Increased rejections - is there a new assessment process?

[quetzal]

I have been submitting to Alamy for nearly 15 years and have over 15,000 images on sale. I only submit wildlife images. Over the last few months I have had more rejections of submissions than usual. The main 'fault' given as a reason was 'soft or lacking in detail'. I have not changed the way I take my images, nor my camera / lens combination, nor the way I process them - except I now use DXO RAW and Topaz denoise and sharpen where necessary. So, as far as I am concerned, the images should be the same 'quality' (actually, possibly better) as I have hitherto submitted. 

When I look at some of the rejected images, the subject (bird, mammal etc.) is clear and sharply in focus. However, there are areas in the image such as out of focus background vegetation, which I believe is causing the problem. My suspicion is that a new system of analysis of images has taken place at Alamy - using AI to assess the images. If this is indeed the case, it puts me in a difficult position. As I only submit wildlife, there is inevitably going to be some part of the image which will not be in the same plane of focus as the primary subject. 

Has anyone else, perhaps also a wildlife worker, experienced a similar problem? It would be great to have a response from an Alamy representative too.

Thanks, Kevin

 

[Mr Standfast]

Sorry to hear that. Although not a nature shooter I have not experienced this. Can you upload some images for the forum to look at.

 

(Nice portfolio!!)

Edited July 30, 2023 by Mr Standfast
Tidy up after coffee kicked in

[quetzal]

Hi - sure but I can't see how I can upload images. Can you tell me please? I can only see "Insert image from URL"

Kevin

[Martin L]

Host them on a website and add the link.

If you haven't access use one of the free ones like imgbb.

I have not noticed any change in QC and I have quite a few macro insect shots / wildlife shots with diffuse backgrounds which always causes an issue if something like AI is used.

I have not failed QC since I have been here, (well I haven't failed yet  )

 

Edit PS: Is the new denoising just too strong creating artifacts rather than softness?

Edited July 30, 2023 by Martin L
Added a bit

[quetzal]

Thanks Martin.

Here are two examples ( I have reduced their size for benefit of uploading). 

https://ibb.co/0KmJjBx

https://ibb.co/Kwk89Mt

 

Kevin

[spacecadet]

8 minutes ago, quetzal said:

Thanks Martin.

Here are two examples ( I have reduced their size for benefit of uploading). 

https://ibb.co/0KmJjBx

https://ibb.co/Kwk89Mt

 

Kevin

We can't see the quality unless you upload at 100%.

[quetzal]

Good point! Here's the full size images

 

https://ibb.co/PQn9cHM

 

https://ibb.co/MGCHG7w

 

Kevin 

[spacecadet]

The kingfisher image has what looks like noise on the plumage but I assume it's actually an artefact of processing.

The warbler looks very "plastic"- it reminds me of the faults that got me failed quite often before I started shooting raw- in my case it was usually excessive luminance noise reduction.

You mention you've started using some new software- if you haven't changed anything else, I suggest you look into that. I don't use those programs myself so can't offer any thoughts on it.

Edited July 30, 2023 by spacecadet

[quetzal]

Thanks Spacecadet. I have had plenty of submissions accepted since starting with DXO and Topaz, so I find it hard to understand or blame those for my recent failures. I look ofrward to hearing other comments. 

[spacecadet]

5 minutes ago, quetzal said:

Thanks Spacecadet. I have had plenty of submissions accepted since starting with DXO and Topaz, so I find it hard to understand or blame those for my recent failures. I look ofrward to hearing other comments. 

Well, the rules say images must be sharp at 100% and those, to my eye, are clearly not. Is it possible that something has changed and affected your ability to assess sharpness- a new monitor, or not viewing at 100% maybe? My own assessments became easier a few years ago when I got new spectacles.

Edited July 30, 2023 by spacecadet

[Mr Standfast]

Kingfisher is definitely a decisive moment. 😄  Well done.

 

To me neither image is quite right. As SC says it looks like processing, not focus.

 

Soft or lacking definition as often used as a catch all comment when something is  not quite right but is hard to quantify. I hope someone else can provide clarity.

 

[Martin L]

In my view both birds are over sharpened, a real 'crusty' feel to the plumage, as James and SC said its been 'over processed', tempting when you try to make it the focus of the image.

