Does Alamy compress our images?

[M.Chapman]

I was looking back at  a couple of my own images that I purchased from Alamy to check how they handled images with sRGB and AdobeRGB profiles (which proved that images are converted to sRGB and the profile stripped) and I spotted something I didn't notice at the time. The full resolution personal use images I downloaded from Alamy have a jpg file size which is roughly 1/2 the size of the jpg files I uploaded. The pixel dimensions are the same, but the file size has been reduced by almost 50% suggesting a higher level of jpg compression is being used. The images I uploaded were jpgs with quality level 10. The files supplied by Alamy appear to be jpgs with a quality level of about 8. Close inspection of the images reveals some loss of detail and jpg compression artefacts. These differences are visible when inspecting at 200% but subtle at 100%. Nevertheless, given Alamy's emphasis on QC inspection at 100% I was quite surprised to find my images had been compressed by Alamy. I wonder if this compression happens before or after QC?

 

I've no idea whether the same compression is applied to all licence types, or whether the compression is relative (if I'd uploaded at Level 12 would I have received quality 10 back?), or whether this is a recent change (to conserve valuable server space). Has anyone else got files they have purchased that they can compare with the originals, or on which they can check the jpg compression being applied?

 

Mark

Edited March 7, 2020 by M.Chapman

[geogphotos]

I'm fairly sure that Alamy have always done this. 

[Harry Harrison]

21 minutes ago, M.Chapman said:

 I was quite surprised to find my images had been compressed by Alamy.

I'm just hypothesising, but we know that our metadata is stripped out, and thanks to your investigations we know that images are converted to sRGB so that rather suggests that the jpegs are resaved in some kind of server based image processor (ImageMagick?), maybe the file size reduction is a consequence of that?

[MDM]

1 hour ago, M.Chapman said:

I was looking back at  a couple of my own images that I purchased from Alamy to check how they handled images with sRGB and AdobeRGB profiles (which proved that images are converted to sRGB and the profile stripped) and I spotted something I didn't notice at the time. The full resolution personal use images I downloaded from Alamy have a jpg file size which is roughly 1/2 the size of the jpg files I uploaded. The pixel dimensions are the same, but the file size has been reduced by almost 50% suggesting a higher level of jpg compression is being used. The images I uploaded were jpgs with quality level 10. The files supplied by Alamy appear to be jpgs with a quality level of about 8. Close inspection of the images reveals some loss of detail and jpg compression artefacts. These differences are visible when inspecting at 200% but subtle at 100%. Nevertheless, given Alamy's emphasis on QC inspection at 100% I was quite surprised to find my images had been compressed by Alamy. I wonder if this compression happens before or after QC?

 

I've no idea whether the same compression is applied to all licence types, or whether the compression is relative (if I'd uploaded at Level 12 would I have received quality 10 back?), or whether this is a recent change (to conserve valuable server space). Has anyone else got files they have purchased that they can compare with the originals, or on which they can check the jpg compression being applied?

 

Mark

 

It has been obvious for a while that Alamy have been doing some extra magic on the images for sale as the compressed download sizes have been tiny. I noticed this quite some time ago but I don't recall when this started.  You can't compress a JPEG much if at all using zip compression or the like so it has to be happening within the file itself.

 

The fact is that for most end user purposes this does not matter as most image use is online anyway. Examining images at 200% is unnecessary and may eventually cause problems with eyesight. I would suggest not spending too much time worrying about it. Life is too short and you are not going to change anything. 

[Phil Robinson]

I've always uploaded JPEGs at quality level 12 - apart from live news at 10 or 11, and then only if I'm in a cafe with slow WiFi.
Seems this may have been a good idea. Back in the olden days, Alamy wanted TIFF format and I suppose I assumed that level 12 would be the best to send after they changed.

[M.Chapman]

2 hours ago, MDM said:

The fact is that for most end user purposes this does not matter as most image use is online anyway. Examining images at 200% is unnecessary and may eventually cause problems with eyesight. I would suggest not spending too much time worrying about it.

