h a l f b a k e r yNo servicable parts inside.
add, search, annotate, link, view, overview, recent, by name, best, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
On sites such as flickr (that one specifically, I use quite a lot), it is
usual to 'tag' photos with meaningfully related words. It's one of
those Web 2.0 folksonomy things, for one, and it helps you find stuff
(your own or other peoples), for another.
I propose an experiment. Build something
like flickr (or maybe
augment flickr itself). Implement tagging, tag clouds, tag searching,
etc. But don't use words for the tag. Only use pictures. There's plenty
of pictures on photo socnet sites. Use them. Associate each picture
with other pictures, and those other pictures thumbnails can
represent the visual manifestation of the tag.
Behind the scenes, the association presents useful data amounting to
'one or more humans think this is something to do with that', so we
know there's already a relationship of some sort (as opposed to
possibly none) which might make it easier to analyse for further
similarities or relationships in other untagged pictures, or pictures
that the tagger wasn't aware of.
*
http://nationalzoo....otos/fpo0720-09.jpg
[ldischler, Jul 01 2009]
(?) Almost full circle.. +
http://www.73553.co...ESKTOPS/DESK08.html
[daseva, Jul 01 2009]
Nice link, [daseva] !
http://timinator.fi...com/2009/06/tag.jpg
Tag, you're it [normzone, Jul 01 2009]
[link]
|
| |
I heartily endorse this product and/or service |
|
| |
I don't think the tag picture set will survive as different from the picture picture set - would you expect it to? It'll just be "these two pictures have something to do with each other". |
|
| |
It would be really interesting to see those connections that cannot be well expressed as words ("this cloud looks like this dinosaur"), but I wouldn't know how to filter those from the many that could ("both images show cake"). |
|
| |
If this was 'tag pictures with pictures as well as words' then you'd get my croissant. I say this because the words are more easily read by machine than images. By tagging a pictire with another picture the machine can associate one picture (and its tags) with another images (and its tags). |
|
| |
Actually this might work better for the average primate user--a universal website that isn't burdened by language barriers. |
|
| |
[-] I don't get it: you already have keywords which (presumably) can be used to xref to other piccies, are you suggesting using icons (in this case a closed set of pictures) instead of words ? |
|
| |
I agree with Jinbish, although I'll give you a doughnut for the concept. One reason why textual tags are important is that they are more generalizable, so you can have specific tags and tags that represented broader categories into which the subject fits. This is in addition to the increased computer-readability. |
|
| |
To modify flickr, there could be a simple addition of a button on a picture, "suggest related photo." |
|
| |
Why should the powerful and more mature world of the
picture be dominated, fenced in and shepherded around
purely on the basis of what smears, blobs and stains of this
laboriously clumsy speech stuff that they have had pinned
onto them? I mean, verbal language tries hard and all that,
but it's not all that. |
|
| |
A picture means something (or maybe
not - maybe it's just an image) which can relate to other
pictures, or at least, their meanings. Demanding that
pictures are sorted, categorised and processed according
to a completely different lexicon - that of words - is no
doubt highly insulting to the pictures, and probably
borders on racism. |
|
| |
Imagine a critical mass where large portions of the picture
atmosphere actually does hook up efficiently to large
other portions of it. The relationships could be expressed
in picture too. This cloud, is a picture, that dinosaur is a
picture, and we also need pictures that denote this looks
like that. Or a picture that denotes that two other things
show cake. Or more likely, more than one picture, to
perform those more abstract tasks. |
|
| |
It would be really interesting to see how the 'picture tags set' evolves over (a long) time. |
|
| |
Hypothetically: some pictures would quickly begin to emerge as 'ideal concept' symbols for basic, primary associations. These would become dominant representations of those concepts. Related concepts would tend to accumulate as relatively common combinations of more basic symbols. |
