Posts Tagged ‘keyword’

Picapp Annoyances

October 8, 2009

I searched Picapp for a picture of a Canon camera for a blog entry I was writing.

This is all I got under the “creative” category:

[picapp src=”0138/fe5a612e-1118-44af-80c1-ed7cde04b280.jpg?adImageId=4764774&imageId=141711″ width=”500″ height=”276″ /]

That’s a cannon – not a Canon!

Searching “editorial” got me lots of pictures of Maria Carey. I don’t know why.

There were also pictures of the Calder Cannons Australian Football team.

[picapp src=”2/f/8/f/TAC_Cup_Final_1e30.jpg?adImageId=4765234&imageId=6616144″ width=”500″ height=”380″ /]

Searching for “Canon camera” wasn’t much better. At least Daveigh Chase has a Canon camera round her neck. (Not that you can tell, but the caption says so.)

[picapp src=”0/b/c/7/Daveigh_Chase_HBOs_b08f.jpg?adImageId=4765354&imageId=6089211″ width=”500″ height=”750″ /]

There was nothing under “creative” for “Canon camera”.

It all goes to prove my point – keywords are pretty useless for searching. Partly because people can’t spell, and partly (my pet peeve) because they are typeless.

My second beef is that the “related pictures” strip under the picture when Picapp displays the picture on the blog often obscures part of the picture. Can’t it appear below the picture?

Tagging is Pointless

September 18, 2009

In WordPress that is.

Lightroom calls them keywords, WordPress calls them tags, but they are the same thing: values applied to information (with no type).

What I mean is that WordPress has a good full-text search feature so if your blog posts are mostly text it will find any post with a term whether it is a Tag or not.

I can’t see any point in adding Tags to my posts. I don’t think it makes them any easier to find.

Of course pictures in Lightroom are not so easy to search. It would so great if Lightroom could ‘recognize’ a Hino AK bus or a Boeing 747 in a picture as I can. Then I could throw all my keywords away. There has been work done on image recognition but isn’t in Lightroom yet.

I wonder if they are something that search engines look for? No – that’s too easy to game. The search engines favour content over probably-manipulated metadata.

Somebody tell me I am wrong!

Controlled Vocabulary

August 24, 2009

I think this is a good idea for the few users who are serious about keywording their pictures in Lightroom. It looks like it is designed mostly for Adobe Bridge, which I have never used, but it works for Lightroom too.

They divide the world into the following top-level categories:

Animals Horticulture
Archaeology House and Home
Armed Forces Housing and Dwellings
Attire Industry
Beauty and Well-being Landscape
Birds Law and Order
Buildings and Architecture Love and Romance
Business and Office Media
Climate and Weather Music and Arts
Communication Nations and Landmarks
Computers and Computing Professions and Occupations
Concepts and Broad Subjects Religion
Education Science and Technology
Emotions and Feelings Shops and Services
Farming Sport
Festivals, Celebrations and Holidays Textures, Patterns and Colors
Food and Drink Time and Seasons
Health Care and Hygiene Touring and Vacationing
Hobbies and Pastimes Transport and Travel
Homo Sapiens


I’d love to try it out but I will not pay £49.99 (US$73.99) for it. Essentially I have developed my own controlled vocab for my areas of interest and I doubt it can be as deep as mine for vehicles (land, sea and air) and places.

I think this is a very tough sell for them and I hope they are not betting their company on it. £49.99 for a text file?!?

I downloaded their demo file and loaded it into my Test Lightroom catalog. It is about archeology. I am sure it is fine but it is so far outside my areas of interest I would not know.

I get the impression though that the Controlled Vocabulary is more for submission to stock agencies than for, say archaeologists. Perhaps the time will come when agencies require keywording against this vocabulary.

Here’s my current keyword hierarchy. It’s still a work in progress but I think I have the structure right for my needs.


I am shocked it is a 148 page PDF file!

