MicrostockGroup Sponsors


Author Topic: Review of AI generated photos are fast now, but many rejection  (Read 7282 times)

0 Members and 2 Guests are viewing this topic.

« Reply #25 on: February 17, 2024, 02:32 »
0
There must be some secret sauce since my images get accepted much faster than a year ago when I started. I submit daily now, I think it helps.

There is a display on of some kind of ranking on the top left on your main screen, for me its 17,583 have no idea if its good or bad or how to measure it.


« Reply #26 on: February 17, 2024, 03:28 »
+1
That is your total rank. You can switch to a weekly rank, which gives you a better indication of how you are doing compared to everyone else.

You see it when you are logged in via tablet or computer. At least for me on my phone it is cut off.

« Reply #27 on: February 22, 2024, 09:54 »
+1
That is your total rank. You can switch to a weekly rank, which gives you a better indication of how you are doing compared to everyone else.

You see it when you are logged in via tablet or computer. At least for me on my phone it is cut off.
Thank you! This week, it shows 1,850 -  surprising that its so much better than 17,538 lifetime rank, yet , its a really slow week for me? So confusingso 1,850 number must not be a good weekly number?
I hope that most people  have better numbers!
« Last Edit: February 22, 2024, 10:02 by Mifornia »

« Reply #28 on: February 22, 2024, 10:33 »
0
I am having a very slow week as well. My current weekly rank is a frustrating 4310, so you are doing much better than me. Last week I was around 3200, which I still consider bad.

Monday was a public holiday in the US, perhaps people are taking a break? Although it is mostly Europe that is buying less.

I have many spring/easter images in good search positions, but the sales are really slow. Positions for easter are actually much better than christmas.

I hope customers get back into working seriously soon.

« Reply #29 on: February 22, 2024, 14:18 »
0
Cobalt, really appreciate your reply.
The rating must be according to the amount of images divided by amount of sales one has?
Or its strictly based on amount of sales?

MZP

« Reply #30 on: February 23, 2024, 02:44 »
+1
Mine is also frustratingly low, at 3,720 currently (way lower than my ATH of 977). Weekly ranking does not take revenue into account, but the number of sold assets in the respective time frame. I don't think it accounts for the number of assets in your portfolio, just the number of sales.

« Reply #31 on: February 23, 2024, 09:34 »
0
Weekly ranking does not take revenue into account, but the number of sold assets in the respective time frame. I don't think it accounts for the number of assets in your portfolio, just the number of sales.
So there is roughly only 4000 accounts that have more weekly downloads than us?
Thats concerning: Ive been working quite hard towards my goal of 200-300 downloads per week, (Im still not there, some weeks are 150) From reading various blogs, I thought its achievable, yet only 4000 other accounts in whole world have more uploads? And how many people contribute to Adobe? Must be millions? That probably puts us in 5-10% top percentage of contributors? (And some contributors been working for years) So there is only a small opportunity for growth, unless the bell curve is really steep ?

(Most of My best sellers are hand painted)
« Last Edit: February 23, 2024, 09:50 by Mifornia »

« Reply #32 on: February 23, 2024, 09:49 »
+1
Cobalt, really appreciate your reply.
The rating must be according to the amount of images divided by amount of sales one has?
Or its strictly based on amount of sales?

I have no idea. This is like asking how does the algorithm work to put my content on page one.

Every agency has their own secret sauce how to handle the review queue and I am sure it is adjusted all the time.

But all agencies want content that actually sells. So producers that upload stuff that sells should see their files moved up faster. It is business basics.

I am sure there is also a lifetime factor or something, probably also seasonal things, maybe now they prefer spring files to christmas etcand certainly they have hot content customers really want and probably place that on top of the queue. Or maybe individual editors that can look into what is coming in and move something up for a high value customer who needs it

And if they bring in more people and increase their team, everything will move faster.

But what you can do is try to make sure you upload what sells.

Obviously newbies need 1000 fresh files of varied content to see what works for them, but uploading 50 images of your dog looking left and right and up and down is not the way to go.

Quality and good editing help a lot.

