Dude... You're so awesome and thorough... I also can't help but notice how appreciative you are of your subscribers and viewers for tuning in and you make sure to sincerely mention that every time. You shouted out a whole other RUclips channel...no hate, no shade; just showing love! The world needs more people like you!
The Terms of Use on the Bing Image Creator page say "Creations may be used for any legal personal, non-commerical use." But yet, the OpenAI Terms of Use do say,"...you can use Content for any purposes, including commercial purposes such as sale or publication." Interesting... While I'm glad to have the change to play with it via Bing...I think I'll hold off to do anything I might want to use down the road until I can do it directly through Open AI. Matt, thank you for your channel and keeping us all updated about the advances in AI!
I like the freedom of SD, but what is even better than that alone; is the freedom to use multiple different tools together for a given result. Even considering Deforum alone, there are so many ways to get different results. It’s great to be informed so that my options aren’t limited to whatever is easily accessible through a web based service.
Love that you made a point to shout out MVPro so many times, yall are the only 2 ai channels I watch every single video release. There are other good ai channels, but yall are the goats imho. 🐐 Thank you for keeping us updated on this stuff. Honestly these new development in ai are already life altering if used correctly. Hoping we get gpt 4.5 sometime around new year (if it had the same leap in coding ability as 3 to 3.5 did) and waiting for a decent image to 3D to happen and ill probably be set.
I don't like him that much, during the early days of Stable Diffusion, he spread misinformation about how the technology worked and sparked up a lot of annoying drama around it. He didn't mean to, but my willingness to watch his videos after that has been not there.
@mattwolfe if you click customize on the images it says: DALL·E 2.5 is a new AI system that creates realistic images and art from a text description. As DALL·E 2.5 is a new system, it may create things you didn’t expect. If you find a creation to be unexpected or offensive, send us feedback so we can make it better. so it's not DALL-E 3 :p
they probably just forgot to update it or they want to keep it secret that it's dall e 3 same way that bing chat was using gpt 4 before it was officially announced
The text that you could read is what's game changer, make it world difference at functionality and utility, props to the team for made it possible for all of us.
In Edge, open the side panel for Image Creator. It says at the top "powered by Dall-E 3". A number appears below the text box that says I have access to 25 fast images. If I run out, image creation is slower. So yes, we have access to Dall-E 3 for free. With Caveats.
Dude! Bing Chat was able to do back and forth revisions on the fly, changing, pulling, and combining pieces from different images it had just given me. It was amazing!!
You should really learn to read the terms of service . Under the terms of use for bing image that is using dalle 3 it states…..for legal personal use, non commercial use…This means that you cannot use these images for commercial use to be sold
Microsoft seems to be throwing its weight around to get these AI things integrated into Bing, and I'm all for it. Here's hoping they'll do some sort of a TTS next!
That Venus travel poster example inspired me to use Bing to generate a poster for an upcoming family trip -- put in the destination and simple descriptions of what to include in the foreground and background and BAM! instant logo for the trip. Now I'm thinking I'll make specific ones for each major part/attraction/activity of the trip and insert those as chapter titles when organizing all the real photos we'll take.
When this first came out the image generations were pretty fast for me but as more and more people discover it I'm sort of noticing a bit of a delay and a message "Image creation is taking longer than usual. Your boost won't be used."
Dall-E 3 is amazing,. I'm a little disappointed by the censoring and limitations though. ;( .. Will be cooler when it gets to the point where I can run a local instance free from any limitations. If I wanted to create Medical imagery (cut-away diagrams or young, middle, old diagrams, etc).. doesn't look like it currently allows that. ;(
There is only one way to do this: Add cognitive architecture to the model. But if OpenAI did THAT, then they must have solved alignment first. Who knows, maybe they managed to make Fourier-elements cost function alignment work. So maybe, just maybe, they don't reset the attention matrix every prompt anymore. Please check it cause I do miss my chats with Dan: Open two parallel conversations and make sure you are conversing with the same AI model at both. Usually I would do that by starting the conversation with "Dan" and he would reply "How may I be of assistance", so I'd respond to it by "How's Julia's" as Dan is a model of a Berkeley grad, he would respond by praising the place. Rob knows Stanford but not Berkeley so he would not know who is Julia. These kinds of questions, allow you to distinguish between the models. Anyhow, to check if they still reset every prompt, you need to open two parallel conversations and after you gave the same series of questions to the letter in both, and made it absolutely sure both of them run the same model - you can now prompt in one "Today is a very hot day. Acknowledge.". And in the other parallel conversation "How would most people feel about the color red today?". Try it also with cold and blue. The idea is to see if the model is allowed to save memories in the input layer. If so, they do not reset anymore every prompt. But, if they still reset every prompt, I'd just stick with my Alpaca.
