<figure><div data-component="image-block" class="sc-18fde0d6-0 EXUng"><div data-testid="hero-image" class="sc-814e9212-1 fcEyBx"><img sizes="(min-width: 1280px) 50vw, (min-width: 1008px) 66vw, 96vw" srcset="https://ichef.bbci.co.uk/news/240/cpsprodpb/538b/live/d57eadd0-514f-11ef-986a-a10b7a6886df.jpg.webp 240w,https://ichef.bbci.co.uk/news/320/cpsprodpb/538b/live/d57eadd0-514f-11ef-986a-a10b7a6886df.jpg.webp 320w,https://ichef.bbci.co.uk/news/480/cpsprodpb/538b/live/d57eadd0-514f-11ef-986a-a10b7a6886df.jpg.webp 480w,https://ichef.bbci.co.uk/news/640/cpsprodpb/538b/live/d57eadd0-514f-11ef-986a-a10b7a6886df.jpg.webp 640w,https://ichef.bbci.co.uk/news/800/cpsprodpb/538b/live/d57eadd0-514f-11ef-986a-a10b7a6886df.jpg.webp 800w,https://ichef.bbci.co.uk/news/1024/cpsprodpb/538b/live/d57eadd0-514f-11ef-986a-a10b7a6886df.jpg.webp 1024w,https://ichef.bbci.co.uk/news/1536/cpsprodpb/538b/live/d57eadd0-514f-11ef-986a-a10b7a6886df.jpg.webp 1536w" src="https://ichef.bbci.co.uk/news/480/cpsprodpb/538b/live/d57eadd0-514f-11ef-986a-a10b7a6886df.jpg.webp" loading="eager" alt="Reuters Fire damage is shown in the Wahikuli Terrace neighborhood in the fire ravaged town of Lahaina" class="sc-814e9212-0 hIXOPW"><span class="sc-814e9212-2 jesyMJ">Reuters</span></div></div></figure>
Nothing obvious stands out in the markup - looks quite similar.
I am no expert at readability.js source code, but from what I tried to debug it looks like during scoring the images get removed - in some articles, the image is above the threshold, in others not.
Thanks for all the example, they will be very useful in debugging this but I also don't see any immediate difference so will have to look see what is up with the scoring for these.
Some pages on bbc.com are missing the hero image.
bbc.com articles where Readability removes the hero image:
bbc.com articles where the hero image is preserved:
Example markup of removed image:
Example markup of preserved image:
Nothing obvious stands out in the markup - looks quite similar.
I am no expert at readability.js source code, but from what I tried to debug it looks like during scoring the images get removed - in some articles, the image is above the threshold, in others not.
Maybe related or maybe not to these: