QC Rejecting Perfectly Good Images

[chicagonature]

Hi!

 

I'm having big problems with the Alamy QC system.

I'm an internationally published photographer with 26 years of experience. When I submit my perfectly detailed, sharp, and high-res images, some get rejected, as well as the entire batch! In fact, these are the same image files that I use to make huge prints in sizes up to 40x60. So you can understand my frustration.

 

Thus far, I've found that the Alamy QC system has problems with highly detailed shots of prairies (which is my specialty) and images of flowing water with wonderful motion blur. I can only imagine what it'll do with my fog shots. To make it even more frustrating, the entire batch gets rejected and I can't move forward with keywording the accepted ones.

If you have experience with this issue, please let me know how to solve the problem.

Thanks so much!

     Mike

Edited November 27, 2019 by chicagonature
typo in title

[dunstun365]

On 27/11/2019 at 19:09, geogphotos said:

 

I would resize it to approx 3800 pixels longest side.

 

You need to post a 100% crop for people to judge the image

 

 

The scan size) is likely too large for 35mm film at nearly 50Mpix. I scanned my old 35mm transparancies at 2900 DPI (the max resolution on my old Nikon scanner) which  gave about 12.5 Mpix sized images & were accepted here they needed retouching though.. Curiously they failed on other well known stock websites so the standards are different so it seems to be a juggling act.

Edited November 29, 2019 by dunstun365

[MDM]

2 hours ago, chicagonature said:

Here's a very insightful article that every photographer should read the debunks the myth about Diffraction Limited Apertures (DLA): https://jonrista.com/2013/03/24/the-diffraction-myth/. As the author (Joh Rista) writes and proves with images, "IQ from a photo taken at f/22 can frequently be restored to such a degree that it rivals the IQ of a photo taken at f/4." My work is real-world proof that DLA does not tell the full story, given that I've been shooting between f/22 and f/32 for over two decades, and I produce prints that go as large (or larger) than 40x60". One client produced a 6x9-foot mural in their visitor center from medium format film image shot at f/32. To shoot at f/8 or f/11, as you suggest, would result in horrendously out-of-focus areas in the foreground and background, which would be far worse that what is seen in the image of the bluebells. And, as Jon Rista suggests, diffraction is easier to correct because it's uniform across the entire image. Software focus stacking is impossible for highly detailed landscape images like these. Unfortunately, just the slightest whiff of air will move the flowers a couple of pixels and cause hundreds of artifacts from software. However, I occasionally Manually focus stack, as long as the foreground flowers don't cross into the middle-ground or background. Images shot at higher f-numbers are often easily restored in post using proper micro=contract techniques, like the Clarity and Texture tools in Adobe Camera Raw. And the then there's actual sharpening that should not be applied to the "master" image. Sharpening should only be applied after the image is resized and ready for output, and this is what Alamy calls for.

 

Just minutes ago, I went into my living room to view a 28x42" print of the bluebell image on the wall. Viewing at just 18 inches away (40% of a common viewing distance equal to the print diagonal), it's very hard to see any blurriness on the sharpened image. It's crazy that Alamy does not take viewing distance and future sharpening into account when assessing IQ. That's just standard optical theory using the Circle of Confusion. Nobody looks at a billboard up close. If they did, the dots would be an inch in diameter! Rather, Alamy should be assessing sharpness at "Print Size," probably at 240ppi to 300ppi. This would take into account the viewing distance and the subsequent sharpening.

Therefore, I may need to apply some Adobe PS Smart Sharpening (using Lens Blur) before submitting to Alamy.

Mike  

 

You can't restore what is not there in the first place in an image. I'll stay with the science myself. No time for this sort of argument I'm afraid,

 

 

 

Edited November 29, 2019 by MDM

[spacecadet]

1 hour ago, chicagonature said:

Viewing at just 18 inches away (40% of a common viewing distance equal to the print diagonal), it's very hard to see any blurriness on the sharpened image. It's crazy that Alamy does not take viewing distance and future sharpening into account when assessing IQ. That's just standard optical theory using the Circle of Confusion. Nobody looks at a billboard up close. If they did, the dots would be an inch in diameter! Rather, Alamy should be assessing sharpness at "Print Size," probably at 240ppi to 300ppi. This would take into account the viewing distance and the subsequent sharpening.

 

You won't get anywhere arguing what Alamy's QC policy should be- it is what it is.

We accept it and modify our methods as necessary. You will need to do the same even if that means unlearning a few things.

