MicrostockGroup Sponsors


Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - Jo Ann Snover

Pages: 1 2 3 4 5 [6] 7 8 9 10 11 ... 291
126
Adobe Stock / Re: Adobe Stock 1099 forms are available
« on: February 02, 2024, 17:39 »
One request, one hint for other contributors:

It would be really good if the link to obtain the 1099 were somewhere in the contributor interface - I looked for it after I saw Mat's post here, but couldn't find one. That interface should be contributor central (and I haven't seen any email about the 1099 either; I did check my spam folder)

If you have a 4-digit contributor number - I do as I joined Fotolia in 2005 - you have to put in a zero before it or the number won't be accepted. I had to figure that out for myself last year (never got a reply from the tax help query) and possibly I'm not the only fossil with such a small number.

127
New Sites - General / Re: Imago-images
« on: January 19, 2024, 10:10 »
...Never realized pond5 also distributes my photos. Should send them more images as well.

I did a catchup upload of most of my portfolio to Pond5 in 2020 (as part of a Stock Coalition thing where Pond5 had specifically asked for more photos to build up their collection). I had a small existing portfolio of some PSD templates and other  "specialty" items that occasionally sold (things other sites didn't handle plus you could set your own prices).

My experience has been that Pond5 doesn't sell photos and so I no longer keep the portfolio up to date. Sales are less frequent than at Dreamstime (which is saying something!).

128
Not sure how long there have been Premium AI images - $249.99 for large, $119.99 for small.Why on earth would a buyer pay that much for AI generated stuff? None of the typical reasons for the premium collection apply to genAI images. Nothing at all wrong with this image BTW, other than the price


129
The same here.  A lot of mass rejection of AI generated photos recently.

I think they just put on their "see quality issues everywhere" glasses for all types of content recently. I had a batch of 7 photos rejected yesterday for quality issues - it only took them about 5 days to reject them though :)

The review system is completely unpredictable

130
I keep track of some stats on Adobe Stock (via Google Sheets) - I started when I noticed the ratio of custom to subs sales was changing a while back. I have seen a big difference in the amounts of royalties over the last month or so - I assume that for whatever reason, customers aren't getting the big discounts they were able to get at various times last year.

January is showing an RPD for custom licenses of $0.99 and subs $0.81 - compared to $0.74 and $0.71 in November. Compare that to March 2023 when custom was $0.58 and subs $0.73

I also keep track of the different royalty "bands" - $1-$2, .90-.99, over $3, etc. - to see if the distribution of sales is changing. I started that when the $0.33 royalties first appeared (and happily they're never more than one or two a month and I've never seen anything lower; I have a minimum royalty amount each month so I can catch if the bottom drops out).

In November, the ratio of sales .90 - .99- to $1-2 was just under 39 to 1 - 39 times as many sales in the $0.9x bracket.

In January, so far, the ratio is just about one to one - very slightly fewer $0.9x sales. That's a very big swing. December was in between - about four $0.9x sales for every $1-2

And this isn't just a January thing. In Jan 2023, RPD was $0.77 for custom and $.086 for subs and the $0.9x sales were about twice the $1-2

It's good to see the RPD numbers coming back up - March 2023 it was $0.58 and $0.73 for subs (and there were zero $1-2 royalties that month!)

131
The girl with the teddy bear is one of 42 similar genAI images you get with a search for girl teddy bear ukraine - sad children in a war zone with a teddy bear.

This clearly isn't in Ukraine or in a war zone and it's the keywords that lead you to it not the title. One of the (unanswered) questions I asked in the thread about the new policy was whether keywords counted - which I think they clearly should. Keywords also determine what shows up for a buyer when they search.

child teddy bear war results in 632 genAI images; child gaza 938 images, gaza 4,323, palestine 10,069. These aren't for tourism promotion...

Not only has nothing been removed since the "change of policy" but the number of pseudo-editorial images has grown - one thousand more gaza images in 6 days, for example.

A rule without any enforcement is just a CYA maneuver.

132
Adobe Stock / Re: Similars policy
« on: November 29, 2023, 16:18 »
There are nearly 300 autumn leaves pictures in this contributor's genAI portfolio - all perfectly pleasant, but lots of repetitive material, not to mention all the other similar non genAI images already in the collection.

