Similar images already submitted!!!

[Sobhy_AI]

Hello there,

I went from a 95% acceptance rate to 20% or less. Almost all of my submissions are being rejected for the "similar image" reason.

The quality of the images is really good and doesn’t have any issues.

I am uploading at most 2 or 3 variations of each object ,PNGs only and a variety of different items across many categories. I also make sure not to upload similar content that I’ve previously submitted. I am 100% certain that my entire account doesn’t have more than three variations of any given item.

Sometimes, I upload the same object in different styles.

for example, a car in a 3D style and another version in watercolor or clay, so they are still distinct. (2 to 3 variations for each style)

Is this a general issue that will be resolved soon, or is it only affecting me?

[inzo_77]

A certain degree of relief for the issue of image rejection can be obtained through these trends.

Create images based on the trends listed below. That way, to some extent, you will be able to avoid the issue of image rejection. ☺️☺️

Trends:

Immersive Appeal

Time Wrap

Levity and Laughter

Fantastic Frontiers

[Verchinskii]

Adobe now clearly indicates the reason for image rejection - similar images in our collection.

[Abambo]

Adobe did that since ever! But they always claimed checking against the whole database, for common motives.

Abambo
via mobile phone.

[Jason37175475psxg]

The system is junk there new meanning means nothing.  I have been testing it to see if I can get images to be accepted and this take it only works for Illustrations.  I have been doing baches of 5 giving it 10 hours to accept the images.  All of my test images have been accepted but it did take several tries for many of them to even get through.  Most of the time it would accept maybe 3 at most I did get an entire batch through once though. It will defintly not work on photographs because they never go through as fast as an Illustration does.  Hope this helps some of you even though it is a pain in the butt and much more time consuming. I was only able to get around 50 in this week done compared too the 100's we were doing before.  So its up to you if you want to wait it out or be consumed by short turnarounds.       

[Jill_C]

Perhaps this is part of Adobe's strategy - submitting hundreds a week generally means quantity rather than quality...

[Nancy OShea]

SUNSETS are awesome

[Jill_C]

I actually have quite a few sunsets that have sold many times - but they're not "just" a sunset - rays of golden light on the Sea of Galilee, paddle wheeler boats silhouetted on the Mississippi River, sailboats in the harbor in Oban, Scotland, a silhouetted couple walking on Pismo Beach, a giraffe striding in front of an acacia tree in Kenya, etc.  However, with the very aggressive "similars" algorithm that is rejecting lots of content these days, I suppose they might be rejected as too similar to the many other sunsets in the database, which is a pity - all sunsets are not the same!

[Юрий253727338kk3]

That is awesome! All those sunsets will be on other stock agencies due to such rejection policy and that would be the diversity and good competition between agencies. that lead to better choice for customers which stock agency to visit)

[Abambo]

There is no need for another sunset! Focus on sunrise pictures. 🤯

[daniellei4510]

A family enjoying a picnic in the country during a total eclipse.

[jie_8284]

That's just black noise

[Svetlana_Lichten]

ha-ha-hah ))) I like it )))

[daniellei4510]

More evidence that the process is random, faulty, or illogical in its decision-making.

This asset was accepted about a month ago:


This asset was submitted last night and accepted today (though nearly identical):


This asset was rejected for being too similar to other assets in the database (at the same time as the accepted one above):

[jacquelingphoto2017]

The two top pictures are totally different. Although it looks like the same lady, one was when she had clean face, the other was when she had a skin issue. Those two images are not similar. Each image carries something different that is of value to the customer. If it were not an error, the moderator saw in your profile another image looking close to the third image in which case this image would not add something different that would be of value to the customer. 

Best wishes

Jacquelin

[Abambo]

The two top pictures are totally different.

By @jacquelingphoto2017

The two images are visually similar. You need to read the whole thread to understand the point @daniellei4510 is making. He is not complaining, or asking why, he is simply adding some value to this discussion by demonstrating the randomness of the new similar rejection wave.

