Image thumbnail generation


I’m reaching out here to see if I can get some advise as to how the preview thumbnail generated when uploading bitmaps are created.

Are there any specific rules around how the 2:1 ratio letterboxed version of the file is created. There does appear to be a level of internal ‘processing’ of the image as it isn’t consistently picking the pixels directly in the vertical centre and it feels like sething is trying make an intelligent guess as to the best thumbnail area to select. Its also not random as generally when uploading the same bitmap bitmap twice the same crop area is selected.

However uploading multiple version of similar bitmaps with exactly the same source dimensions ( say 800*800 pixels ) but variations on the content inside vastly different letter boxing is selected.

Attached an example.of what I mean.

So. Does anyone have any insight as to what’s good on under the hood with images once uploaded?

Thanks in advance



I am not sure, but it might be possible that their algorithm tries to do face detection and crops accordingly, so that the face(s) will not be cut-off. But that’s just a guess based on a little bit research :wink: