Any Vision Plugin
A best practice when publishing or sharing your photos publicly is to accompany them with metadata. Specifically provide a Title or Caption describing the image and keywords to facilitate effective searches. I recently came across a Lightroom plugin called Any Vision that uses Artificial Intelligence to automate the process. John R. Ellis authored the plugin; you can find a download link and instructions for installing, configuring and using it on the Any Vision product page. You may wish to evaluate if automating Titles, Caption and Keywords fits your goals for your photography. Some may object to sending even low resolution image to Google. Or you may find constructing a well-formed prompt harder than writing a good Caption.
In his YouTube video This AI Prompt Writes Your Photo Metadata (So You Don’t Have To) Andy Hutchinson describes how he created prompts for Any Vision that incorporate additional Lightroom metadata. He included the prompts in the video description. Mr. Hutchinson lives in the part of the world that thinks May is mid-autumn and he directed AI to think that way too. So, I modified the original prompts in order to:
- Eliminate the emoji
- Eliminate the Technical Information
- Eliminate the call to action section
- Revise Australia related terms to US relative terms (e.g. months for spring)
- See exercise suggestion below
The plugin features suggest many interesting possibilities with which to experiment and explore for synthasizing image metadata using AI.
Considerations for Titles, Captions and Keywords
Caption
Use sentences and phrases meant to convey what you experienced and felt at the time of capture as the basis for your caption, and go beyond simply describing visual elements.
Title is an abbreviated version, usually a one-liner; imagine it exhibited beside your image. Or you go for the ever popular Title of “Untitled”.
AI Use Case
For your prized images take the time to document your experience creating the image, in order to setup an emotional response from the viewer. More generically for other photos, it may be sufficient to use an AI generated Caption. But in order to create a consistent style for your work, consider building a custom prompt incorporating existing metadata and actions. The Any Vision product page documents how to create a prompt.
Keywords
Apply keywords to a photo in order to make it easier to find again. Simple as that, ask yourself, “how would I ever hope to find this again”.
A common strategy to generate keyword ideas, is to first write a robust Caption and pull out individual words for searching. You may not create Captions first, that is OK. Ask the who, what, when, where questions to come up with words you would associate with the picture.
One factor influencing the words you generate, comes from recognizing the context of your work. For example, I photographed my brother holding his newborn granddaughter.
In my catalog I used keywords: Michael, Emma, birthday . Because I happen to know the people, how old Emma was and also knew it was at a special event, Jim’s Birthday party. And because I would most likely use these words when thinking about this photo again. No need to fret about it too much more. If I do think of something else in the future, I will simply add it as a keyword then.
AI Use Case
If this photo were meant for something like stock photo work, I would not include personal details and write a more generic phrase, “Man holding a sleeping newborn baby.”. But that would be too generic. In order to drive searches to your photos, you need to go stretch a bit. Think mood, or color; go beyond Who, What When, Where include Concepts, Emotions, Actions. Use these concepts to create an AI prompt. Here, I used the provided Any Vision prompt for Adobe Stock to create keywords. I also generated a Title to compare with my original sample texts.
This way of thinking can also be used when tagging for social media platforms.
Generated Title
A peaceful fall evening, a sleeping baby cradled in a warm embrace.
Generated Keywords
baby; newborn; infant; sleeping; cute; adorable; child; toddler; care; love; family; father; parent; human; newborn baby; newborn infant; newborn child; baby girl; baby boy; newborn girl; newborn boy; holding; arms; hands; soft; tender; love; sweet; peaceful; calm; comfort; tenderness; affection; close up; portrait; family portrait; newborn photography; newborn session
In the context of my personal work, I would not think to use words like soft, tender, love to search my catalog. But if I were sharing/publishing to a broad audience, then such keywords improve the chance of that audience finding my photo.
Other Considerations for Lightroom Keyboarding
Since you got this far, I thought I would share some things related to Lightroom Keywording. I recommend viewing Julieanne Kost’s excellent tutorial on Lightroom Classic Keywording
Always keep it simple, repeat the phase “how would I hope to find this again”. Over time you will create a large list of words and may wish to provide some organization or structure. Again keep it simple with some broad categories along the lines of WHO, WHAT, WHEN, WHERE. I mostly muddle around adding keywords, eventually reorganizing them in a way that makes sense to me. Julieanne Kost’s video shows her adding keywords in an ad hoc manner and later organizing them into a hierarchy.
Download my Keyword List template, which provides some “top level” organization. These topic headings appear in upper case and are marked not be exported. These top level categories also suggest questions to ask about a photo in order to come up with a description. The _FIX keyword is the special folder for all new keywords. I enter keywords free-form and periodically move the new words from _FIX into the proper organization. PEOPLE has attribute to be the folder for all new person keywords. Both of these are specific Lightroom attributes you can set. See the Adobe Helpx page Use Keywords for details on these attributes and how to import my Keyword List into your catalog.
Hashtag Exercise
In the prompt I provided for Captions, a set of hashtags appear at the end of the caption. This might be useful for certain social media platforms, but may not fit other settings. You can exclude them out for environments where hashtags in the Caption are distracting. As an exercise, remove the hashtags from the Caption I included, and create a new prompt to put them into the IPTC/ Extended Description (or any other metadata field that makes sense for you).
As a second exercise, create sets of enabled prompts targeting various publishing scenarios you may have. For example, create a set of enabled prompts used to generally get photos organized in your catalog. Create a different collection of enabled prompts when you wish to export/share/publish photos on a specific platform, where you might want to only add a richer set of keywords. You can write AI prompts dictating the audience and style of the response.