Some images do not load via the API and return a vague 500 error. For example, image ID 168287615226281 
You get a 500 response from server with an error like this one:
{
“error”: {
“message”: “Service temporarily unavailable”,
“type”: “MLYApiException”,
“is_transient”: true,
“code”: 2,
“fbtrace_id”: “AXRKafn8F5d7bzTF8GrcgAy”
}
}
I’ve tried via my own access tokens and the API demo website. The bad images do produce a hover preview in the API demo website, but they do not load.
It seems any image in that sequence will produce that error.
Other images still work with the same token or via the API demo website, for example this nearby image from another sequence: 491296285325988 
This is not isolated to that bad image or images in that sequence. I’ve encountered other images that have the same issue in other places. I wish I could investigate the properties of these bad images futher to try to debug the problem a little more, but the API will not return any info about them – just errors.
(For reference, the bad image and working image can both be seen in the center of this view: Mapillary API Demo )
This is frustrating because a bbox or radius query of the API will still return these bad/non-working image IDs and the MapillaryJS viewer in the our client app will try to show them only get to a “Failed to cache image” error.
Is there a reason why some images have this error? Is this a known issue? Is there a way to exclude bad images like these from query results?
