I've been toying around with various formats and how quickly the DR displays them. I'm mainly concerned with image-only documents, such as comics or image scans of entire books, or documents with large components of non-text items (vector graphics and so on). For text-based documents, we already have a good option as I understand it: FBreader.
I am open to creating a separate "optimized" library, preferably created automatically from my original library of ebooks, for displaying on the DR. I wrote a simple script that spiders the entire folder structure of my library and "flattens" all files it finds (pdf, doc, djvu, cbr and so on) using open source tools into a bunch of DR-optimized jpegs (grayscale, orientation corrected, contrast enhanced, 1280x1024, 160dpi etc). The script also automatically combines the jpegs into a pdf.
Everything is "flattened" into a raster image, but since it is optimized exclusively for the DR this should not be a problem (except in terms of disk space), unless you wanted to zoom in really close. In short: this should be a great way to view stuff on the DR.
Here is what is probably well known:
1. A bunch of images in a folder is opened rapidly by the DR, and the other images in the folder are treated as "pages" in the image viewer. Switching pages is very rapid, even if the page is not the next one in sequential order. Very nice indeed!
2. PDFs are SLOW! A large pdf file (with optimized jpegs embedded onto each page) is still very slow to display when jumping pages in non-sequential order.
The large speed difference between (1) and (2) makes me really want to use (1), but this is really disorganized and does not have some of the nicer features the pdf viewer has, such as remembering which page you were on. I don't remember, can someone tell me if the "scribbles" features are available in the image viewer as well?
Unfortunately, (2) may never get as fast as (1), mainly because of the pdf format itself and the fact that the viewer is not necessarily optimized for display on the DR.
Is there any information on efforts to build a DR-optimized image viewer? An 'evince' port itself may end up being a bit heavy and not necessarily DR-friendly (supporting scribbles, minimizing screen refreshes, optimizing processor usage to conserve battery etc.), but would be nice. Is anyone working on this? Perhaps a CBR viewer?
I think we don't really need support for a zillion formats, as long as there is an automatic tool that spiders your collection and spits out opimized files. It might hurt one's sense of efficiency (Rasterize a djvu!!? The horror!), but the original file is left untouched for archival anyway.
A further question: does anyone know of ANY document format that allows for displaying rasterized graphics with embedded text? PDFs do this, but there is no way to tell a PDF that "page 1 is a 1280x1024 image, so for heaven's sake stop playing with it". CBR files (really, a rar-ed bunch of images) are great for this as each page is just an image, but do not contain features for ocr-ed text in the background for instance (so its pretty much only good for comic books, as the name indicates). Anyone know of a format optimized for displaying a bunch of images with text overlaid?
This would be the PERFECT format for displaying scanned documents (technical books, textbooks, comic books) on a specific eReader. Coupled with a tool that converts any document/CBR into this format, we have the perfect optimizer for the DR (or any ebook reader, technically).
If there is a standard format, maybe I'll start looking to see about porting a viewer for it. If not, maybe I'll think about making up one (and a viewer for it)
If anyone is interested in the conversion script I described at the start of my post, let me know, although there are probably tools out there that do things like this already.