Zoom in >100% and slowly scroll up and down in the image, working your way from left to right until you have carefully inspected every area. Incomplete lines, blurry areas, shapes that don't make sense, artifacts, geometry errors - there are so many things to notice. You'll get better with experience.

Generative AI is very bad at checking geometry consistency. If you look at the door, you see that it melts into the wall. Or the window that gets bended, without hinges. Or look at the steps of the ladder. Or the connection of the tube coming down.

Not to talk about the complete nonsense of the pressure vessel at the left.

Looking at the background mess, you also see many tubes and guides and cables, that seam to melt into each other, without clearly defined background and foreground.

Whatever it was, that catched the eye of the moderator, they refuse at that precise first error they see. And as soon as you know where to look, you find something incoherent in such a messy image like this. 

If you want to avoid refusals, you need to check assets at 100% or 200%. Check first the geometry, then check for artefacts. You need to be critical, and surly not hope that the moderator would oversee the issue. Correct it, if you can. If you can't correct, do not submit. Bad assets will hurt your account in the long run.

If you are new to stock, you should consider these resources: https://helpx.adobe.com/stock/contributor/tutorials.html
Please read the contributor user manual for more information on Adobe stock contributions: https://helpx.adobe.com/stock/contributor/user-guide.html
See here for rejection reasons: https://helpx.adobe.com/stock/contributor/help/reasons-for-content-rejection.html
and especially quality and technical issues: https://helpx.adobe.com/stock/contributor/help/quality-and-technical-issues.html

AI is not a replacement for human talent & editing skills.  Despite what you've seen on the Internet, nobody is getting rich from machine generated images alone.  In the right hands, AI can be a good tool the same as a paint brush or palette knife but it can't replace the artist.

Complex AI images contain more errors than simple ones.  After diffusion, closely examine every part of the image at 100-300% magnification & correct errors if you can.  With complex images, this can be tedious & time consuming. Don't rush it, take your time.

Pay attention to small details.  Do the angles, lines and rendering make sense visually? How about the artifacts?  Can they be edited out or would it be quicker to discard the image and start over with a better prompt?  Machines can't make these decisions for you.  It requires human intelligence to correct machine-made errors. 

Read your Stock Contributor User Guide for more tips.  Your goal is to produce commercial-ready artwork that Stock customers will buy.

• https://helpx.adobe.com/stock/contributor/help/reasons-for-content-rejection.html
• https://helpx.adobe.com/stock/contributor/help/quality-and-technical-issues.html
• https://helpx.adobe.com/stock/contributor/user-guide.html/stock/contributor/help/photography-illustrations.ug.html
• https://helpx.adobe.com/stock/how-to/tips-stock-image-acceptance.html
• https://helpx.adobe.com/stock/contributor/help/generative-ai-content.html
  
  

Good luck.

And is there any way to check the image out with Gen AI to catch issues in an image beforehand? Or a process people use manually - or a checklist- to catch these early rather than later? Thanks so much!

 

By @PanOpticus

One thing I do for some images is to create a grid in Photoshop's preferences. You'd need to experiment with the best size depending on the size of the image in question. Then examine each grid section one by one.

Probably due to a lot of minor details that add up and become major. There are a number of objects that or poorly rendered and difficult to determined what they are exactly.

The only way to catch issues in an image is to view it at a minimum of 100% (probably 200% for something as complex as this particular asset).