For my doctorate thesis, I am focussing on implicit knowledge constructions in images. For that I am trying to find a practical way to analyse images in tinderbox and compare the results among the cases (images). I was inspired by a rather old but very eye-opening video on qualitative text analysis https://vimeo.com/8772338 by Tom Webster.
The method I am using includes specific steps that I am trying to lay out on a map view in separate adornments: Currently, I use image adornments because images in notes tend to slow down map view in particular and the tinderbox file in general, dramatically. I also tried to implement my procedure via containers and smaller maps instead of one large map but I was missing the overview functionality of having everything on one map. I draw lines and visual annotations in photoshop and drag the annotated images as screenshots into tinderbox (where they become image adornments due to performace issues). Around these adornments I create notes regarding my findings and assign a Prototype related to the methodological step. This workflow is yet to be tested (which I currently do in preparation for a conference next week).
Is anyone here experienced in image analysis with tinderbox, would like to share their experiences and could walk me roughly through their workflow? When I gained more experience with the processes and developed my own practices, I will also share my workflow, of course
You didnât mention how many images, their resolution, and pixel dimensions that you plan to include in Tinderbox. Those factors, after a point, can affect Tinderbox performance. Since the major element of your process appears to be creating textual notes associated with elements of the image, as well as lines and other marks on the image, is it feasible to use a low-res image on the map for purposes of anchoring your notes, and then point to the original image via a $File or $URL external link?
Tinderbox is probably the perfect analytical tool for your process, IMO. Just a matter of tuning the process for best results.
I agree with the last above. As assumptions/understandings are not as universal as we all may like to suppose, it might help if you posted a small TBX showing the sort of notes/adornments you are using along with image files of the file size/format you will the (the actual picture style in the file is immaterial).
Tom Websterâs method canât be implemented in Tinderbox v6 onwards due to changes in the design of the app (reflecting changes in the underlying OS). However, subsequently in 2022 I did make a TBX that achieved the same style of process but in the new app see this demo TBX featured in this thread: Tinderbox Meet-up 29 May - Map Sorting/Data Triage: the 'Webster' method.
What isnât clear is why an image-containing note needs to be used in this process as a proxy for text.
I recently finished a project that involved analysing and making sense of hundreds of stills from various films. TBX is good with images, very good even, and I spent months and months figuring out how to work with them in my files. But TBX doesnât deal with images the way other programs I knew did, and these differences created some practical limits that kept rearing their heads in odd moments as I worked. When I asked my own questions about how to deal with them, @mwra pointed out that TBX was at itâs best dealing with text and @PaulWalters suggested keeping images elsewhere while continuing to do my work in TBX. For quite awhile and for all kinds of reasons, I resisted these suggestions. Ultimately, though I shifted gears and was glad I did.
In my case, I decided to keep my images in Curio because 1) it allowed me to size and arrange thumbnails on a canvas the way I wanted to do on TBX maps; but 2) it also stored those those images in a package and provided a link I could use from TBX to access the image directly. When I went to work on a still in TBX I just clicked the URL for the file which I kept in a displayed attribute and it opened. (Thereâs also a way have it open in an Apple Preview window ready to be referenced or moved around my screen.)
Your work may require something very different, but I found this division of my image storage and mapping from my analysis and writing much less jarring than I anticipated. It also took a technical question about how images are storedâin a package or as a binaryâoff my plate, which freed me up to spend my time dealing with my stills.
One feature of Curio that is useful is the use of watched folders in the âLocal Libraryâ inside Curio. For example Curio can index files in ~/Documents/Research Images in a sidebar panel, and then those images can be dragged onto the canvas as aliased files. The original resides in its home folder.
From that home folder, Tinderbox can use the âwatched folder in Finderâ feature to access the image if needed. Iâm cautious about doing this a lot. For example, I have an image resource folder with 25 MB of jpg files in it. If I watch that folder in Tinderbox, the Tinderbox file blooms out to an astonishing 210 MB in size. To get around that and facilitate data sharing between Curio and Tinderbox, I have a KM macro that creates a sidecar file text file (with the .md extenstion) next to an selected image in my resource folder with the same file name as the image.