|
| |
The original 'ideal' pictures would start to get compressed, simplified and short-handed to maximise efficient re-usability, and could also start to accumulate new layers of meaning and association |
|
| |
You'd end up with a system that could look an awful lot like Chinese characters (Japanese Kanji) |
|
| |
These characters communicate differently from phonetic alphabets, bringing a whole 'tag cloud' of respresentational associations, many of which are recognised much more immediately and directly than we can manage by reconstructing words from phonetic letters ... which I guess is part of what [Ian] is trying to get at here. |
|
| |
(Interestingly - did you know that if you show a Japanese person a map of an area of Japan they don't know personally, they won't necessarily be able to tell you what the towns are called? The names' symbols can't necessarily be resolved into a specific phonetic equivalence ... rather they will say "could be A or B or C".) |
|
| |
//a universal website that isn't burdened by language barriers.//
Sadly ... as above ... I think you'd quickly find that it evolves into a new language which leaves non-accustomed primates scratching their chins bemusedly |
|
| |
[marked-for-tagline] leaves non-accustomed primates scratching their chins bemusedly |
|
| |
Damn it. This idea, I like it. Simple, and yet I can see this type of information sharing capable of fueling a whole new era of online communication. AI data miners may now have a source for adapting a smart, context-based visual field for their poor robots. So lost for words to express my appreciation for the futuristic, simple appeal of this. If only.... |
|
| |
Ian, I know what you're saying and I am with you (at least in the appreciation of pictures>1000 words type of way). I guess I've got my engineering hat on when I suggested the tag association bit because I was thinking of processing the whole malarky. |
|
| |
Words can be simple pieces of information: a couple of pieces of ASCII or whatever. These are simple to process, compare, index etc. Images are thousands of bits of information about colour, position, and so on. Let alone the aesthetic interpretation... or machine image recognition. Maybe I am reading more into it that you suggest. I'm thinking of tags as meta-data to be automatically processed. |
|
| |
Picture of newspaper tagged with CNN logo. |
|
| |
I can see it sort of working with suitable filters for tag type e.g. colour, shape, concepts represented etc, otherwise it would just be a confusing mess. |
|
| |
Myself, I like to tag Halfbakery ideas with pictures. |
|
| |
I forget what its called but it was set up by Luis von Ahn (I used to say van Aght because of the dutch). It requires users to match pictures in a similar way as to what is proposed. I.e tagging pictures with other pictures based on similarity. He also has a picture game in which you add text tags, and of course there are captchas/recaptchas. |
|
| |
This idea has a lot of relevance to computers and AI and human computational ability. And therefore to Turing, Boltzmann and Cantor. |
|
| |
If a picture paints a thousand words, and a similar picture another thousand, can't we extract some information from that, that would be greater than O(n^2)? ;-) |
|
| |
Interesting - and my PhD was basically a critique of the idea of linking as a set of 'one or more humans think this is something to do with that' tags - I argued for much more semantic content, in a way which would be usable across information domains. I think link semantics would become more importnat with image tags - does the act of linking mean "looks a bit like", or "reminds me of", or "is the same colour as"? |
|
| |
Well, you associate with something that itself denotes "looks
a bit like", or "reminds me of", or "is the same colour as" - as
the relationship node. Or, you associate with whatever it is
that it is a bit like or reminds me of etc - as the entity node,
or associate with attributes, etc. Or a more enwidened yet
fluid model, perhaps, as it's down to not a specific software
architects idea of a rigid schema, which presumes so much in
terms of inclusion and exclusive sets, but rather, an
aggregate of schema design across the populace. Designed by
the peep hole, for the peep hole. |
|
| |
Well, let's take it out of the domain of the photograph.
