Keywording Strategies

[Hert]

The next update (build .147) has this checkbox removed and replaced with a drop down that has 3 options:

1. Write as keyword
2. Don't write as keyword
3. Private for the Catalog

[Hert]

it's understandable why we would call it the same.

Very much agreed....in IDimager this field behaved like Private field, even though the caption only said that it wouldn't write as a keyword. Supreme is a little better in its word-choice

[Mike Buckley]

2. Don't write as keyword
3. Private for the Catalog

When a catalog label is configured in IDimager to be ignored when writing to XMP and IPTC, will it be imported into Supreme as state #2 rather than state #3?

[tstoddard]

1. Write as keyword
2. Don't write as keyword
3. Private for the Catalog

Hert,

Just to be clear, are you saying that you are giving us the functionality to specify privacy on a label by label basis?

I find the wording of number 3 a little confusing. After some thought, it makes sense but I'm wondering if somebody who hasn't followed this thread would understand what you mean by "for the Catalog". Perhaps something like "Make Private (Catalog Only)" would be a little more obvious.

Also, as a test, I created a new category and made it private. I then created a new label. That label still can be configured to "do not create keyword for this label". I would assume that as of build .147 the drop down you've described will be the same for labels in private Categories and that this setting would just be ignored when the label is in a private Category. Is that correct?

[space]

@ Tom

In referring to your first post in this thread, the whole amateur/professional thing doesn't really seem to count for much - we all seem to be using PSu for much the same thing, as in managing an archive of images. In PSu, its all about "housekeeping" and how one cares to use the tools in PSu to perform those "housekeeping" chores. Its maybe whatever you choose to do with your images outside of PSu where the amateur/professional thing comes into play. Some may use Flickr to host and peer review their images. Others may choose to host their images on Smugmug, Photoshelter or lodge the images with stock picture agencies, this with the intent to license their images.

As for your comments about Locations and what to use or choose, there's this. All this is like going back to school and learning about sets and subsets - as in using circles and overlapping them to create subsets. In your example, you have Curacao as a Location. You were looking to separate an image that was specific to that location and another featuring your wife in the same location that you wanted to keep separate. Well, you have now have one 'circle' called Location and you can create another 'circle' called Family in which you have those pictures of your wife and may be of other family members. At some time in the future someone may ask about that picture or moment. Using PSu's Dynamic Search feature, enter in Location AND Family and there you will come up with subset of image featuring both Curacao and your wife.

Of course the above is something of an "over simplification". In keywording that particular restaurant image, you would enter the keywords "Curacao", "wife" and may be more keywords. And, this is the start of your hierarchy - as in Location and Family. Later you may want to "encircle" a set of images featuring events such a birthdays, special days and whatever and call this Events, and so it goes on. Of course, birthdays usually involve family members. So there, you would creating new "subsets" of images. May be this is what this "housekeeping" is all about, as in "creating" neat little piles on things - as in "neat" little subsets of images.

A good start has already been made in PSu with the default catalog categories - as in People, Places (Locations), Events, Objects and Styles. With People there are already several subsets there - as in Family, Friends and Relatives. As a further simplification, you may know a whole lot of Bobs, as in an uncle Bob, a cousin Bob jnr, a friend called Bob and so on. How to separate out this lot? So, therein you have Family, Relatives and Friends all under People and, thereby, create the necessary "separation" - as in looking for images of ONLY friend Bob and NOT the others. Boolean logic by another name.

Taking all this a step forward, there's this. I have this "circle" or set of images called "bands" (as in groups of musicians). I could leave it at that. But then, this is not going to work too well. And, here in lies the rub. There's this word "granuality" - as in grains. In this, how far does one need to separate things? In "bands" have a band called Muse, in Muse there's a guy there by the name of Matthew Bellamy. Could leave it there but don't. Bellamy is a "musician" and Muse is a "rock band" under "contemporary music". Both these subsets are found under the head category label, People. And, its usually goes further.

Arriving at the tail-end of this thread, glad to see these new upcoming features. While "Write as keyword" should be on by default, having the other two features is a nice touch. Need to ask this question, how are these new features going to affect the "Also assign its parents" switch?