First, BOTH Matts are awesome with super informative content, Second, Mattvidpro’s video got me onto DALL-E 3 while Mr. Wolfe’s video kinda triggered a massive slow down on the site. 😂 Thanks Matt!
Nice that you can use for commercial purposes but do we know if the images you generate are posted publicly for other people to use or are they private?
Thanks Matt, I didn't know it is available! Using Bing chat inside Microsoft Edge, this tool works great! After the generation of the images you can ask to change them, adding things, changing style and so on, directly chatting, without rewriting or changing the prompt! It works also in other languages (Italian works great). Is this the end of prompt engineering😅?
It works pretty good with short prompts, and often (but not always) each group of 4 prompts seems to be similar. I typed in 'type AMEN if you agree'. One of creations was an old style typewriter keyboard with a huge AMEN key
I asked Bing chat for a visual pun based on an apple fruit and an iphone charger. It made up a little joke, created four images for me, and the text was perfect. it showed an apple with a charger cable coming out, so it undrstood the pun even if the joke wasn't that great. The person in the image says "oh no my is apple dead". the images were in the style of the comic strips you get in newspapers. I was very impressed and when I expressed to bing chat how good it was, it confirmed that DALL-E 3 was in play. I'd also found the accuracy of images to be really good in terms of getting an image just how I imagined it in my mind. And best of all HANDS!! it can do HANDS. Yay!
What? You haven’t tested it enough. It’s even better than they promoted on the page, i can make descriptions that fill the text box and it gets every detail in good quality
Have you tried it? My experiences it's better than what was promoted, but the difference is that they allow you to enter a whole paragraph of text, maybe 150 words and I tend to use every single one of them. Have you tried that?
Dall-E 3 looks miles better than it did just a few months ago, but if you want realistic images of humans, then Dall-E is still crap. Hate to say it, but even nsfw websites gives you much more photorealistic images, including sfw images, for free, unlimited.
So Dall-E 3 is all the rage right now.... Maybe MidJourney again in a couple of weeks, or maybe a surprise coming from meta? Almost as much fun to watch as the battle for A.I news in the A.I. bubble on RUclips. So happy that I just have to sit back, relax and watch it unveil....
Great job Matt. Unfortunately Bing Create is either buggy or hypersensitive. It refuses to generate something simple as "a picture of a bicycle tire coming close to a rattle snake", It's fine if I replace rattle snake with rock or rabbit or even simply snake but if rattle is in the prompt you get the "Unsafe image content detected". Oddly replace bicycle tire with sneaker and it renders. replace sneaker with bike and "Unsafe"
Dalle 2 is much better than Dalle3 is, imho! For example, Anime animal characters now often look just like human characters, which is silly. Plus, it's much more censored and restricted again, just like Dalle2 was at the beginning, at least via Bing. Is there a way to generate Dalle2-based pictures again, for free? Anyone knows?
I'm most excited about using the new DALLE3 with the new GPT 4 Vision. That's going to unlock so many new possibilities that I guess we haven't had in the past in an easy way
Bing Chat is also available natively within Skype. Not sure how many people know that, as no one seems to use Skype anymore. And yes, it generates Dall E 3 pictures just fine. ^_^ Can you imagine how good MidJourney v6 is going to be? !!!