[JanosJ]

My take on similar images and if the powers to be are listening a suggestion. I could not agree more that an image submitted is similar as a scene, however not as a location.

I recently submitted a river scene in Latvia, Meanders of Upper Daugava. There is nothing or very little unique in the picture as such. However the scene, locality is a tentative world heritage site. As such it can have commercial value. Such sites are key destinations for many travelers, and thus to organisations promoting travel. They will not use just any bend in any river, but one of the real location. I could not find another in Adobe library. Now if I just am so poor in my search this comment is of no value. But if not, then pure scene specific declination hurts Adobe, creators. Yes, I did submit with the description, narrative.

[Jill_C]

I agree. I've had several images rejected recently on the basis of "similars". One is a field of rock cairns high above the fjord in Trondheim, Norway. There are other rock cairns in the database, but none from this particular and very beautiful location. All rock cairns are not the same as the "dumb algorithm" seems to conclude. In their pursuit of eliminating redundant content, Adobe is actually doing just the opposite - screening out unique content!

[Aleks_023]

Hi everyone,

I just wanted to chime in here and share my recent experience, which aligns closely with what many of you have already mentioned in this thread.

In the last couple of weeks, I’ve had a large batch of my images rejected under the reason “Similar content already submitted”. What struck me is not just the volume of rejections (several hundred), but the fact that these images are, in my opinion, quite distinct — not only from each other, but also from anything that’s already present on Adobe Stock.

To better understand the issue, I performed a thorough review of my own portfolio:
-I used Adobe Stock’s own search tools to look for similar images using keywords and tags.
-I ran AI-based visual similarity checks using tools like Google Vision AI and other platforms to compare my images with ones available on Adobe Stock.
-In none of the tests did the similarity go over 70%, and most of the time it stayed well below 60%.
-According to Adobe's own guidelines (Similar vs Spamming), this level of difference should be acceptable — especially when images differ in composition, lighting, angle, color, or framing.

I fully understand that spammy uploads and excessive duplication should be avoided, and I absolutely support quality over quantity. But these automated or overly strict rejection patterns are really discouraging — especially when we don't get a chance to clarify or appeal.

Like many others here, I would really appreciate:
-A clearer explanation or updated criteria if the policy has recently changed.
-A chance for manual review or resubmission of wrongly rejected content.
-More transparency about how similarity is assessed – is it purely algorithmic now?

I’m passionate about contributing to Adobe Stock, and I genuinely care about the quality of what I upload. But this current situation is disheartening — not only for me, but for many contributors who put effort into creating unique, high-quality content.

Thanks to all who’ve shared their experiences. Let’s hope Adobe Support takes notice and provides some clarity soon.

[daniellei4510]

-A clearer explanation or updated criteria if the policy has recently changed.
-A chance for manual review or resubmission of wrongly rejected content.
-More transparency about how similarity is assessed – is it purely algorithmic now?

No's 1 and 2 are never going to happen. It took Adobe a couple of months just to update the canned rejection text to how it reads now (when it stated clearly that the similarities were due to assets in the contributor's own portfolio). For #2, it's just never going to happen. They would need to hire dozens of additional moderators to take time for this. For #3, given the randomness of such rejections, I think we must assume that there is a poorly configured algorithm based on keywords and titles (or AI, based on the actual images, which I doubt) that is acting as a gatekeeper and these rejected assets are never seen by the moderators 

[Aleks_023]

I understand that Adobe's moderation process likely involves multiple steps, possibly starting with AI-based evaluations to assess compliance with contributor guidelines, followed by checks for content similarity and quality. It's plausible that the similarity assessments rely heavily on metadata—such as titles, descriptions, and keywords—rather than on detailed visual analysis of the images themselves. Implementing comprehensive AI-driven visual comparisons for every submission would indeed require substantial resources and might not be feasible at scale.

