How to preview first EPUB embedded image (not cover page!)

[Mdemaillard]

I have following "library cleaning operation" ongoing : bad cover / good cover not correctly embedded.

Some of my low-quality epub books have cover images, which are incorrectly embedded on the book first page : aka not on a separated cover html file.
- some of these images are correct cover-images, that i'd like to keep as correctly formated covers. To do this i use Sigil, split the image on a dedicated html file, and mark both file and image as being covers.
- Some are images which i do not want to keep. Also with Sigil i just suppress them.
- some books have a cover, and also an embedded image on the first page, resulting in "double cover" (identical or not) when viewing.

--> Is it possible to bulk browse the first page of multiples ebooks, so that i only need to open the relevant ones in Sigil ?

What didn't work so far :
- windows explorer doesn't seem to be able to generate previews on epub files
- in calibre i browse easily the covers, but not the first book page. So even if i delete the covers (eather in calibre or editing the ebook itself) i doesn't help.
- Sigil or other viewers i tried only opens 1 file at a time, and don't do previews
- I don't want to blindly use the calibre function "remove first image" as some books have meaningful "non-cover" images, like map, artwork for chapter and so on.


Best method i could think of yet : set calibre to open ebooks with sigil, open ebook from calibre main list, check in sigil, work if needed, close sigil and repeat with next book.

While the book is open in Sigil i also use the opportunity to eventually split some title page, introduction, license page or the like from the book itself, so i guess whatever the preview method i'll keep using sigil to make the actual changes.

(Some books have images correctly defined as covers, so i manage them easily with calibre : i preview them using the cover Browser, and accordingly keep them, replace using fetch metadata or even suppress and replace with a generated jacket.)

[Sabardeyn]

I know you can do a search with cover:true and determine which books have a separate cover file; cover:false for books without. You could use the last search to find the books that still need a separate cover.

After that I think you're stuck looking through the books.

[theducks]

Originally Posted by Sabardeyn

I know you can do a search with cover:true and determine which books have a separate cover file; cover:false for books without. You could use the last search to find the books that still need a separate cover.

After that I think you're stuck looking through the books.

cover:true only looks at Calibre (book details) to see if CALIBRE has a cover.

The Quality check Plugin can look INTO a book and test ('declared?' or first image?) covers (3 choices of check)

[Mdemaillard]

Thanks Sabardeyn, i also already looked at that.
It is even possible to further filter ebooks having a calibre-generated cover using the "Quality-Check" Addon -> Check Structure -> Check calibre SVG Cover
(and do a lot of other things to the covers)

But unfortunately this all works on the correctly identified "cover part" of the ebook.
Thanks anyway for the suggestion, i'll take any further idea gladly.


theducks : this looks promising, i'll check it right away
hum, didn't find what i'm looking for after exploring in details "check-quality". It does have an awful lot of nice features thought, so i learned stuff and will be using it better.

[kiwidude]

There isn't currently a check in Quality Check which will do exactly what you want. There are a bunch of complications with trying to implement it, because of course having an image on the first page doesn't mean that it necessarily is a cover image, it could just be a logo of the publisher or a half cover etc. so you would still need to visually inspect each one to confirm it is what you are after. You would also have the problem case where the ePub has not been split into separate HTML pages for page breaks. So if the logic said "check the first HTML page for an image" then an image partway down would be a false positive. If the logic says "check the first page for only an image" (like what the Modify ePub plugin does when removing broken image link pages) then you will miss the cases where the ePub has not yet been split but still doubles up as a cover image. No clean way beyond eyeballing to know which is which...

And of course you also have the situation where the cover is at the back of the book (when shown in Sigil anyway, due to the linear attribute in the spine of the manifest).

It is an interesting one, as I too prefer to eliminate the double covers as I come across them. Just like I like to eliminate the contents html pages at the back of the book, which if someone has done multiple calibre conversions on without understanding the settings and issues get horribly duplicated. It is something else I may add to Quality Check so if I make the effort to do one I may try to do the other as well, but both have a fuzziness factor to either return false positives or miss certain scenarios which I don't like...

[theducks]

You like to remove covers.

I like to include 'alternate' (other edition) covers. This is probably from wanting to have the cover on the copy in my Dead Tree Library copy.

You do bring up good points about the difficulty of locating:Placement, type (content) of image,

Maybe a Report: Count of Images( broken up by size groupings:small, probably Logos, dingbats; Medium: Maps, charts or half covers; Large: Covers, titlepages )

[Sabardeyn]

Theducks,
Ok, I might not say this properly. Hopefully you will understand what I meant rather than what I said. Anyway...

Wouldn't something like what you're referring to be device screen size dependent? I don't Save to Device/Disk, so I'm commenting without any real experience. But as I understand the conversion process, one factor that calibre takes into account is the screen size, adjusting output accordingly.

So size is relative. Large on a 4" screen is one thing, on a Kindle 3 its something completely different. Unless it could be checked via a percentage of screen size: get image_size. if image_size >=75% report possible_cover:=true.

or something roughly equivalent to that. Margins and automatic image resizing might interfere with that idea however.


PS: I don't exactly care for double covers, particularly: two copies of the same image; or, a generic (calibre?) cover and the real publisher's cover. I might not have a problem with multiple cover images from various editions of a book - provided its a smallish number of images total (<12?). Forget something like LotR/Hobbit with hundreds of editions worldwide.

[theducks]

50% in the CSS is talking to the render engine. It has nothing to do with Calibre resizing (that is something else).

margin-left: 20%, is:
20% of the 600px on my PEz
20% of the 1440px on my desktop