One of my projects involves a lot of pages each with several (up to 20) images. I tried to use Scrivener for it, but I’m running into some trouble at just 45 images.
With embedded images:
Images seem to be embedded as bitmaps so RTF files get huge (100+ MB each). Navigating through them cause Scrivener to crash after a few clicks, saves take a long time – essentially unusable.
With linked images:
RTF files are small and Scrivener is generally responsive BUT Scrivener soon runs out of memory and gives up on loading images at all. From looking at Task Manager’s data, it seems that:
- When opening a page Scrivener loads all images as bitmaps, unscaled (e.g. for a 16-image document it will allocate 300-400MB of RAM)
- When switching to a different document Scrivener does not release the previously loaded bitmaps (i.e. every new document opened allocates additional RAM)
- When memory usage reaches ~1.5GB Scrivener just doesn’t load images anymore - showing big placeholder signs instead.
I’d like to know if Literature&Latte considers any of these bugs that they would fix. The images could be displayed without keeping all of them in memory as unscaled bitmaps, memory used for a closed document could be released and/or Scrivener could garbage collect image memory when running out of RAM. Obviously all of these would require work, and if this is not a use case Scrivener wants to support, I understand.