For security reasons I removed the [PEOPLE] branch.


August 16, 2009

I have made good progress looking at the small two digit numbers on Thai truck and bus license plates and seeing how they correspond to the province. I introduced the problem here.

Thailand Truck License Plate

Thailand Truck License Plate

I have gone through several hundred pictures from my Lightroom database where the province code and the province name (in Thai characters) are visible. It does seem that the two digit code identifies the province (changwat), 01 = Bangkok for example. I have never seen more than one code for a province or a code used for more than one province. (It’s a 1-to-1 mapping in CS jargon.)

Here is that I have so far:

License Plate Name Thai Name
01 Bangkok กรุงเทพมหานคร
13 Ang Thong อ่างทอง
14 Saraburi สระบุรี
15 Ayutthaya พระนครศรีอยุธยา
16 Pathum Thani ปทุมธานี
17 Nonthaburi นนทบุรี
18 Samut Prakan สมุทรปราการ
21 Prachin Buri ปราจีนบุรี
22 Chachoengsao ฉะเชิงเทรา
23 Chon Buri ชลบุรี
24 Rayong ระยอง
25 Chanthaburi จันทบุรี
27 Sa Kaeo สระแก้ว
31 Yasothon ยโสธร
32 Ubon Ratchathani อุบลราชธานี
33 Si Sa Ket ศรีสะเกษ
34 Buri Ram บุรีรัมย์
35 Nakhon Ratchasima นครราชสีมา
36 Surin สุรินทร์
45 Khon Kaen ขอนแก่น
46 Kalasin กาฬสินธุ์
47 Maha Sarakham มหาสารคาม
60 Uttaradit อุตรดิตถ์
63 Phitsanulok พิษณุโลก
64 Kamphaeng Phet กำแพงเพชร
68 Uthai Thani อุทัยธานี
70 Suphanburi สุพรรณบุรี
72 Nakhon Pathom นครปฐม
73 Ratchaburi ราชบุรี
74 Samut Sakhon สมุทรสาคร
75 Samut Songkhram สมุทรสงคราม
License Plate Name Thai Name
1 Bangkok กรุงเทพมหานคร
13 Ang Thong อ่างทอง
14 Saraburi สระบุรี
15 Ayutthaya พระนครศรีอยุธยา
16 Pathum Thani ปทุมธานี
17 Nonthaburi นนทบุรี
18 Samut Prakan สมุทรปราการ
21 Prachin Buri ปราจีนบุรี
22 Chachoengsao ฉะเชิงเทรา
23 Chon Buri ชลบุรี
24 Rayong ระยอง
25 Chanthaburi จันทบุรี
27 Sa Kaeo สระแก้ว
31 Yasothon ยโสธร
32 Ubon Ratchathani อุบลราชธานี
33 Si Sa Ket ศรีสะเกษ
34 Buri Ram บุรีรัมย์
35 Nakhon Ratchasima นครราชสีมา
36 Surin สุรินทร์
45 Khon Kaen ขอนแก่น
46 Kalasin กาฬสินธุ์
47 Maha Sarakham มหาสารคาม
60 Uttaradit อุตรดิตถ์
63 Phitsanulok พิษณุโลก
64 Kamphaeng Phet กำแพงเพชร
68 Uthai Thani อุทัยธานี
70 Suphanburi สุพรรณบุรี
72 Nakhon Pathom นครปฐม
73 Ratchaburi ราชบุรี
74 Samut Sakhon สมุทรสาคร
75 Samut Songkhram สมุทรสงคราม

I have 32 out of 76 provinces – 42%. To find more sans travel I need to look at my older pictures in ThumbsPlus.

I added a Region field to my little Lightroom Vehicle Information Plugin so I can record the information in the database. I have done it in a general way so I can record the same data for license plates of other countries.

