Open nchapman opened 8 years ago
That used to be the approach, but it worked really poorly in some circumstances:
The H&M example feels a lot less broken than this, and I think logotypes are probably a more common use case than photograph-type images that can be cropped reasonably well.
Thoughts?
If we look at this from the perspective of learning styles, we can aid visual recall by displaying images that are intimately connected with the content a user's trying to find or recover. (I'm more of a textual and auditory learner, so, to me, this is more of a theoretical improvement.)
Showing big logotypes or big favicons seems like a less powerful memory aid, so I think we're better off optimizing for the case Nick has in mind.
I don't mean this comment as a verdict or edict, just trying to push the conversation a bit deeper into the human factors. @chuckharmston, what do you think?
Note that we could also file a bug against tiny machine to exclude images which seem to be logos or icons. One cheap approach might be counting the number of colors in the image.
We know how complex the images are. We could background cover for complex images and not background cover for simple images?
It would be nice if images were cropped to fit the preview well. We could treat the image as a background and use
background-size: cover
to get it to fill the area.This is a good example of how it could go bad in that we'd cut off the text, but at this size it seems well worth it.