Opinion on Best Format for Text+Wordclouds

[trefoil]

A personal journal project that makes use of wordclouds to index each entry has a concern about file size. The KFX is awfully large at 250 MB, and so is there a way to reduce the size? What bloats it most of all is the wordclouds being images in jpeg. Here are some things learned:

Using PDF for clouds and text takes the least amount of space because there is no need for image files anymore. The problem there is that the journal loses the enhanced typesetting, and more importantly, the Screen Reader doesn't work with PDF.

Using SVG for the clouds in a KFX would have been fine, except that the Kindle vector graphics engine doesn't allow text directives for some reason. Converting the cloud to vector text outline polygons would defeat the purpose.

Is there some format that would permit both the use of the screen reader and a vector graphics format that permits text? Or else, is there another way to draw a wordcloud without SVG?

[jhowell]

Originally Posted by trefoil

A personal journal project that makes use of wordclouds to index each entry has a concern about file size.

Is this something that you are planning to publish though Amazon?

Amazon produces a different KFX image package for each book based on the capabilities of the reading device. E-ink Kindles will receive low resolution, black and white images. Other devices get better quality images, but in all cases the images are rasterized for KFX format.

Originally Posted by trefoil

The KFX is awfully large at 250 MB, and so is there a way to reduce the size?

Use fewer image, lower resolution images, or eliminate color.

Originally Posted by trefoil

Using PDF for clouds and text takes the least amount of space because there is no need for image files anymore. The problem there is that the journal loses the enhanced typesetting, and more importantly, the Screen Reader doesn't work with PDF.

PDF is not a good format for e-books. It does not allow reflow, making the book unreadable on devices without a large screen.

Originally Posted by trefoil

Using SVG for the clouds in a KFX would have been fine, except that the Kindle vector graphics engine doesn't allow text directives for some reason. Converting the cloud to vector text outline polygons would defeat the purpose.

It wouldn't help. SVG images are converted to JPEG for KFX format.

Originally Posted by trefoil

Is there some format that would permit both the use of the screen reader and a vector graphics format that permits text? Or else, is there another way to draw a wordcloud without SVG?

If you use SVG images containing text the book will not be available in KFX format, but that might be a good thing.

It should work in Amazon's older KF8 (azw3) format. That format supports SVG without rasterization.

Whether or not KF8 will work with a screen reader would depend on the reading device.

[trefoil]

Originally Posted by jhowell

Is this something that you are planning to publish though Amazon?

It should work in Amazon's older KF8 (azw3) format. That format supports SVG without rasterization.

Whether or not KF8 will work with a screen reader would depend on the reading device.

Thanks for the info. I will try again with Calibre's AZW3 converter (starting with HTML + inline SVG), just to see how that goes. By cutting some corners on the wordclouds, the AZW3 file ends up at just about 8 MB. That makes it about thirty times smaller than the KFX file! The VoiceView works pretty well with the smaller file, but I've noticed in the past that it crashes more often with this format than with others.

Now I can listen to my thoughts while working in a crowded room, and the cloud gives me a feel for what's coming.