Spaces:
Running
3 Reasons Why "Too Much Traffic" Message Keeps Occurring
I've seen plenty of community posts within the past few hours regarding this subject, and so I wanted to contribute my own to set everyone straight so they don't keep asking the same question 81 times. (Link to DALL-E 2 at the bottom of the post)
REASON #1:
A few hours ago (as of writing this post), popular youtuber Markiplier posted a video titled "DALL-E Mini: A Meme Nightmare".
In this video he featured this specific project by creating a bunch of questionable art using the Dall-E tool, as well as asking his twitter followers to send him pictures of their generated art as well (More on that in Reason #2).
I myself am an avid Markiplier viewer, so I think I have a right to list this as the #1 reason (as of this thread being posted).
REASON #2:
As mentioned previously, Markiplier tweeted out asking his followers to send him their generated images. This tweet, and other tweets from various creators, understandably got a lot of traction, so of course a lot of twitter users seeing this new AI generation tool are going to come here to check it out.
REASON #3:
I haven't found any evidence to prove this is true, mostly a lot of wild accusations and baseless claims (which is why this is at the bottom), but allegedly a group of 4chan users (go figure) set up some bot accounts (go. figure.) to auto-generate mass amounts of images in order to purposefully slow down the website.
Now let me be very clear, I have NO evidence to support this being true, it may very well be a hoax.
But knowing 4chan, it's also very easily true.
SUMMARY:
Due to the massive popularity that this project has gained recently, way more people than was intended by the developer are now flooding this website in order to get their hands on a cool Darkiplier image. And of course, seeing as this is a (relatively) small project, obviously the servers that host this aren't going to be able to handle 150,000 people all making multiple requests at once. Go figure.
So in order to keep you all stimulated for the time being while this is all being worked out, here is Dall-e 2: https://openai.com/dall-e-2/
I suggest marveling at the realism and beauty of these images (particularly image #7 of "Teddy bears shopping for groceries in ancient Egypt) while these kinks get worked out, and perhaps also trying out some other AI generated image tools. There are plenty of those on the internet, not JUST this one.
If anyone has any other reasons for the site being slow, please reply to this post with those reasons, as these were just the first 3 that came to my mind.
Have a great day everyone.
EDIT:
I forgot to mention this in my original post, but I've seen a lot of people in these forums suggesting that all you need to do to get your image generated during this time of high traffic is to just keep hitting the "run" button until your request is processed...
Don't.
Just don't do that. That doesn't even make any sense. Spamming the servers with more requests is just going to keep adding to the problem. I don't even know why I need to explain this.
Too many requests > "Too much traffic" > Keep hitting "run" > Adds even more requests to the server > Requests keep adding up > Too many requests > etc.
Just leave the site alone for a bit and go try out some other alternatives. There are plenty of AI image generation tools out there, this isn't the only one that exists. A quick google search of "AI image generator" brings you 1.37 Billion results. Go through some of those.
Have a great day everyone.
unfortunately, the best image generator from text is ONLY this one: dallemini
unfortunately, the best image generator from text is ONLY this one: dallemini
That isn't necessarily true. There are others:
https://deepai.org/machine-learning-model/text2img
https://hotpot.ai/art-maker
https://creator.nightcafe.studio/
https://deepdreamgenerator.com/
https://bruzu.com/demos/ai-image-generator-from-text/
just to name a few
I haven't had the opportunity to test dalle-mini yet due to the traffic issue. However, I hope it works better than the links above.
For instance, bruzu merely does a google image search with the specified term and pastes the result untouched (I just tried with "giraffe" and reversed search the image on google and found the exact same one ; the result of bruzu is https://imgur.com/a/veTAsTp ; the original image is https://unsplash.com/photos/QvMRSTkc518 ) -- I wouldn't even call that AI.
Text2img returns results that have nothing to do with the input query, even for a simple query (I asked "a kid eating an ice cream" and got a bird on a lake ; I merely asked "a giraffe" and got a butterfly : https://imgur.com/a/RpPAhCi !...).
Nightcafe gives unreasonable results for simple requests too (for "a bird on a lake", you can barely recognize anything: https://creator.nightcafe.studio/creation/JWa3q7l6oFv3fYfxcmag ).
Deepdreamgenerator doesn't seem to do text-to-images queries.
Art-maker is slightly better, but still not quite good for simple queries ("a giraffe on a car" : https://hotpot.ai/s/art-maker/8-QuUxyCKna4ToOSl or simpler, "a kid eating an ice cream" https://hotpot.ai/s/art-maker/8-tvs3VTBNNWrC5ZS ; "a bird on a lake" https://hotpot.ai/s/art-maker/8-WYGGN0rVTl2M7rb )
I haven't had the opportunity to test dalle-mini yet due to the traffic issue. However, I hope it works better than the links above.
For instance, bruzu merely does a google image search with the specified term and pastes the result untouched (I just tried with "giraffe" and reversed search the image on google and found the exact same one ; the result of bruzu is https://imgur.com/a/veTAsTp ; the original image is https://unsplash.com/photos/QvMRSTkc518 ) -- I wouldn't even call that AI.
Text2img returns results that have nothing to do with the input query, even for a simple query (I asked "a kid eating an ice cream" and got a bird on a lake ; I merely asked "a giraffe" and got a butterfly : https://imgur.com/a/RpPAhCi !...).
Nightcafe gives unreasonable results for simple requests too (for "a bird on a lake", you can barely recognize anything: https://creator.nightcafe.studio/creation/JWa3q7l6oFv3fYfxcmag ).
Deepdreamgenerator doesn't seem to do text-to-images queries.
Art-maker is slightly better, but still not quite good for simple queries ("a giraffe on a car" : https://hotpot.ai/s/art-maker/8-QuUxyCKna4ToOSl or simpler, "a kid eating an ice cream" https://hotpot.ai/s/art-maker/8-tvs3VTBNNWrC5ZS ; "a bird on a lake" https://hotpot.ai/s/art-maker/8-WYGGN0rVTl2M7rb )
I appreciate you taking into account my response as opposed to just ignoring it.
Yes you are correct. Unfortunately AI-driven image generation is very difficult and cannot be achieved easily through conventional means.
Dall-E seems to be the only one capable enough to produce semi-accurate images under the right circumstances.
I also do hope that this server issue becomes fixed, or perhaps the creator develops an app that can render these images on your computer rather than through a server request. Lets just hope that this all gets fixed soon, and look forward to Dall-E 2. Sorry I couldn't provide any better alternatives, I tried my best.
tldr
reddit and 4chan been soamming this garbage ai for days now, youtube ecelebs have nothing to do with it
and its sad zoomers cant just wait
literally justnwait
imagine being filtered by worldgen
lmao
casuls
kys
and its sad zoomers cant just wait
literally justnwait
imagine being filtered by worldgen
lmao
casuls
kys
Above we can see a 4chan goblin trying to communicate with the outside world
"grr, muh evil 4chan doing evil 4chan things for no reason because they're just so evil and edgy 馃が馃が"
and its sad zoomers cant just wait
literally justnwait
imagine being filtered by worldgen
lmao
casuls
kysAbove we can see a 4chan goblin trying to communicate with the outside world
"grr, muh evil 4chan doing evil 4chan things for no reason because they're just so evil and edgy 馃が馃が"
bro I'm so glad we have REAL comedians in this world to help make these forums a better place
Has the team behind DALL-E Mini tried to share a downloadable version of it?
I think that could help deal with the traffic issue.
Has the team behind DALL-E Mini tried to share a downloadable version of it?
I think that could help deal with the traffic issue.
Well that's the problem. There is no "team", it's just one guy.
dang now i cant make "giga chad the rock as a clash royale card" fuck this site tbh 馃