Earlier I looked at over 50 near-identical champagne flutes New Year's images recently approved - one contributor, but it was genAI. Again, perfectly pleasant, but all well covered and no WOW

The rules are reasonable; the problem is that they aren't applied even-handedly.







133
No answers to any of the questions, but in monitoring new acceptances (gaza, hamas, israel war, palestine) the pseudo-editorial genAI collection continues to grow. Terms supposedly not allowed are in the titles and in the keywords.

Rules mean nothing if they're ignored with no consequences.

This Associated Press article headline says it all (this is not about images sourced from Adobe Stock, but at some point something similar will happen given what continues to be accepted)

"Fake babies, real horror: Deepfakes from the Gaza war increase fears about AIs power to mislead"

https://apnews.com/article/artificial-intelligence-hamas-israel-misinformation-ai-gaza-a1bb303b637ffbbb9cbc3aa1e000db47

134
Not surprised to see more coverage - this time in The Washington Post (paywall) - of the masses of pseudo editorial genAI images on Adobe Stock.

"These look like prizewinning photos. Theyre AI fakes."

https://www.washingtonpost.com/technology/2023/11/23/stock-photos-ai-images-controversy/

The article raises many of the issues talked about here, and also points out, after noting Adobe Stock's change of policy and their blog post "As of Wednesday, however, thousands of AI-generated images remained on its site, including some still without labels."

It also appears that Adobe's change of policy came about after the Washington post and other publications contacted Adobe about all these pseudo-editorial images: "Adobe initially said that it has policies in place to clearly label such images as AI-generated and that the images were meant to be used only as conceptual illustrations, not passed off as photojournalism. After The Post and other publications flagged examples to the contrary, the company rolled out tougher policies Tuesday."

I did a few searches just now, and not only has nothing yet been removed, but there are new acceptances that weren't there a day or two ago



It's fine to state a commitment to fighting misinformation, but there needs to be action to follow up for it to mean anything:

"Adobe is committed to fighting misinformation, said Kevin Fu, a company spokesperson. "

Whereever the Washington Post used a photo from Adobe Stock's genAI collection they have slapped a big red banner saying "AI-GENERATED FAKE PHOTO" over it:



They also noted that some results appeared to be AI generated but were not labeled as such, although the example they link to has an image number (281267515) that is way too low to be genAI. Those start with 530+million ... or thereabouts:

"Several of the top results appeared to be AI-generated images that were not labeled as such, in apparent violation of the companys guidelines. They included a series of images depicting young children, scared and alone, carrying their belongings as they fled the smoking ruins of an urban neighborhood."

They also mention other categories such as Maui wildfires and Black Lives Matter Protests:

"It isnt just the Israel-Gaza war thats inspiring AI-concocted stock images of current events. A search for Ukraine war on Adobe Stock turned up more than 15,000 fake images of the conflict, including one of a small girl clutching a teddy bear against a backdrop of military vehicles and rubble. Hundreds of AI images depict people at Black Lives Matter protests that never happened. Among the dozens of machine-made images of the Maui wildfires, several look strikingly similar to ones taken by photojournalists."

I cannot fathom why Adobe Stock would wade into such a mess; the money made cannot be worth the risk of damage.

135
Very glad to see that Adobe plans to tag all genAI content in the collection - any timetable for that and will that include retroactively adding the credentials to the 26+ million images already there?

Also glad to see the note that "...we are committed to making it easier for people to identify which images are generative AI before licensing them on Adobe Stock.", but the screenshot included is no different from the current display. Is there something new to be done? If so, will it be an overlay as there is for Editorial images? And when will this be done.

I did a search for Gaza just now and it doesn't appear that any of the pseudo-editorial genAI images have been removed. Will the existing items that no longer comply with the Nov 21 submission rules be removed?

It's not just titles of pseudo-editorial images that are a problem "Updating our submission policies to prohibit contributors from submitting generative AI content with titles that imply that it is depicting an actual newsworthy event. " Lots of content that has a general-sounding title has keywords that have the image appear in a search for gaza, hamas israel war, etc. Customers may not realize that you do not offer editorial content (outside of the illustrative editorial) and just do a search. I don't think reviewers monitor keywords (given the massive amount of spam) but this new rule will be completely ineffective if keywords allow the unscrupulous to just avoid detection by keeping the title clean.

136
That is bad news - I wasn't sure if it could make a difference because of the various powerful interests who want to just take whatever they find useful, but with that kind of visible damage to an image it's useless for anything but social media stuff.

