Just returned from a trip to Japan and I still have cat on the brain. I know there are other gallery posts, but thought it would be fun to have a dedicated Dall-e 3 Cat Gallery.
How will we know if the image was generated using DALL·E 3?
Aren’t they including meta exif info or something now? Can that be spoofed?
We’ve been toying around with an idea for bi-weekly or monthly DALL-E theme threads…
While I was not expecting an exact answer at this time. If a means becomes possible to uniquely identify the image is from DALL·E by passing it to a verification process or similar than that would be nice.
I have no problems with the topic, just wanted to note that AFAIK there is no way to verify such. As the OP notes,
Yeah, I remembered something about it…
What is the C2PA standard and what does it enable?
C2PA is an open technical standard that allows publishers, companies, and others to embed metadata in media for verifying its origin and related information*.* C2PA isn’t just for AI generated images - the same standard is also being adopted by camera manufacturers, news organizations, and others to certify the source and history (or provenance) of media content.
What is OpenAI’s implementation of C2PA?
Images generated with ChatGPT on the web and our API serving the DALL·E 3 model, will now include C2PA metadata. This change will also roll out to all mobile users by February 12th. People can use sites like Content Credentials Verify to check if an image was generated by the underlying DALL·E 3 model through OpenAI’s tools. This should indicate the image was generated through our API or ChatGPT unless the metadata has been removed.
https://help.openai.com/en/articles/8912793-c2pa-in-dall-e-3
I’m not sure if it can be spoofed or not…
Picture of a cat eating the last ice cream on a hot summer day while all the other animals are watching jealously
Ps. The model really can’t count
Here is the grand and majestic oil painting of a king with a cat’s head, surrounded by its human followers. The scene captures the surreal yet powerful atmosphere of the cat-headed king in a royal palace, with its followers in awe. The painting emphasizes grandeur, with detailed royal robes, an ornate throne, and a luxurious palace setting.
A photo-realistic depiction of an epic cat warrior in full armor, standing in a heroic pose. The cat is anthropomorphic, with a muscular build and fierce, determined eyes. Its armor is intricately detailed, made of metal with a reflective, polished surface, adorned with mystical runes and emblems. The background is a grand battlefield at dawn, with rays of sunlight breaking through the clouds, highlighting the cat warrior. The scene is dynamic, with an air of impending action, and the overall atmosphere is intense and epic.
Tried verifying the image from this post and the result was
Tried the second image, not expecting success either, and this was the result
How I tested and generated the verification result.
- Using Google Chrome on Windows click on the image and drop on the Windows desktop to create a local copy of the image.
- Open
https://contentcredentials.org/verify
- Click on the image from the Windows desktop and drop into the verification site, e.g.
If someone can give a way to create an image that verifies successfully and note the steps here it would be appreciated.
Maybe OP didn’t use DALL-E? Hehe…
I tried mine…
ETA: @EricGT I tried the other one… I didn’t DRAG n DROP, though, so maybe that’s the issue?
Thanks.
I would not be surprised if the meta data is being removed or altered along the way and that is causing a problem. The image from the OP did contain more than just the image data but the viewer I used did not convert the bytes into a human readable form. So I don’t know if the needed meta data is not present by the time the image is posted here.
I tried your image and same result No Content Credential
.
I saw your update, so the needed metadata is there. Now to understand why what I did, did not work.
Can you list the details of how you used the OP image to get the valid credentials? OS, browser, user actions, etc.
No problem.
Yeah, it can be stripped out, so maybe that’s happening somewhere.
Weird.
[quote=“PaulBellow, post:12, topic:912628, full:true”]
Maybe OP didn’t use DALL-E? Hehe…
That would have be hilarious.
here’s mine.
original prompt:
a 3d claymation scene that shows a maneki neko infront of a traditional japanese wagashi shop in autumn
revised_prompt:
Generate a 3D claymation scene that depicts a Maneki Neko, the Japanese good luck charm, standing in front of a traditional Japanese Wagashi shop. The shop should be adorned with autumnal decorations. The ambiance is filled with falling leaves, and the surrounding scene embraces the rich colors of Autumn - robust oranges, deep reds, and golden yellows.
Where is the cat in the picture?
See the title of the topic.