 

Indeed. Nevertheless we are still being asked to carefully inspect all our images at 100%, and then Alamy "squashes them" and discards the colour profile.  All part of a slide toward pile 'em high and sell 'em cheap? Does it start to make "pixel peeping" QC and discussions on this forum (I confess I'm guilty as charged) rather academic? Maybe we should have ignored those "whisps" at the edge of the image that Topaz DeNoise creates (our discussion in the slide copying thread) since Alamy's extra compression is causing (IMHO) similar degradation that's not just at the edge of the frame.

 

2 hours ago, MDM said:

Life is too short and you are not going to change anything. 

 

Yes, Alamy's portfolio continues to grow so this image compression is probably here to stay (to conserve server space and preserve response times). Maybe Alamy will start inpecting at 50% instead of 100% to save time?

 

Mark

Edited March 7, 2020 by M.Chapman

[MDM]

1 hour ago, M.Chapman said:

 

Indeed. Nevertheless we are still being asked to carefully inspect all our images at 100%, and then Alamy "squashes them" and discards the colour profile.  All part of a slide toward pile 'em high and sell 'em cheap? Does it start to make "pixel peeping" QC and discussions on this forum (I confess I'm guilty as charged) rather academic? Maybe we should have ignored those "whisps" at the edge of the image that Topaz DeNoise creates (our discussion in the slide copying thread) since Alamy's extra compression compression is causing (IMHO) similar degradation that's not just at the edge of the frame.

 

 

Yes, Alamy's portfolio continues to grow so this image compression is probably here to stay (to conserve server space and preserve response times). Maybe Alamy will start inpecting at 50% instead of 100% to save time?

 

Mark

 

The effects of JPEG compression to level 8 have little or more likely no discernible effect on the criteria that Alamy are using for QC for the vast majority of images: primarily image sharpness. So that is not really an issue. The real problems with JPEG compression probably start to arise after continued editing and resaving. As this is something that is completely out of my control I am not going to worry about it at all. 

 

As far as Topaz DeNoise goes, I was seeing artifacts within the images, typically at boundaries within the images themselves and not just at the edges. 

 

 

[M.Chapman]

Checking some of my first uploads I see that these also show additional compression (in the quoted download size), in one case to 25% of the size I uploaded. 

 

Thinking about this logically . If Alamy needs to host sRGB jpg files that are more compressed ( for example to level 8 ) to save server space, then why not ask contributors to upload to this standard? This would give a cleaner/better end result as Alamy wouldn’t have to reprocess the image data at all (they could simply strip the metadata they don’t want to pass on). In contributors could go straight from 16 bit data in a large colour space (the best source) in PS/LR etc. to the compressed sRGB jpg that the customer will receive in a single step. At the moment it appears that Alamy opens our jpgs (that are already slightly compressed), then converts the 8 bit data from AdobeRGB to sRGB (if needed) and then re-saves at higher compression. This multi-step process isn't ideal, especially on 8 bit data. If contributors directly uploaded what the customer will receive (sRGB jpg quality level 8?), the customer would get a cleaner file and Alamy would save processing time. A win win?

 

But….. would Alamy want to declare that they want sRGB jpgs at quality level 8? Or would they prefer to preserve the impression that their images are higher quality.... Mmmm...

 

Mark (just musing)

Edited March 7, 2020 by M.Chapman

[Betty LaRue]

10 hours ago, MDM said:

The fact is that for most end user purposes this does not matter as most image use is online anyway. 

How about the “other” uses? Billboards, large display boards seen entering a business?
Just pretend that our higher megapixel cameras are actually retro 6MP?
What happens to the images we’ve reduced down to the minimum already?

Seems bad to me, but what do I know. Nothing, it seems! 