Thanks very much for taking the time to explore this and report on what you observed.

137
I do, occasionally, add to keywords if I realize I forgot something important. Being found is one of the important elements in licensing your work, and regardless of order, if you don't have the keyword, your image won't show up at all.

Sometimes I realize I need an additional term for the same thing (this doesn't apply for iStock/Getty where everything is reduced to the controlled vocabulary anyway). As an example, years ago I was putting remodeling in for all the home remodeling images and not, initially, renovation when that applied to most of the images as well.

Trying to game the secret sauce of search order seems not to make sense when we have virtually zero data to work with. I do put the important keywords first as that's never wrong on any of the sites.

138
At the beginning of October, Adobe Stock was adding around 900k genAI images per week.

From October 16th for the next three weeks, it was just over 1 million each week.

From Nov 6-13, it was almost 2 million!! 1,930,975.

I very much doubt that Adobe Stock is adding buyers at anything like the rate that it's adding genAI images. Or that existing buyers are suddenly going to be buying a significantly larger number of items. I doubt the AI gold-rush enthusiasts have thought too much about where this is all going, but even if this was all novel content I think there'd be a supply & demand imbalance.

Freepik's collection shrank between last Monday & today - from 30.08m down to 29.63m. It had grown from 28.7 to 30.08 the previous week, so I assume they were doing some cleanup?

139
Regarding AI images. I guess they may represent say 10% of the entire collection at Adobe numerically? However, I would venture to guess that "subjectwise" they could potentially be more of a threat than that 10% number suggests since in the AI world, having complex images with "models" and "sets" (e.g. business office, medical office) is a lot easier vs. "real world" photos where the photog would have to get actual models and actual "sets" or locations like offices etc.. Therefore I would expect the "10%" number to be misleading (no "snapshot" images with very low sales potential). In other words, there should be more of a sales impact than 10% might suggest. Having said all that, I haven't seen the impact sales-wise that this scenario would suggest.

I am made an awful lot of assumptions in my statements above. Does anyone care to comment?

I don't have any details about sales beyond sorting the genAI collection by downloads which would suggest that most of the the types of images it could uniquely do are not what contributors are uploading.

I looked at today's new genAI approvals and there are a bunch of rustic wood textures. The contributor has about 100 of those. What's the effing point? Same goes for views of fake-y nature scenes, food photos, abstract backgrounds, birds, puppies, kittens, etc. Even if it is a ton less work to set up food shots via AI, the collections already have tons of hot dog and hamburger pictures, so why add more?

One of the reasons I think the genAI content has stuck with the things already so well supplied is that many of the detailed settings that would work really well as stock and aren't already "full" are very hard for current genAI to get right - flawed details in factories, wind turbines, plane interiors, etc make these functionally useless to buyers.

140
For those who haven't been following the collection sizes, Adobe Stock's genAI collection is almost at 23 million; Freepik's is over 30 million; Dreamstime's is a hair over 7 million.

Shutterstock's is such dreck it doesn't matter, but it's approaching 1.5 million

The "oops" images continue to be everywhere...



 

141
https://petapixel.com/2023/11/05/genai-and-the-forced-evolution-of-photography-from-artifice-to-authenticity/

I like the notion that "Photography is more than just an imageits our connection to the real world, to one another, and to moments that inform our decisions."

His twitter feed references this article

https://futurism.com/adobe-caught-selling-ai-generated-images-israel-palestine-violence

I think the pseudo-editorial images have to be removed - like the 9/11 images that were allowed in and then removed a few months back.

The Futurism article doesn't mention Freepik, but their collection of over 30 million genAI images includes over 40k for a search for "gaza"

https://www.freepik.com/search?format=search&last_filter=query&last_value=gaza&query=gaza&type=ai

And even Shutterstock's sad collection of DALL-E creations includes a couple (along with their standard note "Important information - This content was generated by an Artificial Intelligence (AI) system. Shutterstock does not review AI-generated content for compliance with Shutterstocks content compliance standards. AI-generated"):

https://www.shutterstock.com/image-generated/gaza-city-palestine-2-children-playing-2384676855
https://www.shutterstock.com/image-generated/outdoor-photo-gaza-city-palestine-2-2384676351

Edited Nov 8 to add links to stories about the use of the fake Gaza images

https://www.crikey.com.au/2023/11/01/israel-gaza-adobe-artificial-intelligence-images-fake-news/
https://venturebeat.com/ai/adobe-responds-to-controversy-over-ai-generated-images-of-gaza-explosion/
https://www.vice.com/en/article/3akj3k/adobe-is-selling-fake-ai-generated-images-of-violence-in-gaza-and-israel
https://petapixel.com/2023/11/07/adobe-stock-is-selling-ai-generated-images-of-the-israel-hamas-conflict/

IMO the Adobe statement that they are working on the adoption of Content Credentials is an irrelevant distraction - I don't believe downloadable images had content credentials added, so how can any viewer of the pseudo-editorial stock image have any idea if it's authentic or not?