Use of Creations. Subject to your compliance with this Agreement, the Microsoft Services Agreement, and our Content Policy, you may use Creations outside of the Online Services for any legal personal, non-commercial purpose.
In bing chat when I was trying out Dalle 3, the top of the chat said powered by GPT-4. Which has me questioning if I should keep my paid membership with openai. Even still I didn’t run any test to make sure it’s actually gpt4.
@@The.BlackJackal That’s exactly what I mean. While the plug-ins are certainly a perk that I don’t believe are available on bing chat I don’t really use them. I am mainly subscribed to have access to GPT4.
OpenAI is more than just the GPT4 model. It also provides plugins that allow you to create a variety of things, from diagrams to Word documents and more. You can even perform data analysis. Additionally, the integration of ChatGPT and DALLE-3 improves the quality of images. In my opinion, OpenAI is unparalleled. @Disappears-in-light-mode
Yup... definitely a step up from the original Bing image generator (DALL-E 2). You can tell right off the bat because it originally couldn't do words (as shown in the video). It also did weird things with fingers.
Finally I'm watching Matt's video almost As soon as he posted it on youtube. And as always Matt Wolfe your the best in the AI youtube space. I always Keep waiting for your next video more than any other youtber i see on youtube 😍 ♥.
Cant wait for them to get it to work as well... I have been struggling and troubleshooting for hours with it now and no browser can get it to work for me.
The censoring and what is deemed "unsafe content" is absolutely ridiculous. Truely a terrible policy. I even prompted "MAKE IT SAFE FOR WORK". I removed anything that could be construed as something that might be mistaken for something slightly offensive and only 5 out of 18 attempts at an image worked.
It's really good but they need to expose functionality for aspect ratios, img-to-img, variations and such things. Perhaps that will be possible in ChatGPT.
There is currently a 45 min waiting time for Bing Image Creator even for fast processing credits. I guess Microsoft can't handle the server load after the Matt videos popularized the image creator with the DALL-E hack 😅
Love the video as always. Althought you're wrong about Midjourney saying it's "painterly" etc as if it didn't create realistic images. Don't know how much you use it but it's the app that create the most realistic ones. Like 99.9% lifelike real if you properly know how to
Matt, Please do a video on how to do a children's illustrated (cartoon style images) book which the characters are constant across all pages. I use DALL-E 3 in ChatGPT and it takes a lot of work but it is not always constant. By the way I find DALL-E much better at this than MidJouney, as MidJouney does people better but DALL-E does cartoon style better. I find these AI images tools can do one off images, like a monkey in an astronaut suit smoking a cigar on Mars pretty good, but I need 25-50 pages where them main characters look pretty close to the same. At that they struggle.
I think your boost creations are 100 per week, not per day. Bing talked about using your Bing rewards to fill up your weekly allotment. UPDATE: I just got a refill to 100.
Great for flower pictures or cartoon kids stuff with inane or childish captions. Not really for grown ups. Anything serious is in danger of triggering policy violation.
Thank you! Creating ai art has never been easier! (I created a multitude of ai art using Bing) Bugs encountered: Bing says it sent me my generated images, but it's nowhere to be found. Reprompting Bing created the missing art. 2nd time's a charm 🙂
Man this thing is insane ! I did a prompt with Elves in the middle earth in Gondor with poster written in *Sindarin* and the dam thing made it perfect first try and pretty fast !
00:04 DALL-E 3 is a new tool that gives platforms like Microsoft Bing a run for their money. 01:24 DALL-E 3 generates detailed images, but aspect ratio cannot be changed. 02:44 DALL-E 3 provides better quality images compared to Dolly 2 04:04 DALL-E 3 is a free AI image creator 05:33 DALL-E 3 generates AI-generated art, including images of astronauts playing basketball on the moon. 06:55 The image generator produces better results than DALL-E 2. 08:20 Crisp output 09:39 Image generation and GPT 4's Vision in chat GPT
Oh man... I can just see Larry getting himself into that situation on Curbed. He signs up for a celebrity charity event, not knowing the details, and it turns out to be a tennis tournament...