Some of the most successful stock ports sometimes have very, very few files, sometimes less than 2000 or 4000 but they keep outselling everyone else.

If you look into those ports, literally every single image is a perfect bestseller.

My own strategy is not like that. I admire that very much, but I still upload a lot of midtail shots.

« Reply #33 on: February 23, 2024, 18:20 »
0
So there is roughly only 4000 accounts that have more weekly downloads than us?

That does not take into account what people are buying, if you are the only one with the images that the buyer has wanted then that makes you #1 ... its comparing apples to oranges. If that was 4000 buyers for those 4000 images then that makes everyone #1

« Reply #34 on: February 23, 2024, 21:04 »
0
So there is roughly only 4000 accounts that have more weekly downloads than us?

That does not take into account what people are buying, if you are the only one with the images that the buyer has wanted then that makes you #1 ... its comparing apples to oranges. If that was 4000 buyers for those 4000 images then that makes everyone #1
🤣🤣🤣 omg, yet another confusing algorithm!
 Nothing like no growth prospect to kill a motivation! Thank you for taking time to clarify.

« Reply #35 on: February 23, 2024, 21:29 »
+1
@cobalt
I wish I had your wisdom and a level head, but hope its not too late 🤣 Seems that Adobe doesnt give information easy, I like puzzles, hope to figure this one out as well.
From my limited experience, I can share that a couple of my images are on the first page of search with 5 million results. They sit there and sell poorly. My bestseller is on page 7 and outsells them 6 times over per month. It doesnt move up to better pages. How it is possible, I dont understand. It sells always for 0.37 so may be its from the link outside and Adobe doesnt count it? (but they fade to nothing in about 6 months, thats disappointing for passive income ) I also have a couple images in Featured searches and they also sell poorly compared to bestsellers.
I cant pin point what exactly will sell: I create 5 different images on a subject that I know will sell, but which image will take off - I cant predict. Many times my perfect favorite images didnt sell at all and I spent days painting them. I have only one AI image that became a bestseller (heavily edited) Im yet to repeat that again, but Im trying! Ah, and I make more $ now than when my account was featured on top sellers list.
Its all quite confusing. I would sure would love to see ports that keep on selling the same images for years and years. Its nice to know that they exist
« Last Edit: February 23, 2024, 21:36 by Mifornia »

wds

« Reply #36 on: February 23, 2024, 22:49 »
+3
@cobalt
I wish I had your wisdom and a level head, but hope its not too late 🤣 Seems that Adobe doesnt give information easy, I like puzzles, hope to figure this one out as well.
From my limited experience, I can share that a couple of my images are on the first page of search with 5 million results. They sit there and sell poorly. My bestseller is on page 7 and outsells them 6 times over per month. It doesnt move up to better pages. How it is possible, I dont understand. It sells always for 0.37 so may be its from the link outside and Adobe doesnt count it? (but they fade to nothing in about 6 months, thats disappointing for passive income ) I also have a couple images in Featured searches and they also sell poorly compared to bestsellers.
I cant pin point what exactly will sell: I create 5 different images on a subject that I know will sell, but which image will take off - I cant predict. Many times my perfect favorite images didnt sell at all and I spent days painting them. I have only one AI image that became a bestseller (heavily edited) Im yet to repeat that again, but Im trying! Ah, and I make more $ now than when my account was featured on top sellers list.
Its all quite confusing. I would sure would love to see ports that keep on selling the same images for years and years. Its nice to know that they exist

Could it be that perhaps the search terms you use when you see your images on page 'X' are not the search terms being used by buyers? Perhaps the majority of buyers are using different search terms when they find your images to buy?

« Reply #37 on: February 24, 2024, 03:02 »
+1
that would be my thinking, your bestsellers are probably on page one with another type of keywords.

there might also be regional factors, that your files took off in a specific country or different region to your own.

agencies tend to prefer to show local content to their buyers group because even simple things can look very different. A german apple cake looks very different to the typical US apple cake, different mix of ethnicities etc..

so perhaps your bestsellers are popular somewhere else.

the other thing is our files are also available through api deals. perhaps there is a designer place somewhere that is connected to adobestock and in their system your file has a very high rank?

i have some files that sell consistently for years but when I try to expand the subject ranke, I dont get more sales. even if the new files are better imo.


most important: you have sales. :)

Uncle Pete

  • Great Place by a Great Lake - My Home Port
« Reply #38 on: February 24, 2024, 13:28 »
+1
So there is roughly only 4000 accounts that have more weekly downloads than us?