Adobe Stock is a marketplace that requires all generative AI content to be labeled as such when submitted for licensing. These specific images were labeled as generative AI when they were both submitted and made available for license in line with these requirements. We believe its important for customers to know what Adobe Stock images were created using generative AI tools.

Adobe is committed to fighting misinformation, and via the Content Authenticity Initiative, we are working with publishers, camera manufacturers and other stakeholders to advance the adoption of Content Credentials, including in our own products. Content Credentials allows people to see vital context about how a piece of digital content was captured, created or edited including whether AI tools were used in the creation or editing of the digital content.


Seems like a no-brainer (to me) for Adobe Stock to add content credentials to all genAI images they host. They already strip and alter metadata in what we upload, so it wouldn't be groundbreaking in any way. But I also cannot see anything good coming out of accepting pseudo-editorial images of an ongoing war.

142
"The biggest companies in AI arent interested in paying to use copyrighted material as training data, and here are their reasons why."

https://www.theverge.com/2023/11/4/23946353/generative-ai-copyright-training-data-openai-microsoft-google-meta-stabilityai

"Meta: Copyright holders wouldnt get much money anyway...
Google: AI training is just like reading a book...
Microsoft: Changing copyright law could hurt small AI developers...
Anthropic: Current law is fine; dont change it...
Adobe: Its fair use, like when Accolade copied Segas code...
Anthropic: Copying is just an intermediate step ...
Andreessen Horowitz: Investors have spent billions and billions...
Hugging Face: Training on copyrighted material is fair use...
StabilityAI: Other countries call AI model training fair use...
Apple: Let us copyright our AI-made code"

Let me paraphrase: Because I have to spend huge sums on developing the code and any one of you wouldn't make much anyway, I get to take your stuff for free to train what I'm building. The last part from Apple would add: and all the profits from anything I make is mine to keep.

This is such self-serving B-S-

143
Artists may poison AI models before Copyright Office can issue guidance

https://arstechnica.com/tech-policy/2023/11/artists-may-poison-ai-models-before-copyright-office-can-issue-guidance/

(emphasis mine)
" ...the Federal Trade Commission may have the power to intervene. Andrew Burt, a founder of an AI-focused law firm called BNH.ai, told TechCrunch that the FTC is "already pursuing" what the FTC calls algorithmic disgorgementwhich is where the FTC "forces tech firms to kill problematic algorithms along with any ill-gotten data that they used to train them." It seems possible then that, should artists suing AI makers win or should the Copyright Office provide such a recommendation, enforcers could one day order AI image makers to retrain models using only permitted, licensed data."

I'm not holding my breath, but it's an interesting idea

Separately from copyright law issues (which are a real tangle in an international arena), my thinking about training is that without creators' images generative AI image could not exist. They are wholly dependent on having stuff to ingest and it's stuff they didn't create or pay for. Some reasonable accommodation needs to be worked out (reasonable for both sides, not just whatever the large tech think they have the economic heft to grab).

144
There are one or two additional pieces of gear you'll need, but this video shows how to record video from an iPhone 15 Pro :)

https://youtu.be/V3dbG9pAi8I

145
... have you any thoughts why the software wouldn't pick up on the other images that are very similar in my collection?

I have no information about what the software's rules are, but I did a little experimenting with a series of mine. What I noticed is the following:

1. The series shown is not dependent on the search terms you use for an image - a given image has its "series" (at least the 5 thumbnails shown) however it is found

2. Looking at the first 10 keywords for the images I experimented with, all of them had some keywords in common but not in the same order. I put the most significant terms for a particular image up front and the groupings suggest to me that keyword order, not just presence or absence, make a difference.