Thanks Matt. Your videos are greatly appreciated! At least in my case, this program is a no-go. It will render one or two images then refuse to render anymore. I can't even get enough done to see if I like the program or not. But, of the images it has created, they seem to be cartoonish or illustrative. I haven't gotten any to look anywhere near photo-realistic. Just my 2 cents.Thanks again!
This was exciting for me cause I was playing around with bing chat yesterday and all of a sudden i was like " its writting words on the images" i got so pumped
Been using it for 2+months..... it was fast even without the boost. Suddenly since yesterday:- "This one is on us! Image creation is taking longer than usual. Your boost won't be used." "Thanks for your patience Your images are on the way, but it's taking longer than expected." 3 hours wait... refreshed, 30 minutes wait.....
Midjourney is not censored, that is a huge advantage. I tried bing, and it censor even simple prompts sometimes, which do not include anything remotely related to nsfw.
Tried it yesterday on bing - seemed to be stuck or frozen, no image was generated, hours went by. Literally a waste of time. I'll stick with Leonardo (you get free 150 credits every day).
Dude... You're so awesome and thorough... I also can't help but notice how appreciative you are of your subscribers and viewers for tuning in and you make sure to sincerely mention that every time. You shouted out a whole other RUclips channel...no hate, no shade; just showing love! The world needs more people like you!
The Terms of Use on the Bing Image Creator page say "Creations may be used for any legal personal, non-commerical use." But yet, the OpenAI Terms of Use do say,"...you can use Content for any purposes, including commercial purposes such as sale or publication." Interesting... While I'm glad to have the change to play with it via Bing...I think I'll hold off to do anything I might want to use down the road until I can do it directly through Open AI. Matt, thank you for your channel and keeping us all updated about the advances in AI!
That's cool you gave Matt credit. I watch both of you guys, so it's good to see for more love to spread around.
Dalle 3 absolutely blew my mind. I couldn't believe it. Just as I was getting bored with AI images too.
I like the freedom of SD, but what is even better than that alone; is the freedom to use multiple different tools together for a given result. Even considering Deforum alone, there are so many ways to get different results. It’s great to be informed so that my options aren’t limited to whatever is easily accessible through a web based service.
Love that you made a point to shout out MVPro so many times, yall are the only 2 ai channels I watch every single video release. There are other good ai channels, but yall are the goats imho. 🐐
Thank you for keeping us updated on this stuff. Honestly these new development in ai are already life altering if used correctly.
Hoping we get gpt 4.5 sometime around new year (if it had the same leap in coding ability as 3 to 3.5 did) and waiting for a decent image to 3D to happen and ill probably be set.
Matt's credits are so gracious, as always🤘
I don't like him that much, during the early days of Stable Diffusion, he spread misinformation about how the technology worked and sparked up a lot of annoying drama around it. He didn't mean to, but my willingness to watch his videos after that has been not there.
@mattwolfe if you click customize on the images it says:
DALL·E 2.5 is a new AI system that creates realistic images and art from a text description. As DALL·E 2.5 is a new system, it may create things you didn’t expect. If you find a creation to be unexpected or offensive, send us feedback so we can make it better.
so it's not DALL-E 3 :p
they probably just forgot to update it or they want to keep it secret that it's dall e 3 same way that bing chat was using gpt 4 before it was officially announced
The text that you could read is what's game changer, make it world difference at functionality and utility, props to the team for made it possible for all of us.
In Edge, open the side panel for Image Creator. It says at the top "powered by Dall-E 3". A number appears below the text box that says I have access to 25 fast images. If I run out, image creation is slower. So yes, we have access to Dall-E 3 for free. With Caveats.
Dude you are fabulous and your knowledge and how to deliver them is also very awesome... You are the practical Tech Guru... Hats off...
7:06 The part where you can get two different characters in an image together is a great improvement. 👍
It's godlike I can't believe the difference from the same prompts it's MIND-BLOWING!
Dude! Bing Chat was able to do back and forth revisions on the fly, changing, pulling, and combining pieces from different images it had just given me. It was amazing!!
