Compress an ePub's pictures - an alternative

[rumplestiltskin]

I occasionally run into ePubs that are larger in MB than I think they should be (or, at least, larger than I'd like for uploading to my Kindle Fire via Amazon's webpage or storage in my HD). I've tried a couple of the Calibre-based fixes but nothing worked particularly well.

I did find a website that does very good compression and doesn't seem to adversely affect the appearance of the ePub. It's https://tools.nooapps.com/en/compress-epub.html

I have no association with the site's owner. Just wanted to contribute something helpful. Please let me know if it works for you (or not).

Thanks,
Barry

[theducks]

Why go out of app (calibre) Polish can compress images Losslessly

[rumplestiltskin]

Example :
Original ePub - 54MB
Polish - 34MB
Link - 10MB

The pics in the smallest ePub are, of course, smaller in MB/KB (and resolution) but, as this wasn't an "art book" - the photos were informational - I could live with the smaller pics and having the book 20% of the original size was nice.

[kovidgoyal]

DO an epub to epub conversion and set the max image size to whatever you like in the epub output options for arbitrary size reductions.

[rumplestiltskin]

Originally Posted by kovidgoyal

DO an epub to epub conversion and set the max image size to whatever you like in the epub output options for arbitrary size reductions.

Yes; that worked. Picking a pixel size is, I guess, one way of accomplishing this. As the images in my book are "png", that may be the only "size reducer" available(?). I think I'd like to have an option of compression but maintain pixel size. Does that make sense?

Perhaps the pictures in the ePub must be "jpg" instead of "png"? I write that because a quick web-search only reveals online png compression utilities that make the png's pixel size smaller. If so, that would probably require that the conversion option (ePub>ePub) first convert the picture to jpeg and then apply a certain amount of compression before re-assembling the ePub's components.

Of course, I may be ignorant of something critical about how ePubs shouldn't use jpgs because certain e-readers don't like them. (I'm throwing stuff at the wall here, I admit.)

Maybe I should be posting in the "Plugins Suggestion" thread.

Thanks

[JSWolf]

The solution I use is to load the ePub into the editor and on most images I do a 70% lossless compression. Also, I delete any HTML I don't want as well as any embedded fonts I don't want if there are embedded fonts. For example, an ePub can be large because of the adverts with high-res images that I don't need so I get rid of the HTML and the graphics.

You cannot do this from that website (probably designed to steal eBooks) as you have to do this in the editor.

[JohnnyBook]

Originally Posted by kovidgoyal

DO an epub to epub conversion and set the max image size to whatever you like in the epub output options for arbitrary size reductions.

THIS!! I think you fixed a problem for me. (Did not know about this feature)

BUT a suggestion....

Sometimes the images are sideways, instead of 900x1200, they are 1200x900

Maybe instead of resizing so that instead of reducing dimensions to (for example) 600x800,

It reduces the Longest dimension to the second value.
In the example above, the regular image would be 600x800 But the sideways image would be 600x450.

Using the longest dimension would give 800x600 instead.

Is this a possibility? Or maybe Calibre already does this?

Thanks,

Hmm Maybe just using 800x800 would take care of this problem, since in most cases I cant think of times when the shorter side is the issue.

EDIT: Yes using the same value for height and width does do what I wanted.

[kovidgoyal]

calibre preserves aspect ratio when resizing images the size you set is used as maximum size the actual image size will be a rectangle that is no larger than that size in either dimension.

[JohnnyBook]

Originally Posted by kovidgoyal

calibre preserves aspect ratio when resizing images the size you set is used as maximum size the actual image size will be a rectangle that is no larger than that size in either dimension.

Ok maybe update the tool tip, so instead of saying height and width, it says long side and short side.

Right now it says: "For example, a value of 100x200 will cause all images to be resized so that their width is no more than 100 pixels and their height is no more than 200 pixels"

[kovidgoyal]

Originally Posted by JohnnyBook

Ok maybe update the tool tip, so instead of saying height and width, it says long side and short side.

Right now it says: "For example, a value of 100x200 will cause all images to be resized so that their width is no more than 100 pixels and their height is no more than 200 pixels"

And that's exactly correct. The image will fit inside a rectangle of no more than 100 width and 200 height, which is what I said in my previous post.

[JohnnyBook]

Ok, I was asking it could change so the longest side would use the higher value.

But as I edited into my original post (probably after your post which I had not seen)

using 200x200 does what I was looking for.

Thanks!

[rumplestiltskin]

Originally Posted by JSWolf

The solution I use is to load the ePub into the editor and...snip...I delete any HTML I don't want as well as any embedded fonts I don't want if there are embedded fonts. For example, an ePub can be large because of the adverts with high-res images that I don't need so I get rid of the HTML and the graphics....snip

I've been playing with an ePub that has an advert page and a picture intended for that page. I deleted the page (as it was a discrete ".htm" page) and the jpg intended for that page. No errors found and, of course, that page is no longer part of the ePub. Nice suggestion.

I'm a bit hesitant about removing fonts as there must be some reference to them in a .css entry. Is there a simple method to remove the references once I've deleted the fonts?

Thanks!

[JSWolf]

Originally Posted by rumplestiltskin

I've been playing with an ePub that has an advert page and a picture intended for that page. I deleted the page (as it was a discrete ".htm" page) and the jpg intended for that page. No errors found and, of course, that page is no longer part of the ePub. Nice suggestion.

I'm a bit hesitant about removing fonts as there must be some reference to them in a .css entry. Is there a simple method to remove the references once I've deleted the fonts?

Thanks!

You load the ePub into the editor and remove whatever fonts you do not want. Then you go to the Tools menu and remove unused CSS. Then you edit the CSS to remove the @font for the removed fonts and the font family lines in the CSS. You have to do it manually.

[rumplestiltskin]

Originally Posted by JSWolf

You load the ePub into the editor and remove whatever fonts you do not want. Then you go to the Tools menu and remove unused CSS. Then you edit the CSS to remove the @font for the removed fonts and the font family lines in the CSS. You have to do it manually.

Okay; that actually sounds straightforward. I'll follow your directions. Thank you.