Where is @WILM he understands this and was tracking the rank, long ago.

In 2020 an analysis showed that over 60% of all active portfolios had ten or fewer images. Based on that, I'm already well above the 50% mark.  :)

Predicted Lifetime Downloads / Position

Old data 2019
8,000/4,700
5,000/8,500
1,400/28,000
1,200/32,000
450    /67,000
200    /120,000

From others ranks 2022
40,000/2,230
11,660/7,440
8,030/10,000
2,000/25,000

4,000 accounts with more downloads could be they have thousands more or one more. Next week, your position will change again. And I'm not sure that the positions are by numbers. In fact, I'm fairly certain they are by a range. If someone is position 4,000 that means they could be position 4,000 to 5,000 as a group. It's not precise. Look next week, did you jump down to 6,000 with almost the same number of downloads? Or did you move up to 3,500 with a similar number as the previous week?

Finding your own images, I'd say, start with trying to determine from other sites, what words were used by a buyer, that purchased a license. Then test those on Adobe. I think SS, DT and Alamy will help you with that. Then search for your own, using at least three words. No buyer with any hope or sense would use one or two words only, and I suspect most use a minimum of three words.

If any site has a reasonable search, it's going to include "diversity", so if you have 50 uploads of some subject, the buyer won't see 50 of yours in a row, when they match the keywords. Whether random or based on sales or location, the best searches will show one of yours, then many others, before one of yours will show again. No buyer wants a page of the same variations of the same subject from the same artist.

That's why I personally say, upload the best 4 or 5 and move on. If those start to work, add 4 or 5 more, later. True, we don't know which one is "best" and what buyers will download, but we can generally see for our self, which images are best. For my own, I often upload one and maybe two, of any concept or idea. That's it. Sending 20 or 40 of the same shot does nothing, and just in case, the buyer might see 20 of the same and tend to think "that's common" or ordinary, instead of, that's a nice unusual shot?

Shotgun and hope a little pellet will fly in the right direction, and hit something, or sharpshooter with a accurate single bullet, and target the bullseye?  8)

« Reply #39 on: February 25, 2024, 15:16 »
+1
In fact, Adobe has managed to speed up the review to a maximum of 20 days.
The rejection rate is directly related to the acceptance rate (actually a no brainer).

I am sure that Adobe has an AI automated process as a first step. Every bigger artifact will lead to a rejection.
The second step will include human, manual reviewers. Here every reviewer will absolutely certain look at the acceptance rate first. If this is high, he will be much less strict in order to save time.

My acceptance rate is currently over 99%. The last 800 images have all been accepted since I switched to motives, which create less artifacts.

Uncle Pete

  • Great Place by a Great Lake - My Home Port
« Reply #40 on: February 26, 2024, 14:19 »
0
In fact, Adobe has managed to speed up the review to a maximum of 20 days.
The rejection rate is directly related to the acceptance rate (actually a no brainer).

I am sure that Adobe has an AI automated process as a first step. Every bigger artifact will lead to a rejection.
The second step will include human, manual reviewers. Here every reviewer will absolutely certain look at the acceptance rate first. If this is high, he will be much less strict in order to save time.

My acceptance rate is currently over 99%. The last 800 images have all been accepted since I switched to motives, which create less artifacts.

My acceptance rate, affects my acceptance rate? Sounds like a death spiral. I don't know where to see my acceptance rate. I'm not sure I want to?  :)

Maybe this is useful also?  https://stock.adobe.com/pages/artisthub/pdf/generative-ai-guide.pdf

10 tips for getting your generative AI images approved for sale on Adobe Stock.

Yes, I'd be surprised if any agency now, didn't have a submission review for standards, the intake of files, before a person sees them. Size, file type, and could easily be some checks on obvious quality issues. Now, it could be more sophisticated like focus. But still Adobe or SS, the software used in the review process, is to assist the person doing the review.