The images and that sidecar file are next to one another in my Curio local library and both can be dragged onto a canvas. In Curio I can edit the contents of that file if I want to make notes about the image. And I can watch that text file in Tinderbox so that my notes are available there, too, in read-only mode.
These are just examples of gaming the system to share data access and minimize file bloat.
As a similar âimage + textâ strategy but with different nuts-and-bolts: I also created a âbaseâ note for each image but with two differences.
Rather than using a watched .md file in the Finder to make basic notes on a still, I created that note directly in TBX. When I wanted access to it while working in Curio, rather than working with an alias of that note inside Curio, I opened it from TBX in a text window with opt+cmd+x. I love these windows: theyâre space efficient and easy to work with off to the side of other apps.
I kept my images in a Curio package rather than in an external folder because early on my project files in the Finder were too much of a mess for me to be confident I wouldnât move an image folder around. I was also new to Curio and didnât know if the app would be able to keep track of them if they did. So rather than watching an image folder to create notes with image previews, I added a link to the image file in Curio to my âbaseâ note. The Curio link accessible from the contextual menu would bring me to the still on a Curio canvas. Alternatively, âShow in Finderâ in that same menu would open a Finder window holding the image file, allowing me to drag it to a File attribute in that base note in TBX. Clicking File would then open up in a Preview window.
In practice, text windows plus the two Curio link options gave me the flexibility, depending on the context, to work either with images on a canvas in Curio with a small text window into TBX or, alternatively, in TBX with a small preview window showing one (or more) of my Curio images in isolation.
So again, just a different way of aiming for âthe best of both worlds.â
@chrisH, can chance you can join tomorrowâs meetup? I donât have time to develop a mock-up for you, but I have several ideas that we can go through live during tomorrowâs meetupâwe can see if they work fo you.
My strategy would include the following:
Images stored locally on harddrive, pulled through templates
pMedia prototype
Use $Hover
Use $Fill
function to automatically link images to attribute values (useful for hyperbolic view)
Several user attributes to capture the QDA tagging youâre looking to do
Use $Smartadornments to market the QDA process faster (optional)
What I have in mind would like you to tag/categorize images through various vectors, tags, and links, as well as write up your report and publish it to Word, an Excel table, and even a PowerPoint, or hypertext website.
Trank you (all) for your helpful replies. Currently, I am very busy with preparing the presentation for the conference on Wednesday. And also currently due to our kids (ages 1 and 3) I have difficulties attending to the meet-ups even though I was planning since beginning of the year. I will answer more thoroughly after the conference next week
Good luck at the Conference (I just got back from one!). If youâve any questions on my TB, i.e. the âWester methodâ updated to work within the post-v6 Tinderbox app, then do ask.
Thank you again for the very sound and detailed replies. After all of your replies I thought a lot about my own workflow and setup, growing increasingly more insecure, hence I was hesitant answer to your comments.
Particularly the combination of TBX and Curio intrigued me, as Curio works great using graphics and lines while TBX â though being a visual tool as well â excels (IMHO) in data analysis and working with meta data. My Zettelkasten lives in DEVONthink Pro, my Dissertation âcollection notebookâ is currently growing in Curio while the analysis of the qualitative date should be processed in TBX. I am still not sure wether itâll stay that way but one thing that is a plus for my workflow in Curio is that it is nearly lag-free when dealing with lots of images. One thing that is a huge plus for me in TBX is that I donât have to think about which type of note (âfigureâ in Curio) I am developing/writing beforehand.
As several users wrote in comparison of the two programs is that there are some overlaps but they both serve very different purposes. I am hoping to be able to harness both for my purposes.
That said, I would like to first present my initial workflow based on the methodological necessities and then the current setup.
The method and methodology I use is quite experimental and (among other steps) analyses the formal structure in images. That means that I first draw lines in/over the image to reconstruct certain dominant lines and focus points inside the image. In TBX this functionality is not (yet?) present while it is in Curio.