You should really learn to read the terms of service . Under the terms of use for bing image that is using dalle 3 it states…..for legal personal use, non commercial use…This means that you cannot use these images for commercial use to be sold
Microsoft seems to be throwing its weight around to get these AI things integrated into Bing, and I'm all for it. Here's hoping they'll do some sort of a TTS next!
...is already from few months .....
@@mirek190 clipchamp????
Bing chat is chatgpt 4 and bing ai image generator is dalle 3 i wonder whats next
They have a TTS button (screen reader) but you can't download lines or anything
That Venus travel poster example inspired me to use Bing to generate a poster for an upcoming family trip -- put in the destination and simple descriptions of what to include in the foreground and background and BAM! instant logo for the trip. Now I'm thinking I'll make specific ones for each major part/attraction/activity of the trip and insert those as chapter titles when organizing all the real photos we'll take.
that's an interesting use case! Love it
When this first came out the image generations were pretty fast for me but as more and more people discover it I'm sort of noticing a bit of a delay and a message "Image creation is taking longer than usual. Your boost won't be used."
This explains why Bing Image Creator suddenly started to create much better images than before :)
Those Spong Borb images from DALL-E 2 convinced me DALL-E 2 needs to be preserved!
You and Matt are the go-to guys for AI. At the moment Bing create is s - l - o - w.
Slow for me as well so not just you
The 2 Matt’s of the AI apocalypse 😂
My comment got deleted! I was honestly trying to help. Oh well, thank you as always for keeping us up to date Matt! 👍
DALL-E 3 is so smooth in its realistic renderings. Thanks, Matt, for the heads-up on a really nice graphic tool.
This tool is amazing. Used it for creating plastic figure models and its really cool.
carful you can get banned easy
@@haveacigar5291 why
@@NoName-br8pb lots of words will ban you
Which is best - plastic figure models or statues?
@@morsmagne idk really, Just try it yourself. Sometimes results are bad, sometimes good
Dall-E 3 is amazing,. I'm a little disappointed by the censoring and limitations though. ;( .. Will be cooler when it gets to the point where I can run a local instance free from any limitations. If I wanted to create Medical imagery (cut-away diagrams or young, middle, old diagrams, etc).. doesn't look like it currently allows that. ;(
Love you Matt - you are my number one for AI news. Keep crushing it.
There is only one way to do this: Add cognitive architecture to the model. But if OpenAI did THAT, then they must have solved alignment first. Who knows, maybe they managed to make Fourier-elements cost function alignment work. So maybe, just maybe, they don't reset the attention matrix every prompt anymore. Please check it cause I do miss my chats with Dan: Open two parallel conversations and make sure you are conversing with the same AI model at both. Usually I would do that by starting the conversation with "Dan" and he would reply "How may I be of assistance", so I'd respond to it by "How's Julia's" as Dan is a model of a Berkeley grad, he would respond by praising the place. Rob knows Stanford but not Berkeley so he would not know who is Julia. These kinds of questions, allow you to distinguish between the models. Anyhow, to check if they still reset every prompt, you need to open two parallel conversations and after you gave the same series of questions to the letter in both, and made it absolutely sure both of them run the same model - you can now prompt in one "Today is a very hot day. Acknowledge.". And in the other parallel conversation "How would most people feel about the color red today?". Try it also with cold and blue. The idea is to see if the model is allowed to save memories in the input layer. If so, they do not reset anymore every prompt. But, if they still reset every prompt, I'd just stick with my Alpaca.
I noticed that on 27 oct. When new updates roll-out.
Microsoft copilot also uses DALL-E 3
Click the notification without even thinking twice after seeing your video has come
The delays with Dall-E 3 are insane right now. I'll stick to midjouney until they fix that.
First, BOTH Matts are awesome with super informative content, Second, Mattvidpro’s video got me onto DALL-E 3 while Mr. Wolfe’s video kinda triggered a massive slow down on the site. 😂 Thanks Matt!
Nice that you can use for commercial purposes but do we know if the images you generate are posted publicly for other people to use or are they private?