New Lightroom Region Code

New Lightroom Region Code

I want to learn how to “inject” these name-value pairs as typed Vehicle Tag keywords into pictures that I export as Jeff Friedl does for his Geocoding information:

veh:fleettype=Transport Company
veh:region=35-Nakhon Ratchasima

The Truth About This Whole “Cataloging” Thing

August 16, 2009

Adobe Photoshop Lightroom Killer Tips » The Truth About This Whole “Cataloging” Thing.

Here’s a thought provoking post from the “Lightroom Killer Tips” site.

A few thoughts of my own:

  • I am not “normal”! I actually enjoy keywording my pictures. I enjoy both the process – knowing that I have a keywording scheme that works for me. I get a kick out of Lightroom’s attempts – 75% successful – to predict what keywords I’ll want to add based on past behaviour (the Keyword Suggestions panel).
  • But when it gets it wrong I want to scream at it: “No! That photo already has a Place tag. Now it needs a Type tag. Don’t show me any more Places.”.
  • Keywording is like knitting, it can be therapeutic if your keywords are well organized and you have a plan.
  • I hope the engineers who worked on Lightroom keywording read this. Yes, as Matt says in his post

Nobody keywords. I know that lots of people commented on it here last week but trust me – you’re in the extreme minority (and I mean serious keywording, not just casual “I do it once in a while” keywording).

But I use them to the extent that I am stretching the Lightroom SQLLite database and the user interface to its limits.

  • I enjoy the knowledge that I can find any picture I have ever taken based on the vaguest of criteria because my keywording is “perfect”. Thus when I was trying to work out the system of license plates for Thai buses and trucks I could locate all candidates very quickly and see if I could extend my list.
  • That’s why I wish I could keep every photo I have ever taken in one Lightroom database. See here and here.
  • I use operating system folders to organize photos by date taken. It is easy to set up during the Import function and occasionally I want to look at pictures I took on a certain date. Yes, I could use Smart Collections for that task, but I’d have to set them up.
  • I liked ACDSee‘s Calendar View for the few months I used it. I wish Lightroom had the same. I like to analyse statistics on my photography and numbers of photos taken by day / month / year is useful.
My Lightroom Statistics

My Lightroom Statistics

  • When I used Thumbs Plus I had a completely different folder structure: Camera Identity and then a folder for each block of 1,000 pictures.
Folder Structure for ThumbsPlus

Folder Structure for ThumbsPlus

Here’s the folder of pictures taken with the Canon EOS-30D that I still have catalogued in ThumbsPlus. Each contains approximately 1,000 pictures. Each folder has a sub-folder called “Edited”. That’s because ThumbsPlus does not support the non-destructive editing or stacking like Lightroom so any edits I made went into a separate folder.

  • I accept that some time, maybe by the end of the year, I will have to split my Lightroom database. Doing it by date is the easiest way and trivial if I have the pictures organized by date in an operating system folder.

I am happy with the way I have things set up now. It is great that Lightroom supports my working style (workflow ugh!). But I’ll never criticise anybody for using a different scheme.

And I know I am not a mainstream photographer because of my interest in information management / databases that’s as strong as my interest in photography.

Lightroom Keywording Bug

August 4, 2009

I have become the victim of my complex Lightroom keywording scheme.  It’s grown to over 8,000 keywords (including synonyms). I organized them into a complex hierarchy to enable me to find photographs by two major criteria:

  • WHERE it was taken.
  • WHAT it is of.

I can go into more detail of the structure and the reasons for it in another post. Most of the time it words very well and I can keyword my pictures down to a huge degree of detail.

I wanted to mention here a bug in Lightroom that can be hugely frustrating. What happens is that I add an already existing keyword to a picture but Lightroom creates a new keyword for it instead.

It happens for a small number of keywords. The example I present is for a “Mazda 3” automobile.

The keyword for Mazda 3 is actually deep in the VEHICLES top-level hierarchy in my keywording scheme. The hiewrarchy is


However carefully I apply that keyword, Lightroom invariably creates a NEW keyword in the wrong hierarchy:


I do not know why it chooses that particular position. Other hierarchies have a similar structure, for example VEHICLES > AIR.

