I had been using 3.4.3 and upgraded to 5.1.5. All seemed OK, but the image sizes on the narrated web media page are different. All pictures are set to the width that is specified in the max size slider on the “Images Generation” dialog. The older version didn’t increase image size, which it now does.
All the docs I can find say it should be a max size. Smaller images should be unchanged. The apropos code in 5.1.5 is definitely different. Even setting max width and height to zero to get the default doesn’t work. It actually makes a 0X0 picture. Is this now the intended action? It’s kind of inconvenient.
After some testing, I can confirm what you are experiencing. This from the wiki for the Narrative Web
Max width of initial image: (800 default) This allows you to set the maximum width of the image (in pixels) shown on the media page. Set to 0 for no limit.
Max height of initial image: (600 default) This allows you to set the maximum height of the image (in pixels) shown on the media page. Set to 0 for no limit.
Setting either or both image sizes to zero (0) results in no image generation instead of the expected exact image size.
And instead of the Max size for the initial display, images that are smaller than the max setting are actually stretched to that size.
I never altered these setting from the default when creating a narrative web in any version of Gramps so cannot offer a comparison between versions. But clearly the expected results are not what is being generated. You may wish to file a Bug report.
I do want to resize if the pictures are larger than the maximum, but not if they are smaller. Blowing up small pictures just makes them fuzzy.
Also, there’s the problem of landscape vs. portrait format. Suppose I set the max x size ( in the dialog ) to say 800 and y to 400. A 800 x 400 picture will be fine, but a 400 x 800 picture will be expanded to 800 x 1600, which is too big to view on most monitors. That;s the way it works now.
It didn’t used to.
The problem is that is makes every picture have the same width, as specified as x max in the dialog. That isn’t always the best.
When you have an image containing several people, If you have regions relating to these people for this image, you can click on the regions corresponding to these people to go to their individual page.
For this to work, I need a constant width for these photos.
Some of his portraits (as opposed to landscape orientation) might have a very odd aspect ratio. So that the auto-height is giving more than screen height when the width is scaled to 800 pixels.
Is the CSS capable of being made smart enough to adapt to his image size edge cases? Where if the width is less than 800px (or maybe 400px, since a 200% isn’t blurry), they are not scaled. And if the aspect ratio is significantly taller then the screen aspect ratio, it scales on “fixed height” instead of a “fixed width”?
If I recall correctly, the regions are percentiles; not absolute pixels. (Which probably simplified using scaled cached thumbnails in the dynamically re-scaled Phototagging Gramplet.) This limitation occasionally make it difficult to finely control drawing the rectangular regions.
I haven’t tried using the Gramplet to draw rectangular tagging regions in a very narrow aspect ratio picture. It probably is very weird there too.
(Not suggesting a solution. Just acknowledging the problem is more thorny than the user may realize. The earlier version of the Narrated Web Site probably didn’t support tagged selection regions yet. So it could play “fast and loose” with image sizing.)