Jump to content

Cut out not recognised


Recommended Posts

Just received this response from member services

 

"This is what our cut-out filter looks for: Pure white background, no defined shadows, single object, no negative spaces in the middle (holes in things), clearly defined blank border around object."

 

No quantitative measures then, and I doubt if they really mean no holes in the middle, at least I hope not (see watering can above). What I want really to know is how may pixels are required in the border etc.

Link to comment
Share on other sites

Just received this response from member services

 

"This is what our cut-out filter looks for: Pure white background, no defined shadows, single object, no negative spaces in the middle (holes in things), clearly defined blank border around object."

 

No quantitative measures then, and I doubt if they really mean no holes in the middle, at least I hope not (see watering can above). What I want really to know is how may pixels are required in the border etc.

 

Percentages rather than pixels probably.

 

BY8BPJ.jpgCR6C4B.jpgDFYKG7.jpg
 
No holes in the middle. Right. Luckily the person that devised that algorithm thought it better to ignore that.
(25,000 empty frames found by the filter and perennial best sellers in AoA.)
 
wim
Link to comment
Share on other sites

So I have 71 isolations (isolation is perfect as rendered against transparent background or from an alpha mask created at run time).  The cut out filter shows 52 in a search.  Something under half the missing images aren't on white so that's fair enough if Alamy want to define isolation differently to the rest of the world but the software selection is clearly faulty within its own parameters for the remainder.  Slightly irritating is the fact that searching with the keyword "isolated", which is present on all the images, returns the same images as using the cut out filter.

Link to comment
Share on other sites

So I have 71 isolations (isolation is perfect as rendered against transparent background or from an alpha mask created at run time).  The cut out filter shows 52 in a search.  Something under half the missing images aren't on white so that's fair enough if Alamy want to define isolation differently to the rest of the world but the software selection is clearly faulty within its own parameters for the remainder.  Slightly irritating is the fact that searching with the keyword "isolated", which is present on all the images, returns the same images as using the cut out filter.

The cut-out type keywords seem to be disabled for images not picked up by the filter.  I suppose that is so contributors can't override the system, which would be all very well if the system worked.

 

Pearl

Link to comment
Share on other sites

Has anybody noticed that the zooms from the zoom page are cutting off a part of the bottom of all images? That way some of the borders may not be borders.

 

wim
 
edit: here is an example . This one was not recognized as a cutout.
Click on the image and the medal is touching the frame, while the white upper and lower border are the same width in the original.
The new one did pass, but the lower border is still missing half of it's width at least.
Link to comment
Share on other sites

That's certainly how it appears.  Wonder how it affects folks who legitimately use isolated, isolation etc where it's not referring to a cut out picture - "isolated country farmhouse" sort of thing,,,

 

 

As I said try isolated island or isolated tree. I was wrong there:

 

Luckily clients aren't that dumb. They look for a lonely tree in stead: 3x.

 

A bit more -3300x- are looking for a genuine cutout, when they use the keyword isolated. Ok maybe just 99% of them, because the odd client is looking for a black background or someone isolated in a crowd.

 

wim

Link to comment
Share on other sites

Hello Mr Alamy, anybody there?

 

There are clearly major problems with the software filter for cut outs. It needs to be sorted out. It appears that there are potentially thousands of genuine cut outs within the system that are not being recognised, and therefore business lost, bad for both contributors and agency.

 

In the short term it would be very helpful indeed if contributors could have access either to the program to test their own images before submission, or the detailed set of criteria (quantitative data please not vague descriptions) that are used to judge whether or not an image is truly a cut out. 

 

I've tried asking via member services and have received speedy and polite replies, but nothing of any substance.

Link to comment
Share on other sites

 

Has anybody noticed that the zooms from the zoom page are cutting off a part of the bottom of all images? That way some of the borders may not be borders.

 

wim

 

edit: here is an example . This one was not recognized as a cutout.
Click on the image and the medal is touching the frame, while the white upper and lower border are the same width in the original.
The new one did pass, but the lower border is still missing half of it's width at least.

 

Nothing constructive to offer but, I looked at you example an wondering how to tell where the frame of an isolated on white on a white screen is  :unsure:

Link to comment
Share on other sites

 

 

Has anybody noticed that the zooms from the zoom page are cutting off a part of the bottom of all images? That way some of the borders may not be borders.

 

wim

 

edit: here is an example . This one was not recognized as a cutout.
Click on the image and the medal is touching the frame, while the white upper and lower border are the same width in the original.
The new one did pass, but the lower border is still missing half of it's width at least.

 

Nothing constructive to offer but, I looked at you example an wondering how to tell where the frame of an isolated on white on a white screen is  :unsure:

 

 

Click on the image on the zoom page and you'll get a larger zoom with a slightly darkened background. This is where the cropping happens.

In some browsers and operating systems, you can drag an image over a darker background temporarily, just to have look.

 

If by coincidence the filter happens to work on the image we see when zooming from the zoom page (or in the same way), that would explain the strange results of the filter.

 

However it is beyond me why these

 

childrens-hands-forming-a-round-frame-is

united-hands-dtw1mj.jpg

 

are considered a cut out, but this: 

nobel-prize-medal-e7w3t6.jpg

 

is not. If the major rule is: there should be a white border all around and no hole in the middle.

 

I have solved the problem for mine (the medal)  by increasing the white border by a couple of pixels.

 

wim

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.