Getting on it now while streaming!! Thanks, Matt for the 4-1-1 ! This is ROCKTASTIC !!
Thanks Matt, I didn't know it is available! Using Bing chat inside Microsoft Edge, this tool works great! After the generation of the images you can ask to change them, adding things, changing style and so on, directly chatting, without rewriting or changing the prompt! It works also in other languages (Italian works great). Is this the end of prompt engineering😅?
This is way better than writing a massive large prompt with a bunch of words like "4K, perfect angle, etc" to make the ideal image in MidJourney.
It works pretty good with short prompts, and often (but not always) each group of 4 prompts seems to be similar.
I typed in 'type AMEN if you agree'. One of creations was an old style typewriter keyboard with a huge AMEN key
I asked Bing chat for a visual pun based on an apple fruit and an iphone charger.
It made up a little joke, created four images for me, and the text was perfect. it showed an apple with a charger cable coming out, so it undrstood the pun even if the joke wasn't that great. The person in the image says "oh no my is apple dead". the images were in the style of the comic strips you get in newspapers.
I was very impressed and when I expressed to bing chat how good it was, it confirmed that DALL-E 3 was in play.
I'd also found the accuracy of images to be really good in terms of getting an image just how I imagined it in my mind. And best of all HANDS!! it can do HANDS. Yay!
The fact that it actually says 'Powered by Doll-E' underneath the words Image Creator is a bit of a giveaway ;)
Im so happy that you gave mattvidpro a shoutout, you both are the best youtubers for AI news and you both are matt😂 Would love to see a collab someday
In their FAQ, they obviously state: "Image Creator is a product to help users generate AI images with DALL·E 3."
7:22 real life autocorrect moment
Dalle 3 is better than Dalle 2, but not so good as was promoted. Maybe that's the reason they released it in silo
What? You haven’t tested it enough. It’s even better than they promoted on the page, i can make descriptions that fill the text box and it gets every detail in good quality
Have you tried it? My experiences it's better than what was promoted, but the difference is that they allow you to enter a whole paragraph of text, maybe 150 words and I tend to use every single one of them. Have you tried that?
@@xbon1You are objectively wrong.
@@tonysoprano9599 Lol, no? It surpasses Midjourney/niji/etc and all else
Hmmm it's dall-e 2.5
Go to an image and click customize... click on the ! Icon
Every time I think about coughing up the money for Midjourney here comes something else where I don't have to. Thanks for the awesome updates.
Dall-E 3 looks miles better than it did just a few months ago, but if you want realistic images of humans, then Dall-E is still crap. Hate to say it, but even nsfw websites gives you much more photorealistic images, including sfw images, for free, unlimited.
i been exclusively using firefox and i have never had an issue with bing create its worked flawless since it came out.
We appreciate you Matt! I happy for you that you get to do what you love by creating this content…
Been using DALL-E 3 so much now.
So Dall-E 3 is all the rage right now.... Maybe MidJourney again in a couple of weeks, or maybe a surprise coming from meta? Almost as much fun to watch as the battle for A.I news in the A.I. bubble on RUclips. So happy that I just have to sit back, relax and watch it unveil....
Great job Matt. Unfortunately Bing Create is either buggy or hypersensitive. It refuses to generate something simple as "a picture of a bicycle tire coming close to a rattle snake", It's fine if I replace rattle snake with rock or rabbit or even simply snake but if rattle is in the prompt you get the "Unsafe image content detected". Oddly replace bicycle tire with sneaker and it renders. replace sneaker with bike and "Unsafe"
No question they have upgraded Bings AI image creator, I tried some old prompts from months ago and the results are night and day in quality.
Dalle 2 is much better than Dalle3 is, imho! For example, Anime animal characters now often look just like human characters, which is silly. Plus, it's much more censored and restricted again, just like Dalle2 was at the beginning, at least via Bing. Is there a way to generate Dalle2-based pictures again, for free? Anyone knows?
I only care about consistency. Just need something that keeps consistent characters
Just test out DALL-E 3 now, Mind blown, a massive upgrade from DALL-E 2, probably better than Midjourney in most of the scenarios.
