Balance between detailed metadata and benefits of Cloud Optimized GeoTIFFs

[drnextgis]

I'm looking for a way to find a balance between employing geographically compact STAC items for highly detailed metadata descriptions, while still retaining the benefits of Cloud Optimized GeoTIFFs. Consider a scenario where we have a large image and wish to provide extremely fine-grained metadata (such as cloud coverage). To achieve this, we divide the image into smaller sections (tiles) and utilize them as the basis for STAC items. On one hand, this offers a significant advantage: granularity that enables precise searching. However, on the other hand, if a client is primarily interested in visualization and not metadata, desiring to view all tiles associated with the same image (each STAC item possessing a property with the original image ID), it would entail a substantial number of HTTP requests to retrieve data from assets across corresponding STAC items. This can severely impact user experience.

Is this a common challenge that others have encountered in the past? Are there established best practices? I've contemplated the idea of having these geographically compact STAC image assets reference an original image and storing "ranges" (or perhaps just a single range with the original resolution) in the "properties" section pointing at the file location corresponding to the item extent. This approach would allow us to reduce the number of rasters we need to manage while still preserving highly detailed metadata.