Let's say there was a way of perceiving dance as raw
comprehension (though not second-hand by perceiving a
video of some dance - but in reality, I suppose this'd have
to do). Now, if you witnessed Michael Jackson's
"moonwalk" you might be a bit amazed. Especially the first
time. But if you could associate that with Shalamar's "Night
To Remember", you might be able to add extra metadata
such as the date (1982) and some names (Jeffrey Daniels)
and demonstate that he did that same dance to that tune
on Top Of The Popes in that year. However, in order to do
that, one would enter the literary realm by describing it in
words (which itself would have a literary criticism payload
by inevitably qualifying and framing it - "oh, well of course,
Jackson didn't invent the moonwalk, he got it off Jeffrey
Daniels" etc (which now creates a value alteration of the
original experience of Jacko's moonwalk)) - you see, in
order to do anything reasonably abstract or complex, one
resorts to the leverage of verbal language. |
|
| |
What if there were a direct association (with minimal
loading of opinion or rhetoric) between Jackson's
moonwalk instance and Daniels rendition. And what if
there were further attribution tags to the amount of
professionalism Jackson employed, and the time that he
contacted Jeffrey Daniels to learn it, etc. And similarly,
what if there were a way of tagging the Shalamar
performance with Jeffrey Daniels striving to learn a dance
from a by now obscure 1930's dancer, demonstrating a
similar effect, and tags denoting research, study, historical
interest, hard work, perfecting a style, etc. |
|
| |
But all without using words. |
|
| |
That's etymology, innit? Just not with words... |
|
| |
So everything still remains invented by the Picts... |
|
| |
I find this concept extremely interesting, Before, we have had language to intimate etymology, before that I suppose we had verbal approximations at origin. With the advent of ubiquitous, pervasive and "beligerent" visual recordings, I would assume we are going to have to derive some meaning fromm motion. That meaning is necessarily going to be traced to its last recorded visual root. |
|
| |
Because pictures/stills are similar in some respects, I could imagine the same "etymology", if you will. |
|
| |
This is an interesting subject. As I see it, this boils down to a desire to start written language over again from scratch. Early written languages used pictographs rather than words which, eventually, proved inadequate for communication because they lacked context. Hence, over time, they were altered and embellished and gradually became full on written language. Although a picture may indeed be worth a thousand words, the problem is that it's not the same thousand words for each viewer. hippo's questions about what the 'tagging' actually means really gets to the heart of this, so I'm with kindachewy on this one.
However, at the heart of this issue there is a real problem I think. As I see it, the point of tagging is to allow people to gain quick access to information in a system where the data you are after is stored in many disparate styles and with many different purposes. It's not quite like going to the index of a library where there is a single purpose at work (i.e to find a particular book or author) and a uniform method of indexing. People use tags for all sorts of different reasons, to represent many different things and consequently there are ever more and more tags. Tagging is handy but not a reliable guide to navigation. Which rather illustrates, to me at least, the point of why written language has evolved as it has.
As a final aside, it seems to me that there are two different views that you could take on the way that written language seems to be evolving currently. On the one hand you could argue that things like txt speak and tagging are an attempt to simplify an overly complicated and difficult to learn set of modern languages or, on the other hand and somewhat more portentiously (and certainly more pompously), you could argue that they represent a debasement of language, to remove the shades of grey from it, and are a sign of the decline of civilisation. |
|
| |
It's not really driven by the desire to reboot language.
Maybe that desire exists in me, but this isn't really it. If
anything, this is borne of two things: |
|
| |
a] that words get all the attention these days. The
internet leans heavily toward the word as the granular unit
of communication, and if we're not careful, this will
delegitimate art of various forms as being a way of
communicating things that don't or shouldn't fit into words
(ie, art will become the ceremonial or the decorative, but
it won't be the message, because all you have to do is
write down what it means and have people talk about that
instead - the art itself disappears). |
|
| |
aa] tackling such a thing as a 'picture description language'
- much like PostScript™ is a page description language.
Except that I feel that in a way, analysing the content of
pictures from the inside is probably not the place to start.
Drawing inferences and connections between pictures and
other pictures might prove a better way, from the outside
- ie, without actually having to see inside the picture file
in any significant way. Although, maybe this semantic non-
verbal web of meaning would present a more refined
starting point for such subsequent internal analysis, one
after the other. |
|
| |
b] oh yes, and the thing that stimulated it into being was
the sheer amount of photos I put up onto flickr (which is a
selected subset of the amount that I scan from film I've
developed as a result of shooting it). I was thinking about
the titles. There's a space for the title of the photo, then
a caption, then all the tags, then people will caption it.