It does not happen for all keywords – else my tree would be in a real mess. It seems to happen for 1 or 2 character leaves. It happens for the Mazda 6 – but you don’t see many in Thailand. I have also had it happen for models of Hino truck which are two letters, e.g. FE or FF.

I avoid photographing Mazda 3s so as not to have to clear up the mess.

I find I have made the problem worse since I added another level to the tree structure for certain vehicle manufacturers. Companies like Mercedes-Benz have many model names for their cars, buses, coaches, and commercial vehicles. To make things clearer when I browse the tree I added categories CAR, BUS-COACH, TRUCK. I did the same for some other manufacturers like Volvo. I do not export those names – they’re only there to make the tree easier for me to read.

Now when I see a Volvo 940 and add the 940 keyword under Volvo it ends up keyworded as a Mercedes-Benz CAR with a new keyword.

Clearly I have exceeded the design specs of the keywording process either in number of keywords or the breadth of my trees. It is failing in a strange way – not a crash, but a logical error. At least it is reproducible: not caused by an alpha particle.

Maybe I have to reduce the depth of my keyword tree. Lightroom seems to have no problem with the number of keywords, but once you allow users to build trees it gets hairy. (To mix my metaphors shamelessly).

I could circumvent the problem if Lightroom had an option to prevent users from adding new keywords in the Keywording panel. I suggested that weeks ago here.

Keywording and Tagging in Lightroom

July 27, 2009

The Name Game – Keywording and Tagging in Lightroom | x=blog+stay+informed.

Here’s one of the most comprehensive discussions I have seen. But Brandon’s requirements are completely different to mine so he has taken a different approach.

It’s good that the keyword / tag concept is so flexible that it can accomodate everybody’s needs. But without a plan you end up with a mass of keywords and aren’t much the wiser.

I’ll write more about my scheme in another post. It seems to be standing the test of time in that I can keyword my photos now without finding I have to change the design. Lightroom’s “keyword suggestions” seem psychic sometimes – but at others remarkably dumb.

Vehicle Tagging

July 23, 2009

I have started to add what Flickr calls “machine tags” to my pictures of road vehicles and railways. The concept is similar to aerotags for aircraft and geotags for location information.

So far I have defined the following tags:

veh:man= (manufacturer)
veh:model= (model)

rail:man= (manufacturer)

I used “man” rather than “make” for compatibility with aerotags.

I did not have to do much work in Lightroom to add them. I added the tags as synonyms for the keywords I normally apply and set the option for Lightroom to export them.

Lightroom Ford KeywordLightroom Focus Keyword

The purpose is to add some structure to the normally “typeless” concept of a tag or keyword. It makes searching easier. For example a picture with tag “veh:man=Ford” is definitely of a vehicle made by Ford Motor Company. A tag of “Ford” could be a that or it could be picture of a river crossing.

Here’s a picture of a Ford Focus in Lightroom with the tags that will be exported.

Lightroom Ford Focus keyworded

Click for a larger version

(Note: I also added Thai language synonyms for “Bangkok” and “Thonburi” just to see how well Lightroom supports Unicode and how well they export to the web. That’s not relevant to this discussion.)

Maybe I’ll if I can get others who take pictures of vehicles to adopt the format. One way would be to create a group on Flickr similar to the aerotagging and geotagging groups.

I can add more names later. For example: “veh:body=sedan” and maybe “veh:use=taxi”. But for now I am keeping things simple.

Flickr recognizes the tag format “prefix:name=value” and hides them in a tag list unless you elect to show them.

Here’s how the photo shows on Flickr:

Click for a larger image

Click for a larger image

There is no registry of machine tag prefixes, so someone could come up with their own set of tags that are incompatible with mine. That could be confusing.

The Trouble With Keywords

July 16, 2009

Is that they are typeless.