3. Clicking on "See More" gives me varying numbers of items in the series, again suggesting the keywords are being used to collect items. Items that are not part of the series (as I see it or intended it) show up for some images but the content is related-ish

I think keywords are the area for you to focus on - making sure you're keeping them accurate to the image content, but ordering them consistently across the series

146
https://petapixel.com/2023/10/29/ai-and-me-how-image-generation-is-changing-my-role-as-a-photographer/

One photographer's perspective on use of genAI images, including composites where the "set" is genAI and the studio is used to shoot the model who will be photoshopped into the set.

147
I do have an iPhone but have not tried to write to an external flash drive - so the short answer is, I don't know.

However, both those links you provided say they support Android phones; nothing at all about iPhones (and I assume they'd have said something if they were supported)

This is what Apple has to say - and it doesn't say anything at all about whether you could use the camera app with the right hardware connection to write to the external storage:

https://support.apple.com/guide/iphone/external-storage-devices-iph95baac91f/17.0/ios/17.0

There is a video about this (I didn't watch it) that might tell you how that guy did it

https://www.youtube.com/watch?v=Pr3DqO-NOG8

148
Adobe Stock / Re: This is highly unprofessional
« on: October 27, 2023, 06:52 »
Maybe the dog ate Adobe Stock's homework, but I don't think that's why searching for Berlin - spelled the same in English and German - gets two different totals in two different regions. It's a bug - as are all the other examples in all likelihood.

Fotolia led all the other agencies in handling local sites where you could do business in the local language with local currency. The other agencies followed after they saw what a success that was. Having established that with both contributors and customers, it's expected that this will continue over time. As a buyer in Germany, you're not going to switch regions to see if there are more files available - you'll just assume that what you're shown is all there is.

And these bugs aren't new. I think when tailoring search results to promote "local" content was introduced a while back people started reporting problems with content not showing up (versus just different sort order preferences).

Adobe is a huge company and Adobe Stock is a very small cog in a very large system. Sometimes it doesn't get attention quickly/at all.

149
iStockPhoto.com / Re: August 2023 statements are in early
« on: October 27, 2023, 06:43 »
Question for those whole upload to Istock regularly :  do new files sell at all?  or is the largest part of your Istock income based on old(er) files?

After a (years-long) break I started uploading to iStock again earlier this year. So some of those "new" files weren't literally new, but were new to iStock, and they are selling. My portfolio is small - about 1.8k - but making payout nearly every month.

150
Shutterstock will be announcing their Q3 results next Tuesday (31 Oct) so perhaps it's not surprising that they are banging the drum about new AI features in the Shutterstock library:

https://www.prnewswire.com/news-releases/shutterstock-integrates-creative-ai-into-library-of-700m-images-to-offer-first-ever-marketplace-of-fully-customizable-stock-301967961.html

A couple of days ago, one analyst report said the results weren't expected to be good:

https://finance.yahoo.com/news/analysts-estimate-shutterstock-sstk-report-140042530.html

"Wall Street expects a year-over-year decline in earnings on higher revenues when Shutterstock (SSTK) reports results for the quarter ended September 2023. While this widely-known consensus outlook is important in gauging the company's earnings picture, a powerful factor that could impact its near-term stock price is how the actual results compare to these estimates."

Blathering on about "infinite options", the CEO opined:

"This is an unprecedented offering in the stock photography industry," said Paul Hennessy, Chief Executive Officer for Shutterstock. "Now, creatives have everything they need to craft the perfect content for any project with AI-powered design capabilities that you can use to edit stock images within Shutterstock's library, presenting infinite possibilities to make stock your own."

It's in beta, but considering the abysmal quality of the Dall-E output that is currently in their AI generated section, it'd have to be much better to actually be useful to customers

Babies   -   Typewriter  -  Podcast illo

They will give a demo 9 Nov

Edited mid-day Friday to note that SSTK (Shutterstock's stock) is around $34.20 which is what it was at the beginning of June 2020 - when Stan Pavlovsky pursued his plan to boost the stock price by looting contributor royalties. Or "margin optimization" as they referred to it.

The stock ended 2020 around $70 and reached as high as $123 in Nov 2021. Except for a brief lurch up to $70 at the beginning of this year, it's been downhill since.

Pages: 1 2 3 4 5 [6] 7 8 9 10 11 ... 291

Sponsors

Mega Bundle of 5,900+ Professional Lightroom Presets

Microstock Poll Results

Sponsors