Why so much words? I really don't agree with titling my
photos. They retain the title that my scanner software
gives it as a sequence, but that's just mechanical cruft,
not me giving it an explicit title. I began to think that it
would be wrong to give the photograph - any photograph -
a title. As soon as it has the title, and perhaps a caption,
that title is more of a meaning than the photo is. It
outweighs the photo. It does all the communication, and
you start to think - why has he applied this curious photo
to this powerful title? Couldn't he have chosen a different
photo to go with this title? |
|
| |
It's partly because I production line my photography so
prolifically (which I never really realised until today) that I
never place importance in titling any of them - never mind
those ones, there's some more coming along soon. That
got me thinking where the 'weight' is. Without titles, it's
now in the right place - the disposable and brief photo
that communicates whatever it was occurred to me when I
took it. With a title, that would all be a staged lie. |
|
| |
I get what you are saying on your point aa] but on your point a] , isn't that because the internet isn't really the proper place to view art that originates from the 'outside' world? As things stand currently, it can only supply a two dimensional representation of an artwork. Viewing a picture of, say, Rodin's 'The Kiss, is not the same as seeing the sculpture in a gallery. The web is all very well but it's no substitute for the real thing! If, assuming for the sake of argument, art is a search for truth then how much truth can be garnered by associating a two dimensional representation of a three dimensional object with another such representation?
Ooh, you've added point b) whilst I was typing. Yes, I agree with that one! You should complain to Flickr and ask them to make titling & tagging optional instead of compulsory. |
|
| |
This reminded me that art is the job of the audience more than the creator ... a picture, presented naked and unadorned by words, stimulates a wide range of potential responses at many different levels. Some may be individual only, others may be the equivalent of in-jokes for smaller groups, yet others may tap into broader cultural, social and/or species roots. |
|
| |
It is only in the 'eye of the beholder' that these responses and associations are actually made and realised, and each beholder will take away a different experience, insight, learning or other impression. |
|
| |
Words (alone or as arranged groups) do the same thing as pictures, if somewhat differently. However, words tend to be more precise and may provide fewer dimensions of association potential, especially if arranged prosaically (in series, progressively narrowing potential associations) rather than poetically (in parallel, encouraging multiple allusions and associations) |
|
| |
Adding another element (a picture or word[group]) has simultaneous destructive and creative potential. Destructive, because it effectively 'focuses out' all the potential associations of the first that don't also match the second. Creative because it effectively 'amplifies' and focuses attention on all the potential associations of the first that DO have relevance with the second. |
|
| |
The job of the artist is to choose elements and arrangements carefully, so as to get the right balance between amplification and focus to maximise the chances that the beholder will experience something in line with what the artist intends. |
|
| |
[Human intelligence seems to operate mainly via the process of continually taking elements and putting them together to see if doing so results in a net destruction or creation of insight. Our sense of humour has probably evolved to encourage us to experiment with seemingly ridiculous linkages, discovering common ground worth exploring further] |
|
| |
As per [Ian]'s concept, a picture may need to be untramelled with words to speak for itself. However, adding anything else to it will impact on the way in which it does this ... whether the additions are pictures or words. |
|
| |
The process of tagging for discovery (which is typically a 'scientific' or 'technical' endeavour seeking precise focus rather than allusional richness) can interfere with the associations process, as the tags used are generally not selected for their 'artistic' impact. |
|
| |
So ... an intereting corollary of this idea emerges. We should create a search engine which ITSELF has artistic capabilities in sorting through, selecting and combining pictures, words and other elements that have greater potential to generate the optimal response/insight in the 'searcher' ... and a search interface which allows the searcher to define the kind of response/insight which they are looking for, as opposed to the specific content items which they want as a mechanism for getting there! |
|
| |
I see the point about titling photos, although there are many artworks where the relationship between the title and the artwork is part of the expression of the artist (e.g. Damien Hirst's "The Physical Impossibility of Death in the Mind of Someone Living", or René Magritte's "Ceci n'est pas une pipe").
Culturally though we are so accustomed to labelling things with words, that even when an artwork is labelled "Untitled" or if someone on Flickr leaves the title of their photo blank or with a automatically generated sequence number, it is interpreted as a conscious choice by the artist and the viewer reads some meaning into that choice and adds it to their overall interpretation of the image. So, there's no escape - even if you ignore titling pictures with words, people will still think your decision to do this in some way expresses something of your intention with respect to the picture. |
|
| |