 

 

[M.Chapman]

On 29/11/2019 at 17:11, chicagonature said:

Here's a very insightful article that every photographer should read the debunks the myth about Diffraction Limited Apertures (DLA): https://jonrista.com/2013/03/24/the-diffraction-myth/. As the author (Joh Rista) writes and proves with images, "IQ from a photo taken at f/22 can frequently be restored to such a degree that it rivals the IQ of a photo taken at f/4." My work is real-world proof that DLA does not tell the full story, given that I've been shooting between f/22 and f/32 for over two decades, and I produce prints that go as large (or larger) than 40x60". One client produced a 6x9-foot mural in their visitor center from medium format film image shot at f/32. To shoot at f/8 or f/11, as you suggest, would result in horrendously out-of-focus areas in the foreground and background, which would be far worse that what is seen in the image of the bluebells. And, as Jon Rista suggests, diffraction is easier to correct because it's uniform across the entire image. Software focus stacking is impossible for highly detailed landscape images like these. Unfortunately, just the slightest whiff of air will move the flowers a couple of pixels and cause hundreds of artifacts from software. However, I occasionally Manually focus stack, as long as the foreground flowers don't cross into the middle-ground or background. Images shot at higher f-numbers are often easily restored in post using proper micro=contract techniques, like the Clarity and Texture tools in Adobe Camera Raw. And the then there's actual sharpening that should not be applied to the "master" image. Sharpening should only be applied after the image is resized and ready for output, and this is what Alamy calls for.

 

Mike  

 

Thanks for posting a link to that interesting and thought provoking article. I find myself somewhat conflicted by the idea that you can shoot at f/22 for example (to maximise depth of field) and then simply apply some extra sharpening to recover the sharpness lost through diffraction. I can see various several supporting and conflicting observations.

 

Photographers have been happy enough to use sharpening to improve images taken on cameras with an Anti-Aliasing filter (reduce moire) in front of the sensor, so that seems like a very similar issue. But... if it was possible to recover all the detail by simply applying extra sharpening, why have camera manufacturers started producing cameras without AA filters? And why, if I sharpen an image of fabric (for example), doesn't the moire suddenly reappear?  It seems like the blurring of the AA filter has permanently removed some (the high frequency) information and I would argue diffraction softening will do the same.

 

Whilst I don't disagree that sharpening an f/22 image can appear to improve the appearance of clearly defined bold edges (as in the Nikon logo example in the video) but what about very fine details subtle details (animal fur, textures etc.). I fear that the diffraction at f/22 softening will turn these subtle details into "mush" which sharpening will never recover. MTF graphs of lens performance, at increasingly small apertures, clearly show the reduction in the ability to relay high frequency components.

 

Applying extra sharpening in an attempt to recover detail lost through shooting at f/22 will increase noise in the image (extra noise due to longer exposure and due to sharpening algorithm). Sure it's possible to remove noise with filtering but this will also remove finer subtle detail from the image. It's also not a technique that will work well on scanned slides as the film grain will become overly visible if additional sharpening is applied. More sophisticated noise filtering might help, but it's clearly a trade-off versus the benefit of increasing depth of field. Personally I tend to manually combine (stack) two images shot with different focal points.

 

In the end it's all a trade-off and I'll be sure try some tests to see what happens in the sort of images I take. Thanks!

 

Update: I just took 2 test test shots of a flat subject containing a number of fine textures (brush bristles and coins on a rattan mat). The first was taken at f/8 (where my lens performs best), the second at f/22, which is quite a bit softer due to diffraction. I then tried various methods of sharpening the second to see if I could match the quality of the first.  I failed. I tried various sharpening and noise reduction settings in LR, I also tried smart sharpening in Photoshop CC, and Deconvolution sharpening in RawTherapee. In the end I gave up. Nothing I did gave me the quality of the image taken at f/8. Maybe I'm just not that good at sharpening, or it's not as easy as the article makes out. If you have any examples that demonstrate how to restore the quality of an image taken at f/22 to match that of an image taken at f/8 it would be useful. The example of a Nikon lens logo and barrel taken at f/22 in the video in the article shows noise and halo problems after sharpening, so isn't a good example.

 

Mark

 

Edited December 3, 2019 by M.Chapman

[M.Chapman]

18 hours ago, chicagonature said:

Just minutes ago, I went into my living room to view a 28x42" print of the bluebell image on the wall. Viewing at just 18 inches away (40% of a common viewing distance equal to the print diagonal), it's very hard to see any blurriness on the sharpened image. It's crazy that Alamy does not take viewing distance and future sharpening into account when assessing IQ. That's just standard optical theory using the Circle of Confusion. Nobody looks at a billboard up close. If they did, the dots would be an inch in diameter! Rather, Alamy should be assessing sharpness at "Print Size," probably at 240ppi to 300ppi. This would take into account the viewing distance and the subsequent sharpening.

 

One possible reason Alamy insist on the image being sharp when viewed at 100% is to give the customer maximum flexibility in how an image is used. It's quite common for customers to only use a portion (a small crop) of an image to pick out a particular person or feature. Looking through the "found images" threads I've often been surprised at just how small a crop from a scenic landscape image (for example) a magazine will use, and they will blow it up to 1/2 page. Under these circumstances the normal "viewing distance" argument of the whole image doesn't apply.  Alamy aren't crazy.

 

As has been mentioned before in this thread, it's all about "calibrating" one's workflow to meet Alamy's requirements.

 

Mark

Edited November 30, 2019 by M.Chapman