Shame as they are great captures but not all lost as they can be re-processed to make them a little more natural which for me is a little less sharpening to give the feathers a little softer appearance. I know a bit strange to make them softer when the failure was 'soft and lacking definition' but as James said, its a catch all comment for 'not quite right'

 

 

21 minutes ago, quetzal said:

Thanks Spacecadet. I have had plenty of submissions accepted since starting with DXO and Topaz, so I find it hard to understand or blame those for my recent failures. I look ofrward to hearing other comments. 

 

Don't forget, QC only spot check so some others may have slipped through that if spotted may also have failed. Also once one fails you can go on the naughty step so further submissions maybe scrutinised more carefully.

[IKuzmin]

13 minutes ago, Martin L said:

In my view both birds are over sharpened, a real 'crusty' feel to the plumage

+1

Kingfisher just out of order completely, you cannot rescue a capture that you missed with either DXO and/or Topaz, none of them is a "magic bullet".

[spacecadet]

You could also downsize to 3250 long side. I only do it with high ISO images, and then only on an image I'm really keen on- like the kingfisher.

[M.Chapman]

1 hour ago, quetzal said:

Good point! Here's the full size images

 

https://ibb.co/PQn9cHM

 

https://ibb.co/MGCHG7w

 

Kevin 

Looking at 100% view. IMO massively over-processed giving an illusion of sharpness with little "natural detail" left. I often use Topaz Denoise AI but find I have to be very careful with it. It's great at removing noise, but even on the lowest sharpening setting it has a tendency to over-cook things. So I tend to apply on a new layer and then blend back 50/50 with the original. It's particularly bad if applied to an image which has already been sharpened.

 

Looking at the EXIF data I can see the images were shot at ISO 3200 and 6400 on a Sony A9 at relatively small apertures (f/9 and f/13). Great for depth of focus, but might be better to open up the aperture a bit, reduce the ISO and rely a bit less on Topax and DXO. Alternatively just reduce the processing and down-size to 2000 x 3000. I'm sure the Warbler would pass then. Not sure about the Kingfisher without seeing the original.

 

Mark

Edited July 30, 2023 by M.Chapman

[IKuzmin]

Disclaimer: I never used Sony a9, and do not know how good is the lens used. My opinion below is based on Canon use (several  bodies and L lenses).

 

2 hours ago, M.Chapman said:

Topaz Denoise AI but find I have to be very careful with it. It's great at removing noise, but even on the lowest sharpening setting it has a tendency to over-cook things. So I tend to apply on a new layer and then blend back 50/50 with the original.

That's exactly how I use Topaz Sharpen AI because it also denoises and gives additional benefits compared to Denoise AI. I filter through 50-99% layer masks, often using brushes of different density on these masks as well.

 

2 hours ago, M.Chapman said:

ISO 3200 and 6400 on a Sony A9 at relatively small apertures (f/9 and f/13). Great for depth of focus, but might be better to open up the aperture a bit

These ISO should not be a problem on a ff 24 Mp sensor, as well as F=9=11, diffraction-wise. Something above 10 would worsen images on high-resolution sensors but not that one, IMHO and this site:

https://www.photopills.com/calculators/diffraction

I often use F=13 on a crop 20 Mp sensor (macro with flashes) and do not visually see any degradation even though it should be present based on the math above.

Edited July 30, 2023 by IKuzmin
Disclaimer added

[M.Chapman]

3 hours ago, IKuzmin said:

These ISO should not be a problem on a ff 24 Mp sensor, as well as F=9=11, diffraction-wise. Something above 10 would worsen images on high-resolution sensors but not that one, IMHO and this site:

https://www.photopills.com/calculators/diffraction

I often use F=13 on a crop 20 Mp sensor (macro with flashes) and do not visually see any degradation even though it should be present based on the math above.

You're right, noise shouldn't be a major issue (FF+24MP stacked BICMOS sensor). So I wonder why OP used Topaz Denoise? Perhaps it was predominantly for the sharpening function? I note the lens being used for the warbler was a FE 200-600mm zoom with 2x convertor. The Kingfisher was the same lens with a 1.4x convertor. I'm not familiar with these lenses and convertors, but maybe they caused some softness that the OP has tried to recover with DXO and Topaz? If so, downsizing might have been a better way to go to pass Alamy QC.

 

Mark

Edited July 30, 2023 by M.Chapman

[IKuzmin]

8 minutes ago, M.Chapman said:

downsizing might have been a better way to go to pass Alamy QC

+1, I do this routinely.