« Reply #41 on: February 27, 2024, 10:30 »
+2
Finally my recent uploads were approved today after 20 days or so.  I think this is the new norm, 20 days.  It's better than 30 days or so last year.  But new year fast review is over.  Also, the approved photos were from a batch of 9 photos and all were approved.  I upscaled 2x using Midjourney upscale.  I found 4x upscale on Midjourney makes photos look artificial.  So, 2x is the sweet spot imo on Midjourney upscale function.
« Last Edit: February 27, 2024, 10:33 by blvdone »

« Reply #42 on: February 29, 2024, 09:48 »
+1
My acceptance rate, affects my acceptance rate? Sounds like a death spiral.

Yep it's a death spiral. I am quite sure if you have a high rejection rate, you will get many rejection rates because the reviewer will be likely much stricter. Now when I managed to get less rejections most files are accepted even if I would spot myself some AI generation errors.
What I did is to shift to more simple images and other Stable Diffusion models, which create less image noise and artefacts.
You should first try to decrease your rejection rate by submitting much less complex images and then start to submit your favored images. Chances will be more that they then will be accepted.

What Adobe Stock does not examine are AI generation errors like false hands, nostrils, ears, etc. They just seem to focus on technical errors like image noise, etc.

« Reply #43 on: March 03, 2024, 10:05 »
0
Lately I have 100% approval, but my AI photos hit the road block, sitting there for more than a month already

« Reply #44 on: March 04, 2024, 04:44 »
0
Anyone else's review times are screwed up now? Today, in my to-be-reviewed tab, my oldest illustration now say "submitted over a year ago" and my oldest photo "submitted over 2 months ago".

MZP

« Reply #45 on: March 04, 2024, 06:36 »
0
Apparently some batches get into a black hole or somethnig, and never get evaluated. If a submission is older than 2 month, you better delete and resubmit those assets. At least that's what the guys from Adobe recommend on Discord. I also had few images that were 2 months old, but I just deleted them, without resubmitting, because I didn't like them anymore

bpawesome

  • 3D artist & Full Stack Developer

« Reply #46 on: March 04, 2024, 06:53 »
0
I don't know if it's related,
but, I'm getting quick reviews now that I'm submitting batches of 10 images (1-2 days for the last 10 batches)

There are few images that are stuck in review, but they are maybe problematic and the reviewer didn't know how to handle them (somewhat news or current world events related AI images) not rejected but are stuck in review for a month or so...

« Reply #47 on: March 04, 2024, 08:01 »
0
Apparently some batches get into a black hole or somethnig, and never get evaluated. If a submission is older than 2 month, you better delete and resubmit those assets. At least that's what the guys from Adobe recommend on Discord. I also had few images that were 2 months old, but I just deleted them, without resubmitting, because I didn't like them anymore

Thank you, but that wasn't what I meant. I mean, I have files I submitted 3 weeks ago, but suddenly they show as "submitted over a month ago" and files I submitted 2 months ago that show as "submitted over a year ago". The times are worng.

MZP

« Reply #48 on: March 04, 2024, 08:48 »
0
Ah, got it now. In my case they show correctly. If it's a glitch, maybe it will go away soon.

« Reply #49 on: March 04, 2024, 08:55 »
+1
So, it's just me? Now that's worrying me even more.
I just double checked, actually everything I submitted more than 5 days ago is displayed incorrectly, with a month to a year added to the time.  :o


 

Related Topics

  Subject / Started by Replies Last post
2 Replies
3215 Views
Last post December 21, 2006, 07:38
by melking
4 Replies
4075 Views
Last post May 27, 2012, 14:28
by sam100
2 Replies
2618 Views
Last post May 02, 2014, 04:08
by ruigsantos
31 Replies
16235 Views
Last post October 23, 2023, 20:03
by PigsInSpace
4 Replies
1623 Views
Last post February 26, 2024, 13:47
by cascoly

Sponsors

Mega Bundle of 5,900+ Professional Lightroom Presets

Microstock Poll Results

Sponsors