Previously, I discussed methods of clustering photos into geographically and temporally related groups, and some of the trade-offs involved in various clustering algorithms and parameters. Since then, I have arrived at a workable set of defaults, and created an interface that allows clusters to be merged and split. I have not yet built a tool for custom cluster editing, and the split tool does not allow the user to select a split point. Instead, it uses k-means clustering with k=2 to try to find the most logical point to split the cluster. This is a user interface issue more than a technical issue – I am not yet sure how to integrate a more customizable cluster creation tool into the application.
Here are some ideas for future development of Vignette. These have been floating around in my head and in my notebook for a while, and it was time that they got written down. Elaboration and blog posts on implementation of some of these features are to come.
- Conversation is important, and a (significant) step up from conversation with a chatbot is conversation with a real person. Creating a “photo-album” mode for video chatting that allows all users to flip back and forth through a jointly controlled photo album while talking and sharing stories. Optionally, the conversation could be recorded and associated with particular photos to create a kind of oral narrative to accompany viewing those photos in the future. You can almost think of this as similar to current online photo album sharing and communication tools (e.g. Facebook and Google) with the advantages of audibility and immediacy.
- Photos are important traces of our lives, but so are all of the other digital artifacts we leave behind. In scrapbooking, one often uses boarding passes, ticket stubs, and other physical artifacts to tell the story. Those things are now stored on our phone – why not archive them into the photo library on the phone so that they can be viewed and remembered like photos?
- Simple interactions can help people get more out of their photos. Trigger daily reflection on a mobile device at the end of the day when it detects a picture has been taken. The interaction could be brief and straightforward, with questions perhaps being similar to my Twitter bot, @SnapshotReflect.
- Take the conversation engine from Vignette and make it spoken, with text-to-speech and speech-to-text.
- Use the spoken dialog to create better video slideshows, that involve the user telling their own story.
- New output formats from the written story in Vignette, including low-cost printed pamphlets, video slideshows, interactive websites, and a “journal mode.”
- Suggesting stories within large photo archives by
- Finding loops/journeys that end where they started after traveling a long distance
- Finding the first time a person appears in a photo, or when someone appears after a long gap
- Finding the first time a place appears in a photo, or when the user returns after a long gap
- Physical computing interfaces that allow users to browse their photos on real pieces of paper, sort them, shuffle them, flip them over and write notes on them. The photos would be printed, or projected from a ceiling mounted project (e.g., like the Realtalk system at Dynamic land, though projects have a problem of resolution.) Notes and order would be detected by a camera, and everything would be used to organize the “real” digital collection.
- More human mannerisms in chatbot dialog
- Long pauses in attempts to encourage further elaboration like a real interviewer would
- Typing notification (…) appearing and disappearing
- More robotic mannerisms in chatbot dialog
- Play up the fact that it’s a robot through font and visual design choices. This may make people more comfortable with errors and mistakes that the robot might make. Make the robot socially awkward, because it will be anyway.
As I explore and develop Vignette, my photo narrative project, I have also been reading about theories of narrative structure. Narratives are vital to human understanding. Narratives describe a connection between a series of events by convincing a reader of a progression or cause. Without the ability to construct a narrative, there would be no stories, only unrelated events.  Research in cognition shows that narrative is strongly linked to memory, and that understanding of many complex relationships takes the form of narrative.  Modern narratology is heavily influenced by cognitive models and “considers narrative to be the conscious and unconscious mental framework that supports and directs human thinking.”  Furthermore, narrative creation is understood to be a critical part of personality building and self-definition.  Constructing coherent life-narratives is strongly associated with greater life satisfaction and reduced risk of depression,  and assistance with narrative is now a common psychotherapeutic approach.
In Vignette, I am exploring methods computers can use to understand narrative relationships between personal images. I wrote a good deal about how I’m approaching the project previously, but I didn’t talk much about the “kit of parts” I am building it with. Today, I’ll rundown some of the tools that I’m using to develop exploratory prototypes.
Vignette is an open source project investigating new ways of encouraging narrative creation from personal media. Vignette will be exploring data-driven methods for managing large quantities of photos and videos, visualizations that highlight the importance of place in narrative, and interfaces that engage users conversationally and emotionally.
In 2014, carrying maps marked with waypoints and a backpack laden with supplies, I embarked upon a long distance solo hike through rural Turkey. It did not go as planned, and I spent days grappling with questions of loneliness and purpose arisen from just under the surface of the everyday routine. After chancing upon new faces and walking away from my isolation, I was scrolling through photos on my camera’s LCD and jotting down observations and realizations, when I began to feel that there might be a structure to the experience that I had not seen while enveloped within it. The ensuing process of extracting a narrative from my images, frustrations and feelings, shaping something coherent from it, and sharing this story proved immensely cathartic. This catharsis was not short lived. It helped me understand that the photographs I’ve taken are more than a jumble of prints in a shoebox — they are a story captured in slices, and the composition of these fragments allows me to share that story with others.