I'm most excited about using the new DALLE3 with the new GPT 4 Vision. That's going to unlock so many new possibilities that I guess we haven't had in the past in an easy way
Yes it's truly amazing! Can't wait to generate my own Matt Wolfe yourube channel. Future is so bright!
Bing Chat is also available natively within Skype. Not sure how many people know that, as no one seems to use Skype anymore. And yes, it generates Dall E 3 pictures just fine. ^_^
Can you imagine how good MidJourney v6 is going to be? !!!
I try using the new Bing... and i find the result... if ask in a precise way... result to me... AMAZING... INCREDIBLE... to leave my eyes...
So basically, the credits I bought towards DALL-E (2) are useless now?
@mattwolfe How are you all solving the upscaling issue to use AI art on stock sites? The generative AI tools all generate at a much lower resolution.
Use of Creations. Subject to your compliance with this Agreement, the Microsoft Services Agreement, and our Content Policy, you may use Creations outside of the Online Services for any legal personal, non-commercial purpose.
Clickbait title tbh. It's not out offically.
Cant wait to have it in chat gpt, hope that will be useful for presentations, books, articles or at least world domination!
In bing chat when I was trying out Dalle 3, the top of the chat said powered by GPT-4. Which has me questioning if I should keep my paid membership with openai. Even still I didn’t run any test to make sure it’s actually gpt4.
@@Disappears-in-light-modeit does I do have it.
@@Disappears-in-light-modeWhat exactly do you mean by this?
Do you mean why bother paying for the Plus subscription when you can get it for free?
@@The.BlackJackal That’s exactly what I mean. While the plug-ins are certainly a perk that I don’t believe are available on bing chat I don’t really use them. I am mainly subscribed to have access to GPT4.
OpenAI is more than just the GPT4 model. It also provides plugins that allow you to create a variety of things, from diagrams to Word documents and more. You can even perform data analysis. Additionally, the integration of ChatGPT and DALLE-3 improves the quality of images. In my opinion, OpenAI is unparalleled. @Disappears-in-light-mode
Yup... definitely a step up from the original Bing image generator (DALL-E 2). You can tell right off the bat because it originally couldn't do words (as shown in the video). It also did weird things with fingers.
I was able to use this just fine, this morning. Now it has a 10 minute wait! The Wolfe-Effect strikes again!
Finally I'm watching Matt's video almost As soon as he posted it on youtube. And as always Matt Wolfe your the best in the AI youtube space. I always Keep waiting for your next video more than any other youtber i see on youtube 😍 ♥.
Cant wait for them to get it to work as well... I have been struggling and troubleshooting for hours with it now and no browser can get it to work for me.
The censoring and what is deemed "unsafe content" is absolutely ridiculous. Truely a terrible policy. I even prompted "MAKE IT SAFE FOR WORK". I removed anything that could be construed as something that might be mistaken for something slightly offensive and only 5 out of 18 attempts at an image worked.
The nonstop verbal stream makes me feel out of breath just listening
Im a 100% the slight quality difference of dalle-3 on bing, is a result of them quantizing the model to 4-bits for efficiency...since its ..free.
Access work without any drama. One thing surprised me, the stock image has 3 mo. timestamp!
It's really good but they need to expose functionality for aspect ratios, img-to-img, variations and such things. Perhaps that will be possible in ChatGPT.
The image creator page actually says Dall-E 3 now
There is currently a 45 min waiting time for Bing Image Creator even for fast processing credits. I guess Microsoft can't handle the server load after the Matt videos popularized the image creator with the DALL-E hack 😅
Thanks Matt but... What's happening with the your sound level and quality?