First, I set up a group with documents (in Markdown format) in DEVONthink corresponding to the analytical steps (iconographic description of the image â what does the image show?; iconic analysis â how are the shown aspects presented in the image?) and add them as file-backed sync text figures in Curio. This way, all written analysis is synced to DTP. Next, I will create canvases (âidea spacesâ in Curioâs language) for each analytical step and start drawing the lines. Curio has a way to work with layers. As there will be a lot of lines that can get confusing I send groups of lines to separate layers that I can turn on/off as I need it to.
The method is based on iterative comparison of cases, hence I will have several groups/folders based on the specific images of a case. In a next step â yet to try out â I want to replicate all MD files to a watched group in DTP and this way add them to TBX. I am not sure yet, how I will conduct this analysis but I will figure something out and report it back to the community
Yes and No. No, because Tinderbox doesnât use feature parity with Curio so you shouldnât assume Curio features are in Tinderbox (I donât mean that unkindly, Iâm just trying to be unambiguous). Yes, becauseâif I understand the above correctlyâyou can draw linked notes over a background image (of a network/graph layout).
To do the latter, youâd export or make a screen grab of the diagram saving it as a JPG or PNG. You then drop or paste that onto your Tinderbox map view. Doing so will generate an image adornment. Now you can add notes to the map and place them over the items is the âbackgroundâ image. Note donât double-click on the image adornment as Tinderbox will (understandablyâ ) think you are trying to edit the adornment. Rather, make the new notes alongside the adornment and then move them so they are over the right place. Once the notes are doe, you can add links with (link type) labels if needed.
HTH
â . Do bear in mind that a Tinderbox map is not a simple drawing app, so donât assume all tasks will use the same technique as in [some other app].
I should have added, if you have specimen image of the sort of diagram you are trying to âmapâ it might help in terms of giving a more precise answer as to what you could do inside Tinderbox.
Also, Markdown. Tinderbox does support Markdown but unlike many very recent apps it isnât built around Markdown. IOW, itâs textual use doesnât assume MD but does tolerate/handle it. So, if Tinderbox is intended as the final analysis space, Iâd also check if use of Markdown is needed or simply what you are used to. Again, this is about surfacing untested assumptions. Markdown or not isnât a zero sum good/bad decision. But if Markdown isnât needed (it might be needed, because reasonsâŚ) then you might do better to work with normal text and do all the styling on export from Tinderbox.
Thanks for the replies! First of all, of most of the things you wrote I am well aware and have them considered, especially re: the uniqueness of TBX. Thatâs why it took me so long to figure out a workflow. Also, I donât expect TBX to act as other software; I use TBX in so many occasions, but most of them are text-related (even though one could argue that map view is just partly text). Probably due to my English-as-second-language-situation I couldnât express my point clear enough. I tried your approach already by myself when I first thought about a workflow using TBX several months ago and it left me with too many notes on the canvas that I didnât need at the stage of the process.
Yet, I do want to use TBX in analysis as I think it works perfectly with text. Thatâs why I am currently building my workflow (status: work in progress) around Curio, DTP (as data center/data hub) and TBX as comparison tool of the data.
The research method I am using (documentary image analysis = Dokumentarische Bildinterpretation) consists of three necessary stages (iconographic image description, iconic analysis, iconologic-iconic interpretation) in which the second step (iconic analysis) needs extensive drawings, hence the move to Curio canvas.
In fact, it is not necessary to use MD format, itâs just what I use for my standard documents (notes and drafts due to the simple formatting). Also, it is highly recommended by my university to use md/plain text.
I will, thank you! I just wanted to give sth. back to the community even after this time (8 months) and after the community helped me find my ways (see the posts above)
Thank you for your community spirit in doing so. Even if peopleâs exact path differs, it is very helpful to know at someone else has achieved all/most of what we might be trying to do. So, the progress report and giving back to others is much appreciated.