Given the vast number of images uploaded daily, it's conceivable that contributors inadvertently use similar or identical metadata, especially when adhering to standard tagging practices. This overlap could lead to the algorithm flagging distinct images as duplicates, resulting in rejections that may not accurately reflect the uniqueness of the content.

To address this, it would be beneficial for Adobe to provide more transparency about the criteria used for similarity assessments. Understanding whether the evaluations are primarily metadata-driven or if visual elements are also considered could help contributors tailor their submissions more effectively. Additionally, clearer guidelines on optimizing metadata to minimize unintended similarities would be invaluable.

In my latest batch of submissions, I noticed that 100% of the images were rejected under the "Similar content" reason. This was unexpected, especially considering that each image had a unique title, crafted to reflect its distinct content. And images themselves were notably different, with no more than three sharing similar themes or compositions... while the tags used were relevant, it's possible that some overlap with existing content on the platform occurred.

I understand that with the vast number of submissions Adobe Stock receives daily, automated systems play a crucial role in the initial review process. It's plausible that algorithms assess similarities based on metadata, such as titles and keywords, rather than conducting in-depth visual analyses of the images themselves.

Aleks

[JanosJ]

My impression, as described earlier in this thread, was that image similarity is based on visual analysis, - ignoring the title, description, keywords. I submitted a riverbend, also called meander, of a specific locality. This locality is a Tentative World Heritage Site. These the prime tourist assets of a country. Heritage sites also have relevance depending on criteria they are chosen, say in my case historical, cultural significance.

Based on visual analysis I would gladly agree the similarity, as well as there being artistically etc better meanders in the library already aplenty. However if one would search Meanders of Upper Daugava, the name of the Tentative World Heritage Site, there are none.

So, I would think this is fixable. It is possible to teach AI easily to recognise uniqueness and relevance in the title, description, keywords, while the visual attributes indeed can be more challenging. Metadata should be easy, and my example would indicate it is not up to the task currently.

[Aleks_023]

Honestly, if it's really the way you're saying... that the system just auto-rejects based on metadata like titles and tags... then I really don't know what we're still doing here, trying to upload new work!?

There's already such a massive amount of content on the platform that it's nearly impossible not to run into similarity at some point, no matter how different the images actually are.
I even tried generating images on the same theme but in totally different styles... as referencing specific painters, using different drawing techniques... and still, I get hit with “similar content.” LOL

Feels like the library is getting saturated, especially with so many new users uploading aI-generated stuff in bulk.

I’m honestly wondering… is there even a point in continuing?
What do we do from here?

Aleks

[Aleks_023]

And now imagine the fact that everyone’s chasing the same topics Adobe promotes in their 2025 Creative Trends.

We’re all working around the same themes, so how much “uniqueness” can really exist anymore?

It’s no surprise the system sees everything as similar... lol.

Aleks

[Nancy OShea]

Look through Stock inventory for keywords that don't have millions & millions of competing assets.

It's better to be a big fish in a small pond than a tiny one that nobody will see.

Make a list of things that interest you.  Below are a few ideas to get you started. Be open to new ways of looking at things.  Experiment, find your own style. 

• Aging,
• Animals & wildlife,
• Beauty & cosmetics,
• Banking & finance,
• Computers & technology,
• Construction,
• Cultural diversity,
• Drone footage,
• Education & learning,
• Entertainment, dance,
• Environment & climate change,
• Fitness & exercise,
• Food, drink, nutrition,
• Holidays,
• Home & office,
• Martial arts,
• Medicine & medical technology,
• Mental health,
• Personal relationships,
• Poverty & homelessness,
• Real people in social settings,
• Science & math,
• Sports,
• Travel,
• Yoga, meditation, religion.

[Kristina987]

Yes, I've already noticed some improvement just by aiming to be more original and trying to come up with ideas I haven't seen before, even without doing any research.