Rogan

[tstoddard]

Regan,

Thank you for your input. I completely understand the concept of organizing and assigning my labels so that I can find specific images that reside in those intersecting sets and subsets. My query was more specific than that. I know that if I'm searching for images in PSU, I can simply use my labels to find what I'm looking for because they are easily accessed whether or not they have been written to my files as "keywords". What I'm trying to decide is for which labels I don't want to have corresponding keywords. Some articles I've read indicate that it is redundant to put names of locations in the keyword space if it is already recorded in the appropriate IPTC location space. When I'm using PSU this doesn't matter because when I search PSU will find a location name as long as I have a label created with that name. However, if I happen to be using Picasa, and I search, it appears that Picasa only looks at my keywords and captions and not at the IPTC fields. Therefore, the question of whether or not to write keywords is significant only in the context of what I might do with my files outside of PSU.

I mention being an amateur because I assumed that a professional will often have his or her photos in other places and those place will have their own methods for searching those files. I am curious about how those other hosting sites work. Should I, at some time in the future, decide to use some sort of external hosting site, I would like to know what the ramifications of my label configuration in PSU will be. For example, do stock picture agencies search metadata beyond keywords? Do they maintain their own database of keywords external to the files they host? If so, do they extract those keywords from the embedded keywords in the files or do the users provide specific keywords only for use by that stock picture agency?

Many of the features provided by PSU and other cataloging software are designed to make image files more useful when accessed outside of the cataloging application itself. Standards provide multiple ways in which to indicate the locations related to the image. My original question was about when to write those locations to keywords and when not to. And, if I decide to write them in some instances and not in others, what is the best way to accomplish that. I am leaning toward the suggestion that Jim made in his post early in this thread which was to have some location names duplicated under different category hierarchies, one for locations and another for subjects.

Perhaps it was a mistake to use the amateur versus professional context to begin with. The real question is whether my labeling structure is designed to prepare my files for my own use only or whether I want them to be useful to others at some point in time, and if it is intended for others, who is that audience? The keywords I would embed would be different If my target audience is my family versus the general public. Clearly, there is no one best practice that can anticipate these issues but in some instances, such as location names, there may be a strategy that works better most of the time.

[Hert]

Just to be clear, are you saying that you are giving us the functionality to specify privacy on a label by label basis?

That is correct.

Hert

[Hert]

When a catalog label is configured in IDimager to be ignored when writing to XMP and IPTC, will it be imported into Supreme as state #2 rather than state #3?

If I remember well, that setting in IDimager create completely private catalog labels. In PSU it will then be treated as #3.

Hert

[gcoupe]

The next update (build .147) has this checkbox removed and replaced with a drop down that has 3 options:

1. Write as keyword
2. Don't write as keyword
3. Private for the Catalog

I thought I was following this discussion, but now I'm lost. Does #3 mean that a Catalog label also won't be written as part of the ICS schema in the image metadata, whereas it would be for #1 and #2, assuming that the write ICS schema is checked in all cases?

[Mike Buckley]

When a catalog label is configured in IDimager to be ignored when writing to XMP and IPTC, will it be imported into Supreme as state #2 rather than state #3?

If I remember well, that setting in IDimager create completely private catalog labels. In PSU it will then be treated as #3.

Hert

The only way to make a catalog label completely private in IDImager is to enable ignoring the label when writing to XMP and IPTC and disabling writing to the ICS. So, I'm back to assuming that a catalog label configured as described in my question will then be imported into Supreme as state #2...unless I'm completely missing something.

[Hert]

Mike, it will be behave identical in PSU as to how it behaves in IDI. If you say that it behaves like #2 in IDI then it will be set to #2 in PSU. If it behaves like #3 then it will be set to #3.

[space]

Hmm, seems like we're running two different conversations here in this thread

@ Tom and back to Keywording Strategies

Apologies if I sounded patronising in my previous response but get your point now.

In mentioning Picasa, this is owned by Google and Google more or less "owns", well, at least defines SEO (Search Engine Optimisation).

