When building up knowledge or writing some sort of content from my notes, I often want to re-use images I have. They might be images I picked up elsewhere or diagrams I’ve made myself with excalidraw or similar tools.
So I came to realize that images can be also an atomic unit of knowledge that can be composed with others such as we do with notes and I think most of the tools don’t have a proper way for managing them. For logseq they are just assets, and as of now for me they belong to an specific note and I re-use them block-referencing the picture anywhere else but in my opinion we need more tools to proper handling them.
As a quick brainstorming, some features I think will be needed for this could be:
Images need to be indexed too: searchable by name, maybe also query-able.
Images can be a sort of special block, accepting metadata properties. This is already possible I think but paired with the other feature this will be enhanced.
Images have backlinks too. You can open an image as it was a page and see where it is referenced. Again this is possible with block-refs but not as convenient.
Renaming images, updates all relevant notes
I think there is also an OCR plugin that could take something like this to another level.
I agree that OCR of images is a “necessity” as it is provided by most other products at this point. But, as you noted, figuring out a “proper way” to capture that data with blocks / block references / backlinks would take quite a bit of work. And I believe the current development team is working on stability, speed, reliability issues at this point - so once v1.0 is actually released, perhaps there will be an active roadmap to request this - and to get feedback on timing of implementation - if this is even feasible.
First let me say what a TERRIFIC contributor YOU are to the Logseq community.
And yes, I’ve seen your plug in. Feedback is that it works well. Thank you!
My “necessity” reference was pointed toward the idea of making “images as a first class unit of knowledge”. Ultimately, things like automatic background scanning of images and automatic inclusion of data found into the DB structure would be a great core feature.
That is, in fact, the hazard of a product like Logseq… There are SO many possibilities that getting it “right” in the core product is (I’m sure) quite a challenge.
Oh sure, I’m not meaning this is a “must be done today”. Just a feature I was thinking could enhance the product, with something none of the alternatives do either.
I’ve written this post not just for feature request, but also for some discussion about images as an atomic unit of knowledge like evergreen or permanent notes. Since I think I’m not alone having this images/diagrams as also a source of truth or valuable information I thought about discussing making them a first class citizen in Loqseq.
As you said, the main focus is to get the core functionality right and performant and build up from there. The team is doing a great job with that. Later we can think about features like this one but I think is valuable to have time to discus it an get the details right so the implementation is smooth
Such a great suggestion! As a drawing artist I would love Logseq to be able to manage image knowledge and notes as well, as mentioned in the original post.
In my case, I am learning to draw and I am storing a lot of screenshot images showing thickness variations in lineart, different colors used for shadows in drawings, and step progress recording images showing how my drawing looks like after drawing the shadow layer + showing also what the shadow layer itself looks like.
These are very image-based information which cannot be easily stored in words, and I would love to have a tool that manages these notes well. Not sure if any tools already exist for similar features, but would look forward to Logseq being able to do this if this could be a feature needed by many users.
I came here thinking about this very thing and looking to re-find a tool someone else mentioend (thanks to Logseq, it was already in an old journal page) - https://tropy.org/
I’m wondering if there’s room to simply integrate -with- tropy and let it do all of the heavy lifting but share data sources?
, which could play a big role if having images more integrated into the system.