[M.Chapman]

8 hours ago, MDM said:

The effects of JPEG compression to level 8 have little or more likely no discernible effect on the criteria that Alamy are using for QC for the vast majority of images: primarily image sharpness. So that is not really an issue. The real problems with JPEG compression probably start to arise after continued editing and resaving. As this is something that is completely out of my control I am not going to worry about it at all. 

 

As far as Topaz DeNoise goes, I was seeing artifacts within the images, typically at boundaries within the images themselves and not just at the edges. 

 

To my eyes (looking at 100%) I see a not too dissimilar level of artefact to those produced by Topaz if I save at jpg level 8. But it does depend (in Topaz) on how sharp the edges are to start with. The sharper the input edges the more noticeable the artefacts in Topaz output, but I see a similar effect with jpg compression.

 

Mark

[MDM]

I am seeing different artifacts in Topaz then as there is no way to produce this sort of thing by saving as level 8 or any other level JPEG. I will see if I kept the stuff I did and post next week.

[MDM]

28 minutes ago, Betty LaRue said:

How about the “other” uses? Billboards, large display boards seen entering a business?
Just pretend that our higher megapixel cameras are actually retro 6MP?
What happens to the images we’ve reduced down to the minimum already?

Seems bad to me, but what do I know. Nothing, it seems! 

 

BIllboards use a totally different type of printing and don't require very high res images. Something I know very little about except that that is a fact. 

 

But we are not talking about downsizing. We are talking about JPEG compression here which is entirely different. Where it might matter is in larger prints (inkjet or wet chemistry).

[meanderingemu]

1 hour ago, M.Chapman said:

Checking some of my first uploads I see that these also show additional compression (in the quoted download size), in one case to 25% of the size I uploaded. 

 

Thinking about this logically . If Alamy needs to host sRGB jpg files that are more compressed ( for example to level 8 ) to save server space, then why not ask contributors to upload to this standard? This would give a cleaner/better end result as Alamy wouldn’t have to reprocess the image data at all (they could simply strip the metadata they don’t want to pass on). In contributors could go straight from 16 bit data in a large colour space (the best source) in PS/LR etc. to the compressed sRGB jpg that the customer will receive in a single step. At the moment it appears that Alamy opens our jpgs (that are already slightly compressed), then converts the 8 bit data from AdobeRGB to sRGB (if needed) and then re-saves at higher compression. This multi-step process isn't ideal, especially on 8 bit data. If contributors directly uploaded what the customer will receive (sRGB jpg quality level 8?), the customer would get a cleaner file and Alamy would save processing time. A win win?

 

But….. would Alamy want to declare that they want sRGB jpgs at quality level 8? Or would they prefer to preserve the impression that their images are higher quality.... Mmmm...

 

Mark (just musing)

fine .  what's level 8 equivalent in %?

[M.Chapman]

10 hours ago, meanderingemu said:

fine .  what's level 8 equivalent in %?

 

Mmmm.... you may have found a flaw in my suggestion. The relationship might vary depending on the software package being used. In Adobe products there appears to be some consistency.  The Photoshop jpg quality scale goes from 0 - 12 whilst Lightroom goes from 0 -100% and this article gives shows a relationship table https://photographylife.com/jpeg-compression-levels-in-photoshop-and-lightroom. Assuming this article is correct a LR jpg quality level of 62% - 69% appears to be equivalent to level 8 in Photoshop. But I can't comment on other packages.

 

Mark

[meanderingemu]

3 hours ago, M.Chapman said:

 

Mmmm.... you may have found a flaw in my suggestion. The relationship might vary depending on the software package being used. In Adobe products there appears to be some consistency.  The Photoshop jpg quality scale goes from 0 - 12 whilst Lightroom goes from 0 -100% and this article gives shows a relationship table https://photographylife.com/jpeg-compression-levels-in-photoshop-and-lightroom. Assuming this article is correct a LR jpg quality level of 62% - 69% appears to be equivalent to level 8 in Photoshop. But I can't comment on other packages.