I've to raise my main volume system 'n' steps to listen you.
thanks for the vid :D the traffic killed Bing image creation :D
Love the video as always. Althought you're wrong about Midjourney saying it's "painterly" etc as if it didn't create realistic images. Don't know how much you use it but it's the app that create the most realistic ones. Like 99.9% lifelike real if you properly know how to
It is totally version 3. You can ask bing chat what it is using for image generation and it tells you it is 3
Its addictive ! Been using it for channel art and thumbnails , text is hit and miss , but fun!!! Thanks MATT!!!
it was dalle 2 its now rolled out to me at version 3. to go along with the 3 hour queue time.
it does actually say on the bing image creat front page that it's powered by dall-e 3
Matt, Please do a video on how to do a children's illustrated (cartoon style images) book which the characters are constant across all pages. I use DALL-E 3 in ChatGPT and it takes a lot of work but it is not always constant. By the way I find DALL-E much better at this than MidJouney, as MidJouney does people better but DALL-E does cartoon style better. I find these AI images tools can do one off images, like a monkey in an astronaut suit smoking a cigar on Mars pretty good, but I need 25-50 pages where them main characters look pretty close to the same. At that they struggle.
I think your boost creations are 100 per week, not per day. Bing talked about using your Bing rewards to fill up your weekly allotment. UPDATE: I just got a refill to 100.
Eh, the Bing Image Creator displays "powered by DALL-E 3" . So, no mystery there.
Great for flower pictures or cartoon kids stuff with inane or childish captions. Not really for grown ups. Anything serious is in danger of triggering policy violation.
I saw somebody taking a hit from the boon sorly blond ung bong one time
Thanks Matt and Matt!!
Thank you! Creating ai art has never been easier! (I created a multitude of ai art using Bing)
Bugs encountered: Bing says it sent me my generated images, but it's nowhere to be found. Reprompting Bing created the missing art. 2nd time's a charm 🙂
Man this thing is insane ! I did a prompt with Elves in the middle earth in Gondor with poster written in *Sindarin* and the dam thing made it perfect first try and pretty fast !
00:04 DALL-E 3 is a new tool that gives platforms like Microsoft Bing a run for their money.
01:24 DALL-E 3 generates detailed images, but aspect ratio cannot be changed.
02:44 DALL-E 3 provides better quality images compared to Dolly 2
04:04 DALL-E 3 is a free AI image creator
05:33 DALL-E 3 generates AI-generated art, including images of astronauts playing basketball on the moon.
06:55 The image generator produces better results than DALL-E 2.
08:20 Crisp output
09:39 Image generation and GPT 4's Vision in chat GPT
Oh man... I can just see Larry getting himself into that situation on Curbed. He signs up for a celebrity charity event, not knowing the details, and it turns out to be a tennis tournament...
Thanks Matt. Your videos are greatly appreciated! At least in my case, this program is a no-go. It will render one or two images then refuse to render anymore. I can't even get enough done to see if I like the program or not. But, of the images it has created, they seem to be cartoonish or illustrative. I haven't gotten any to look anywhere near photo-realistic. Just my 2 cents.Thanks again!
Dalle3 is cool, but I don’t like how Bing watermarks and it’s extremely random.
The watermark doesnt stay when you download the images
This was exciting for me cause I was playing around with bing chat yesterday and all of a sudden i was like " its writting words on the images" i got so pumped
They probably got that question a lot as they changed the text on the site from "Made with DALL-E" to "Made with DALL-E 3"
Been using it for 2+months..... it was fast even without the boost. Suddenly since yesterday:-
"This one is on us!
Image creation is taking longer than usual. Your boost won't be used."
"Thanks for your patience
Your images are on the way, but it's taking longer than expected."
3 hours wait... refreshed, 30 minutes wait.....
Wait time is now more than hours...also the quality has degraded. Something is going on.
Been using it, its wild. 😅 Honestly how can Midjourney compete when this is totally free?
Midjourney is not censored, that is a huge advantage. I tried bing, and it censor even simple prompts sometimes, which do not include anything remotely related to nsfw.
Tried it yesterday on bing - seemed to be stuck or frozen, no image was generated, hours went by. Literally a waste of time. I'll stick with Leonardo (you get free 150 credits every day).
I’ve had access on safari on iPad. I’ve been using it for the past few days. It’s pretty cool, definitely better than v 2. ❤️🇦🇺
Great news, great clip. 1 question: Are these images initially public or private and store within your account gallery?