increase image filesize limits

[beijbom]

I was thinking about image filesize limits. I don't think there is any reason to have limits anymore with the new backend. I'll conduct some tests, and if they work out we can change the policy. A lot of folks are asking about hugs mosaics (eg 20.000 x 20.000) pixels.

[kriegman]

It's good to be able to support, but it opens an abuse opportunity. Probably we're well below the radar screen on this, but "unlimitted" and "unmonitored" scares me.

David

On Fri, Nov 11, 2016 at 8:14 PM, Oscar Beijbom [email protected] wrote:

I was thinking about image filesize limits. I don't think there is any reason to have limits anymore with the new backend. I'll conduct some tests, and if they work out we can change the policy. A lot of folks are asking about hugs mosaics (eg 20.000 x 20.000) pixels.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/beijbom/coralnet/issues/93, or mute the thread https://github.com/notifications/unsubscribe-auth/AFQOgVZE6unVMlV--dWvnbULTfufWgFjks5q9T09gaJpZM4KwUiI .

[beijbom]

Yeah, sorry, I was unclear. We would definitely still have limits. They would just be higher.

[StephenChan]

Wondering if you have decided on a new max resolution yet.

I do feel like at a certain point, though, there are a couple of questions:

1. Does it make sense for the user to start splitting up their images into multiple smaller ones, not unlike how coral reefs are analyzed by transects and quadrats; or is there a good reason it needs to stay as a single image?

2. Do they really need to upload that level of detail on CoralNet, or can they work with uploading a downscaled version to CoralNet while keeping the original on their hard drive? Although the user has to take the time to resize the images, it means the upload, thumbnail generation, patch generation, annotation tool performance, and vision backend performance will be faster, so it might not even slow down their work overall.

Edit: On further consideration, I might be starting to come around here. Currently discussing with C.C. about unlocking higher resolutions for sources on request. More details to come.

[StephenChan]

Okay, so previously I didn't really understand the applications of orthomosaics (#192), which are the main reason people are asking for these higher limits. But now it does seem like a pain to have to split those mosaics up. In general though, treating those mosaics like standard images probably isn't going to be satisfactory, because:

• It's standard to keep the mosaics' geo-referencing info
• These mosaics may be possible to view in 3D
• These mosaics can just get huge. Thinking "raising 8000 to 30000 will solve mosaics in general" would be far too naive.

So in conclusion, if anyone's asking for higher limits for applications other than mosaics, then we could see what limits they are looking for. Otherwise we should focus on the larger issues, so to speak.