 

Mark

 

🙂  i use C1.  I know some images i do at 90% when i want speedy uploads and i have bad Wifi,  but now i'm curious how low i could go.  

 

i'll try some tests later. 

[M.Chapman]

34 minutes ago, meanderingemu said:

 

🙂  i use C1.  I know some images i do at 90% when i want speedy uploads and i have bad Wifi,  but now i'm curious how low i could go.  

 

i'll try some tests later. 

 

I wouldn't go too low. Even if Alamy do compress the jpgs to level 8 (or the C1 equivalent), it probably wouldn't be a good idea to submit images at that level because Alamy (with their current process) will still open (decompress) the image, convert the data to sRGB (if image is AdobeRGB) and then re-save (compress). Putting lower quality data into such a process isn't a good idea. It's generally reckoned that it's best to submit at quality level 10 or above to Alamy.

 

I guess one way to check for rough equivalence between the levels of compression in different software packages would be to save the same file at different compression levels in PS, LR and C1 etc and then compare the file sizes.

 

Mark

 

[geogphotos]

This may be very simple but.......since  I know that Alamy are compressing then that gives me more reason to send in at the highest quality I can.

[spacecadet]

I found out by accident (LR decided to reset itself after a hard restart) that the difference in file size between 60% quality (apparently LR default) and the 90% I usually use is a factor of 4 or 5. It is definitely noticeable at 100% on film scans. On a digital original it takes the edge off sharpness but not enough to compromise QC on the example I tried. I certainly can't see the difference between LR 90 and 100% quality at 100% enlargement.

It might have mattered once, but not now that marginal bandwidth and storage are effectively free. I'll stick to 90%, which by the look of that chart is about PS 8.

Although whenever I take a jpeg from PS I use 10. Hey ho.

Edited March 8, 2020 by spacecadet

[M.Chapman]

6 hours ago, geogphotos said:

This may be very simple but.......since  I know that Alamy are compressing then that gives me more reason to send in at the highest quality I can.

 

Simple but effective, especially as I don't think we know whether Alamy's compression level is relative.

• For example upload at jpg level 12, Alamy compress to level 10. Upload at jpg level 10 Alamy compress to level 8

Or absolute

• For example upload at jpg level 12, Alamy compress to level 8. Upload at jpg level 10 Alamy still compress to level 8

I suspect their compression is absolute (always to a fixed level), but currently have no evidence to support that.

But it's always good to upload a quality that's higher than Alamy needs methinks, so that any losses their processes introduce don't cause problems.

I'm sticking with uploading at level 10.

 

Mark

[Robert M Estall]

I've always uploaded at jpg level 10 Adobe RGB mostly due to pretty slow modem speed. DIAL UP! It's pegging along about 70Mbps down/15Mbps  up these days but I see no real reason to go to level 12. There has long been a belief that Alamy supply as level 8 but I don't recall ever seeing that confirmed. It's still a huge amount of server space no matter how you slice it.

Edited March 11, 2020 by Robert M Estall

[Klinger]

On 08/03/2020 at 15:56, M.Chapman said:

 

Simple but effective, especially as I don't think we know whether Alamy's compression level is relative.

• For example upload at jpg level 12, Alamy compress to level 10. Upload at jpg level 10 Alamy compress to level 8

Or absolute

• For example upload at jpg level 12, Alamy compress to level 8. Upload at jpg level 10 Alamy still compress to level 8

I suspect their compression is absolute (always to a fixed level), but currently have no evidence to support that.

But it's always good to upload a quality that's higher than Alamy needs methinks, so that any losses their processes introduce don't cause problems.

I'm sticking with uploading at level 10.

 

Mark

 

Same here, I'm always at level 10, sRGB, but that doesn't mean anything scientific, just that I'm happy with that size for storage and uploading.

 

Note: metadata can be removed, without changing the image. I think someone had asked about that? Lossless operation, where the image data is not altered.

 

Edited March 16, 2020 by Klinger