Quote:
Originally Posted by jdg-calibre
Hi Covid,
I did look at the EPUB output section of the Conversion workflow, but I do not see something that is obvious to make the image changes that used to happen without me taking any actions other than to convert the docx to epub.
Did you make changes to Calibre between 6.14.1 and 7.x that affected the way images are processed in a docx file during the conversion to epub?
Many thanks again for you insight and intimate knowledge of the tool!
Jonathan
|
I wanted to provide an update. I have solved the problem. I compared the images that were generated using version 6.14.1 and then used the largest image dimensions as the "Shrink images larger than:" factor when using 7.12.0 for each of the platforms that I care about. I was able to get more or less the exact same sizes though there were some minor variations in the generated images despite the guidance. Here are my notes from that test:
======================
Starting with 6.14.1:
Apple: 749x842 for Map of England, 749x694 for Scandinavia and Gotland and the trip of 749x842. Runes are 113x113 as they should be...
B&N: 576x648 and 576x534 and runes are 113x113
Kobo: 512x576 and 512x474 and runes are 113x113
Now, upgrade to 7.12.0 and try to generate with the correct shrinkage for each platform:
Apple: using sizes from 6141: 749x842, get 749x842 and 749x693 (odd) and runes and dingbats are correct.
B&N: using sizes from 6141: 576x648, get 576x646 (odd) and 576x533 (odd) and runes and dingbats are correct...so there is some minor difference on Nook
Kobo: using sizes from 6141: 512x576, get as expected, 512x576 for the Map of England, 512x474 for others and runes and dingbat are correct at 113x113 and 150x28.
======================
Notice the minor differences in the height of the reduced images on Apple and Kobo for some of the images. Again these are minor size issues and must be related to the algorithms used in Calibre I believe since the source images were the same exact dimensions in the MS Word source documents.
So I think I am all set to use 7.12.0 and later with the shrink factors I was able to determine based on 6.14.1 output.
If anyone has any further suggestions or if there is in fact something that Covid might want to review and tweak that would be much appreciated.
Thank you to everyone who already contributed to this thread.