I have a separate Scrivener Project for my Characters and a few other odds ‘n’ ends common among multiple projects.
The more character sheets I’ve built–I have around 160 that a couple of books in a series feed from through hyperlinks–the exponentially the larger my project became. I include my visualisation of who my character might be in a little 68 x 90 points scaled picture. I produce the pictures (generally a celebrity) through sizing and cropping it in PowerPoint according to a frame with the dimensions 4.5 cm by 6 cm. The original size of the picture counts so I generally have to scale it to 68 x 90 when pasted into Scrivener. Anyway, it works, and the effect is as follows–the picture(s) are pasted into a single table cell:
The drawback is that I ended up with a 28 MB file that was taking longer and longer to backup to Dropbox. Reading recent posts, I looked at a suggestion on using the Image Linked to File option. The user manual never presented a clear enough example to tell me I was on the right track, so I went about experimenting.
I created a folder on my computer called Series Character Images and stuck it in a Research folder, also outside of Scrivener. I’m a firm believer in small Scrivener projects with content like pictures, PDFs, and whatnot saved outside of Scrivener (and accessible through links) to keep my projects small. Then I saved one Character’s picture from Scrivener to the Series Character Images folder outside of Scrivener and deleted the picture in Scrivener. Next, I used Insert > Image Linked to File and selected the image. Once in place in Scrivener, all I had to do was scale it to my choice size 68 x 90 points.
When done, I saved and exited. Once the backup completed, I compared the current backup to the prior one and noted a reduced ZIP file size. This confirmed I was on the right track. After a few hours of following the same process with all my Character Sheets, I exited and found I’d reduced my project size from 28 MB to 879 KB and reduced the backup speed to mere seconds compared to a few laborious minutes. Furthermore, there is no visual difference in my Character Sheets in the revised project.
I further checked what would happen if my project became corrupted. Would I need to rebuild the links if a reverted to a backup? Opening a backup, I found the answer is positively no.
In a synched between computers and/or operating systems workflow, this might not be the best option, but I work on a single machine, and it meets my needs quite comfortably.
If you have two computers using the same project, on the second computer the link to the file is broken EVEN IF the linked pictures/pdf/etc are saved on a cloud folder accessible to both computers. the Link is unique to each computer, however if you keep the research files in one folder you can highlight multiple files to import into the computer at once. Or you can edit broken links and reconnect on the new computer.
I work on one computer only. However, for syncing purposes you can “address” all the broken links at once with an app like Atom Editor with it’s Find and Replace in Buffer feature, for a duplicate setup purpose.
In my setup, my projects are on my computer and the ZIP backups are auto synced as a ZIP to Dropbox.
I don’t trust cloud services with “open” documents and their guarantee that “we’ll determine when your computer is in need of space and auto-move what we think is right to the cloud.”
Try asking an LLM like Copilot to explain the difference between a cloud, green tick with white background and white tick with green background. It will BS you, and when you question the integrity of the answer it becomes significantly evasive, make you out to be stupid with language like “let’s slow it down and take it one step at time” and any further queries on that it treats like you never asked for clarification in the first place. It’s not that it hallucinates, it’s programmed to be wilfully misleading to promote sub-standard junk they call cloud services. The big guys have been trying this for decades, with somewhat better success of late.
FWIW, Dropbox’s own documentation (for example) is pretty clear about what the various icons mean:
LLMs generally explain everything poorly, it’s not just a conspiracy to promote cloud services. They are not search engines and treating them as if they were is just asking for trouble,
I’ve seen a recent example where changing one word in the prompt gave an entirely different answer. Change a NO (Correct answer) to a YES.
Stunned by how many ‘intelligent’ people I know are using them in place of Google, including a CEO of a multinational who insisted my correct answer to him was incorrect based on an LLM.
Yeah, I kind of had enough the other day, and wrote a massive itemisation of just how inaccurate and unhelpful a bot response was. What I found fascinating is that when I initially pointed out to the person that pasted it that it had accuracy issues, they were unaware of any and challenged me to produce them. So I did.
Be careful. I’ve experimented by having it summarize topics I know well and things I’ve already read. The results vary from “grad student with good research skills” to “gibberish.”
Something that Google is actually encouraging, for no reason that I can figure out. For a company whose entire reason for being is “index the world’s information” to promote a tool that routinely gives inaccurate results seems insane to me.