[M.Chapman]

28 minutes ago, IKuzmin said:

+1, I do this routinely.

Me too. Downsizing allows me to use a 10x zoom (14-140mm) as my normal "travel" lens instead of having to carry several lenses around.

 

Mark

[MDM]

I agree with most of the comments - both images have been fundamentally destroyed by overprocessing. However, looking at the metadata, it becomes clear that the OP is an FRPS which makes the post all the more surprising. I have wondered at times if Alamy are using some sort of AI on frontline QC but there is no doubt here that both images have been mashed in post. 

 

The one thing I totally disagree with is the idea that noise should not be an issue at those sorts of ISOs. I'm a Nikon user but the Nikon and Sony FF sensors are very similar (most made by Sony) and lumimance noise will most certainly be an issue at those ISOs and well below those ISO values. It is easy to treat but please dispel any notion that it is not necessary to do so.

 

[IKuzmin]

1 hour ago, MDM said:

lumimance noise will most certainly be an issue at those ISOs and well below those ISO values. It is easy to treat but please dispel any notion that it is not necessary to do so.

I do not think that a moderate noise is necessary to treat. It used to be in photographs, starting with grain during the film era. If noise/detail ratio is good, and noise is not something really hurting my eye, I do not treat it (and where we all were before the invention of the modern noise reduction software, when only some initial versions of NeatImage and Imagenomics were the only options to consider?)

[M.Chapman]

8 hours ago, MDM said:

The one thing I totally disagree with is the idea that noise should not be an issue at those sorts of ISOs.

Thanks for confirming. As you know I'm not a FF user, so When IKuzmin said noise shouldn't be a problem, I went and looked at DPReview's test images for the Sony at ISO6400 and although there was clearly noise, I felt it wasn't excessive. So not a "major issue" and something that normal NR processing would happily deal with, without the need to resort to Topaz Denoise AI which can be a bit of a "sledgehammer" and can easily create other problems (e.g. artificial looking synthesised "detail").

 

Most of my Topaz Denoise AI use is on digitised slides - to reduce film grain.

 

Mark

Edited July 31, 2023 by M.Chapman

[MDM]

8 hours ago, IKuzmin said:

I do not think that a moderate noise is necessary to treat. It used to be in photographs, starting with grain during the film era. If noise/detail ratio is good, and noise is not something really hurting my eye, I do not treat it (and where we all were before the invention of the modern noise reduction software, when only some initial versions of NeatImage and Imagenomics were the only options to consider?)

 

It's a matter of taste. I don't like digital noise so I treat it with a simple LR preset the strength of which depends on the ISO and the degree of underexposure if any. I'm not using the new LR AI thing as it adds a new step to my workflow and for the most part I don't need it.

 

I think there is also a degree of subjectivity in terms of perception of luminance noise which varies between photographers. The monitor used to view is also important Noise is much more apparent on a matte monitor than on a shiny iMac screen. 

 

The comparison with film grain is a little more complex as well. Fuji Velvia 50 was virtually grainless when processed normally in comparison to an underxposed, push-processed film. Again a matter of taste as to what was acceptable.

[MDM]

1 hour ago, M.Chapman said:

Thanks for confirming. As you know I'm not a FF user, so When IKuzmin said noise shouldn't be a problem, I went and looked at DPReview's test images for the Sony at ISO6400 and although there was clearly noise, I felt it wasn't excessive. So not a "major issue" and something that normal NR processing would happily deal with, without the need to resort to Topaz Denoise AI which can be a bit of a "sledgehammer" and can easily create other problems (e.g. artificial looking synthesised "detail").

 

Most of my Topaz Denoise AI use is on digitised slides - to reduce film grain.

 

Mark

 

I've not looked at the Sony A9 test images (will do when I get a chance) but I know the Nikon Z6II (24MP FF sensor) has a serious problem with noise at anything above ISO 1600 or therabouts. Yet I know other photographers who think that noise is not problematic on the same camera at ISO 12,800. Again a matter of taste and perception. 

 

Yes we had the conversation several times. I have not even had Topax Denoise installed now for several years.

Edited July 31, 2023 by MDM

[Inchiquin]

I looked at the images before I read any of the comments and my first impression was that they both look somewhat artificial. I put it down to over-sharpening and now that I've read the comments  it looks as though others feel the same.

 

Alan