For a moment here, let's forget about web pages, web page design, content development, blogs and whatever. Let's just focus on images. In SEO there are basically 3 metadata fields that really matter - content in the Headline field, content in the Description field and Keywords. Keywords tend to be the KEY in this regard. There's some indication that the Title/filename may serve to gain some points/ranking. However and personally, I'm not going to start creating filenames like "Curacao-at-night-2012-011.jpg". While this might work for some, its too far removed from the way I name and file my images.

Other than the "who done its", the IPTC would love to have image hosting sites, content aggregators and image buyers "respect" and make far better use of IPTC image metadata - especially when it comes to copyright. This is not really happening.

As for the placement of images on image hosting/sharing sites, stock picture agencies and the like, we have this. At Photoshelter for instance, they tend to "ingest" most of the content in the IPTC fields and from many but not all the EXIF fields. Photoshelter goes on to emphasize that only the content in the Headline, Description and Keyword fields and to some extent filenames are useful for SEO purposes. The rest of the data is of little interest otherwise.

With Photo Deck, where I now host my image archive, they make use of even less metadata - no EXIF metadata and just a small subset of the IPTC data.

On the other side - as in the picture agency/stock library side of things, Getty is BIG on keywords and content in the Description field as well as few of the more "specialised" IPTC fields. Alamy, on the other hand, almost strips out everything save for the Description, Headline and Keyword fields. Other than of some interest to image buyers, little "weight" is given to any content in the Description field. In Alamy they have a 3 tier keywording "structure". The first being limited to 50 characters (including spaces), the second to a 180 characters and the last is limited to 2,000 or so characters. In that first tier, which ranks the highest, here you may well put in something, "Curacao night Dutch Antilles Caribbean", etc. And yes, I too had to look up Curacao

The thing with Alamy is that run their own internal image search operations and one based around their own customised search parameters. They have no need for SEO and thus, no dependency on external search engines. Getty and others are a bit different. They have their own search facilities but also make use of external search engines to drive traffic to their sites. While Picasa (for obvious reasons - as in Google) and Flickr seem to score favourably when it comes to image search, others seem to struggle.

While entities like Flickr use some of the EXIF data - camera data and the like being of particular interest to this image sharing site, others seem to forsake EXIF metadata altogether. Conversely, Flicker seems to forsake content in the other key matadata fields. What works for Flickr may not work for Picasa, Smugmug, 500px and others. The point here is this, other than those 3 key metadata fields, there seems to be a decided lack of consistency on how metadata is deployed and used across the Internet. As they say, your mileage may vary from site to site.

As for Facebook and other similar image sharing services, you may indeed need to exercise some caution here. While conditions vary (as in reading those Terms and Conditions thoroughly) and, other than the most basic info - usually contained in the Headline and Description fields, seems that many of these sites tend to strip out all metadata including copyright info.

Long story short, make the best use of your metadata fields where you can - especially keywords. In your "Curacao" example, it may come down to this, as in creating the following keyword hierarchy and a catalog category for each of the following, as in Caribbean > Dutch Antilles > Curacao. Forsaking things like Greater and Lesser, Leeward and Windward, etc. - this is the easy part.

The next part comes in handling things like "night/night views", "street scenes", "landmarks", "famous buildings", etc. The easy way to do this is to lump all these lesser "categories" under "Curacao". However and at a later date, your may well end up with lots of "night views" of other locations as well. And so, I guess this is the way categories evolve and how the catalog grows.

Hope this helps - Rogan

[tstoddard]

Rogan,

Your response is extremely helpful to me. Thank you for taking the time to share all of that information. That's exactly the kind of information I was hoping to get when I started this thread.

You're correct in pointing out that there are two different conversations taking place in this topic. Since this is a PSU forum, however, I believe it's completely understandable how this topic could have led to a discussion of how to use PSU to create (or not create) keywords from labels. I am also impressed by the fact that a conversation like this resulted in an enhancement to the application in a very brief period of time.

Thank you for getting back to the original conversation. I hope others will read all the way through this topic so that they may benefit from your contribution to it.