I’m an Obsidian newbie. I’ve been using paper sketchbooks for a long time, an have now gotten about halfway through scanning them all, making png’s.
I want to start using Obsidian, and to start linking, or referring to, those sketchbook scans in my Obsidian notes. I’m using Obsidian by storing my vault in a git repo, on Github.
But I don’t necessarily want to upload all those scans to github. Well that would be my first question, whether there are good established patterns to integrate image files that are held elsewhere. I could just have them locally and use file:// references, but that could be problematic if I switch operating systems and the file references would look different.
Or I could have a local webserver, that way they’d be http references and wouldn’t change…
Basically for now I want to avoid hosting them in a cloud service somewhere.
The other part of the question is whether there are well established naming patterns for such a thing.
So far, I’ve just put them, in order, with scanning timestamps, in folders, where one folder is one notebook, names for instance 2017_1 , 2017_2, and so on.
Instead of the timestamps as names, I could auto rename them to some pattern, while preserving their ordering.
So if there is an established pattern in Obsidian I would probably follow that.
Also I should point out I just want to reference / link to, the individual pages. They are drawings interspersed with some handwritten text, and I don’t really need to OCR them or convert them to a PDF representation.
Your ideas sound reasonable to me.
I would probably try and place the images inside my vault and exclude them from the git export. This way, you can make relative links to the images that will always be consistent with the rest of your vault, without having to worry about OS changes and the like. Of course, you would need two backups strategies, one for git, and one to another location that would have your images backup.
As far as naming the images go I think your suggestion is reasonable. You can use obsidian to make an index that link topics to the relevant pages. I saw someone do something similar on another thread recently, i’ll try and dig it up!
Found it!
Thanks for the tip! Excluding those directories from git export, and arranging for their backup won’t be a problem.
I won’t be able to use this ZK pattern
20230102.1350
since the original creation timestamps of my sketchbook pages are unknown and the timestamps I have are from the scanning, i.e. meaningless.
So if I create a folder inside my vault for the images, would there be a best-practice regarding the nesting, or absence thereof? Or is it all up for grabs basically?
I could have one folder with all of them, and make the sketchbook name part of the filename, such as
/scans/2017_1_005.png .
Or I could nest the folders. such as
/scans/2017_1/005.png
I’m guessing either of those would work, am just wondering if there is a canonical way…
How are you going to search for them when you are trying to find an old sketch? Are you going to know what date/range of dates it was saved? If I were saving sketches, I might start with the ISO date and a counter so that they sort in the same order they were in my sketchbook, but I would also give them a descriptive and keyword stuffed name. So something like
2023-09-10-001 elephant holding a travelling trunk comic pun humour
2023-09-10-002 mccray house before renovations
2023-09-10-003 flow chart showing Mark LLC corporate structure organization hierarchy
2023-09-10-004 dad John Smith holding grandson Spencer Smith age 2 realistic warm family
Then you can search for the subject, style, a keyword, etc.
That’s a good question. For the existing sketchbooks, the best time range info I was able to achieve, was, to know the order of sketchbooks in each year. And for the pages I have the order. But no way of knowing a date for a page, usually. For new sketchbooks, I could adopt writing down dates. Each sketchbook doesn’t have a lot of pages, usually 15-30. So it’s not overly stressful to page through an entire sketchbook looking for something.
Regarding the pattern:
2023-09-10-001 elephant holding a travelling trunk comic pun humour
Are you referring to this as the filename, or some other way of adding the description?
I was thinking of adding tags to the images, i.e. to keep the filenames simple and to add tags somehow…
I suppose this might mean creating one Obsidian note for each image, but that could be easily done, I could write a script for it.
I think I would rather not make the description part of the filename, since I view this as part of ‘classification’ and classification structures can change over time.
I use tags in my regular bookmarking (I use pinboard) and I sometimes re-tag whole swathes of bookmarks, when my idea of an ideal classification changes. Also, with a sketch being relevant to multiple concepts, I can add as many tags as needed.
It was my suggestion that would be the file name. If you want to be able to search from your file finder/explorer, then it is really helpful to have that information in the filename so you can see in an instant whether it is what you are looking for or not. And tags are not ubiquitous across platforms and systems. I’ve moved from one system to another over the years and always have in mind what information/features will move with the file no matter what system it is in, and what is in a closed system (like Obsidian.)
It also makes them a lot easier to find with quick switcher within Obsidian.
Thanks for the clarification.
I hope Obsidian will stick around, or there will be a successor system at least, because I’m intending to make heave use of the in-vault linking. Since I decided I need to have tags, I can then add a description, maybe as an ‘alias’ into each note associated with each scan.
So what I’ve done now, for a scan called 2009_1_01.jpg , I have a page called 2009_1_01.md which looks like this
---
tags:
- sample
---
![[2009_1_01.jpg]]
and now I just have to make a script to create all the other md files in the same way.
IIUC, this kind of in-vault link will not work based on general markdown, but relies on the indexing within the Obsidian vault.
But I’ll be using that a lot anyway between all my notes, so I suppose none of my notes would work if Obsidian itself went away.
My naming pattern for my notebooks is AAYY-123, where:
AA= any two distinct characters, typically some descriptor of the notebook (eg, BL for black, LT for Luechtterm, SP for spiral)YY= last two numbers of the year in which the notebook was started (eg,BL23was started in 2023)123= three digits for ascending page numbers beginning with001(use leading zeros)
So, for example:
- BL23-001
- BL23-002
- etc
I number each physical page sequentially, so the file name tells me exactly which notebook and which page number. If it ends in .md, it is the transcription. If it ends in .png, it is an image of the actual page.
Because these notebooks are daily journals, each page has a date associated with it, so I put the date at the beginning of each filename, without dashes, so 20230922 BL23-001. I sometimes will add keywords or descriptors to help find something easier, though I find this is often unnecessary because those words are likely to be in the text of the note.
Because your pages are mostly non-text (drawings), you would probably want to add a searchable description…
BL23-015 elephant holding a travelling trunk comic pun humour
But you probably would not want to add actual page numbers on the drawing, though perhaps on the reverse?
I thought about whether, apart from writing dates, I might number my notebook pages in the future as you suggest. But I think I won’t, I realized something when I scanned my old notebooks:
I basically curated/redacted the pages at scanning time, on average just scanning two thirds of the notebooks or so.
There were just many pages which with the benefit of hindsight turned out not worth scanning, and that’s not just to save work while scanning, the scanned notebooks are more useful to page through, with crappy pages already removed.
With the method of scanning in the right order, and having the app create filenames with timestamps (sortable), and then later running a script to rename them, I ended up with contiguous naming 01,02,03,04 etc. whereas if I had already chosen numbers when drawing the note, it would have been a really annoying job of manually renaming all the files, having to make sure the numbers match what’s on the page.