I guess I’d better explain that.

When you apply a keyword to an photograph or anything else it is a single value, usually a string. For example “Bangkok”.

That tells the person doing the search that the photo has something to do with Bangkok. That’s okay as far as it goes but what about “3”?

I have the keyword “3” several times in my Lightroom database. Sometimes it means a Mazda 3 car. Other times it refers to a bus on route #3. But that shows up in at least three bus companies I have pictures of.

In Lightroom it keeps track of the position of the keyword in the hierarchy I have built – aka nested keywords.

So I have


PSVs-Operators-Asia-Thailand-Bangkok MTA-routes-3

PSVs-Operators-Europe-England-London-Transport for London-routes-3

and I can distinguish them because of that hierarchy. But the hierarchy is lost when Lightroom exports the pictures, say to Flickr. On Flickr keywords (or tags as Flickr calls them) are just values with no structure. So I will just see “3” in Flickr’s tag list.

Combine that with the fact that many people don’t bother tagging their images, tag them inconsistently, mis-spell tags and even apply the same tags to all images whether they are relevant or not, tags are pretty useless for locating pictures.

Groups, which are akin to Lightroom Collections, are much better especially if they’re moderated, because often a human has ensured that everything in the group is relevant to it. Still mistakes slip in – I have seen many Skytrain pictures in a group on Subways for example.

Some people have tried to impose some order on keywording with efforts like aerotagging and geotagging.

I know aerotagging the best. In addition to adding the keywords Boeing and 747 the poster of the picture adds aero:man=Boeing and aero:model=747.

In computer science this is a simple example of typing. I know now that the second aerotag is an instance of the type “aircraft model” and not “aircraft construction number” (aerotag aero:cn=747).

Geotags are similar for encoding position information (lat/long) as keywords.

Lightroom’s keyword synonyms are excellent for aerotagging. I have “aero:man=Boeing” as a synonym for my keyword “Boeing”. Every time I add Boeing to a picture it is exported with the correct aerotag as well.

I have done with same with locations. So for example London Heathrow Airport has synonyms “aero:airport=EGLL” (aerotags use the 4 character ICAO airport code rather than the 3 character IATA code that’s used on passenger related stuff like tickets and baggage tags.)

Lightroom lets me add any number of synonyms to each keyword and to decide if they are exported or not.

As Lightroom supports a keyword hierarchy I have “747” as a child of “Boeing” so when I add keyword “747” to a picture it will be exported with

Jumbo Jet

and a few others. This also provides a modicum of protection against errors. I bet I’ll find as many pictures of girls if I search Flickr for “pretty gril” as I would for “pretty girl”.

If I have more than one keyword with the same value – like the Mazda 3 and BMTA bus route #3 then Lightroom gives me the context when adding the keywords so I am sure I am adding the correct instance of 3. (There is a Lightroom bug here but I will talk about that elsewhere).

Andrew Ferguson has a site devoted to aerotagging here. There’s a Flickr pool for aerotags here.

And Flickr keep a list of their most popular tags here.

Lightroom Import (Part 4)

June 27, 2009

These days I travel virtually since I don’t have the opportunity to travel in the real world. Sometimes I import to Lightroom photos I did not take from sites like Flickr.

If the original photographer has put keywords on her image then Lightroom attempts to import them. Sometimes it makes a real mess of it – as in the illustration above.

It failed to parse the keywords and made one very long one. This is the confirmation dialog box I got when I tried to delete it. I wonder if Lightroom can get a buffer overflow in this situation.

I think this shows what people complain about in the IPTC and XMP specifications – different photo apps are not truly inter-operable yet.

In my case I want to put my own keywords on my images. Even if Lightroom had parsed them correctly I would have deleted 95% of them. It did me a favour as I didn’t have to seek tham all out.

But what I want is an option in the Lightroom Import process not to import existing keywords from an image.

Sometimes I import photos I haven’t taken.