterygota, neoptera, odonata, zygoptera, coenagridae, pyrrhosoma, nymphula,
> imago, 30.6.2002".

Sure, and KPA makes that very easy *and* allows the complete hierarchy to be 
implied by the single selection.

> But only, when not intermixed with the pyramids of Gizeh.

Luckily, KPA does not require that every photo have a place in every category.  
In this example, pyramid photos would simply not have anything in 
the "Classification" field, and so when you drill down through 
classifications, pyramids would not appear.

Again, I think KPA, as is, handles this perfectly.

> > It's also worth noting that when KPA presents a list of categories they
> > are sorted in lexicographical order, so if you name your batches with
> > leading zeros (i.e. 00123 instead of 123), the list will be sorted in
> > numerical order.
>
> As I said, the naming of my files is devoted to my relatives, and they have
> to work even outside of the context and outside of linux.

You misunderstood me.  I was not referring to the naming of files, but to the 
naming of categories inside the "Scan Batch" category.  Filenames are 
irrelevant to categorization in KPA.

	Shawn.