Thanks! Right now it relies entirely on getBoundingClientRect(), so if an element has no dimensions yet the block gets skipped.
For images/videos I added a load event listener that re-measures once the media loads and takes up space.
For elements that are truly dynamic with no initial size, the best workaround today is setting explicit dimensions via CSS (min-height, aspect-ratio, etc.) which is good practice for CLS anyway.
That said, something like data-shimmer-width / data-shimmer-height attributes to manually override sizing could work well for that use case.
For images/videos I added a load event listener that re-measures once the media loads and takes up space.
For elements that are truly dynamic with no initial size, the best workaround today is setting explicit dimensions via CSS (min-height, aspect-ratio, etc.) which is good practice for CLS anyway.
That said, something like data-shimmer-width / data-shimmer-height attributes to manually override sizing could work well for that use case.
Adding it to the backlog!