Dalle 3 - The Biggest Update EVER In ChatGPT! Goodbye, Midjourney...

Dolly 3 is finally out and I believe that it's going to be a game changer to millions of people I've tried it tested it and the potential is truly endless so you better cancel your mid-journey subscription right now unlike the second version of this AI it is not a standalone website that you can use but that is a good thing if you are a Chad gbt plus subscriber since the beginning of October you should see a new option in the drop- down menu called Dolly 3 Once you switch to Dolly 3 you can just typee in your request for image generation in the chat so let's try something interesting for the first image I wanted to generate a photorealistic image of a capibara in a business suit reading a burning newspaper so I will type that and hit enter what's interesting is that since Dolly 3 is integrated into Chad gbt it starts rewriting my prompt a couple of times to give me a few options to choose from for each image it creates a separate and unique version of a prompt with different specifications and details I see that as a huge advantage over the competitors since you can not only see how AI thinks but also care less about how you write your prompts just take a look at the generated images look at the details the quality the consistency I've tested many image generation AIS but this is something else and do you want to see a magic trick I just type in the request to change capibara to the bald eagle and the AI instantly generates me a few images and look at them they are absolutely perfect it's exactly the same style the anatomy is spoton and even the font on the newspaper is similar to the real one it is a really good pack of images if you don't think that's impressive enough take a look at how Dolly 2 performs I will ask the same thing Kara in a business suit reading a burning newspaper and the first results are not ideal the second result after adding photo to The Prompt is a bit better even now it is clear how much better the third version of Dolly is than the second one struggles to maintain the style and adhere to my requests it is just all over the place very clearly AI generated and now let's change the capara to Bald Eagle instead of just typing what I want in Dolly 2 I have to click on a dedicated button and then manually select the area of the image I want to change and the AI will change only the selected Parts nothing else and what's worse I cannot just add a part that I want to change I need to change the entire prompt it takes more time and is much less intuitive a real hassle if you ask me and the generate results they are just not very good as I said dolly 2 changes only the selected part of the image and the head of our ego looks super weird and small Dolly 3 did much better in this test and it also was much

faster before we move on let's take a quick break and talk about this video sponsor gamma app if you are a knowledgeable worker office worker or run your own business you are surrounded by presentations and PDFs every time they're expecting more and more from you but you are already swamped and have no time so your presentations look super simple well with gamma app you can supercharge your presentation game free of charge this AI part service is super creative and creates presentations like you have never seen before from the design of the pages and layout to fonts and clever color schemes creating the presentation is super easy I just type in the topic and let AI do it job after that the platform gives me an outline to check and edit if needed after which the decks get generated these decks are basically slides of presentation but the feel is different more refined presentations here are mobile friendly by default which makes them easy to share and others view them from any device I can either manually change the presentation or ask AI to do the job for me I can ask it to change the presentation or add a new SL slide or even generate interactive multimedia content from scratch to fit my needs it's super easy to do just type in what you want and boom it's done what I like is how fast it all is the AI answers momentarily and all the generation happens in seconds even images take a couple of seconds to get generated which is faster than dolly or mid journey and the best part is that this platform can also design websites and even documents and doing all that takes literally seconds saving a ton of time when you are on a tight deadline what you also might find interesting is that you can collaborate with others further speeding up the work and the platform gets updated every week new features improved stability and so on so if you don't really know how to create a beautiful presentation try gamma app I will leave a link in the description for you basically the same thing happens when I use my journey the only way to use this AI is through Discord chat and before each prompt I need to type in in/ imagine yes the generate images do look really good that's the main selling point of the service however for the same price as TBT Plus subscription you get no option to chat with your AI artists with mid journey I need to create a separate prompt for each image and if I need to change the capibara to Bald Eagle like I did with Dolly 3 here I need to write a custom prompt from scratch and this is not intuitive or usable when there is such an easy way to do things like with Dolly 3 and chedi to prove my point let's change the style of the image with Dolly 3 it is as easy as just asking Chad gbt and boom FAS and FB style images are complete Dolly 2 needs a separate prom to do it and the results are wildly inconsistent only one image out of four is somewhat similar to what I needed mid Journey almost nails to style but usability is seriously lacking since none of the newly generated images are versions of the previously generated ones so it's not changing the style of the image it is generating images from scratch I think you get my point now but

let me show you a few more interesting examples of what Dolly 3 is capable of I will ask you to generate me a photo of highrise builders in 1950s New York doing acrobatics and a beam I imagine something like that old photo of Builders eating lunch sitting on a beam a few thousand feet in the air and would you look at that dolly3 nailed all four photos from the first try the framing the style the accuracy every image looks great really hard to pick one Dolly 2 in the same task performs quite poorly with inconsistent and strange results I really do not like how it performs so far much worse than I expected to make things even more interesting let's change the setting to Medieval China Dolly 3 has no issues with changing the time frame and visuals while still retaining the core ideas that I needed and wanted to see the builders are doing gymnastics and Scaffolding and the background scenery is definitely Chinese the buildings have texture and details the perspective is nice and even people look like people dolly2 here once again performs really poorly the builders are levitating in the air their bodies and faces are crooked and unnatural and the images just scream AI do you know what else I've noticed how different the style of the images is Dolly 3 tries to stick closer with the prompt and give the closest possible interpretation of what I asked for dolly2 just gives strength and obscure images that are very obviously AI generated and do not look like something you would want to use anywhere and mid journey is just too artistic even when I ask for photo realistic images it still gives drawings and illustrations and it also does not always stick to the prompt sometimes completely missing the core ID of the image I do not say that M journey is bad it is great but you have to understand its strengths and weaknesses before you use it Dolly 3 is much more forgiving and flexible and do style is much better as far as I can see but I'm not saying

that Dolly 3 is perfect it can miss the point sometimes too for example if I ask you to generate an image of a goth Pony expecting to get a photo realistic or at least somewhat realistic image it will give me the cartoonish version from the series for little girls and only if I deliberately ask it to make it photo realistic it will give somewhat realistic results maybe it is trained on more difficult tasks but this is not a very satisfactory result and as you can obviously guess Dolly 2 here is a complete mess what is this and this is just super bad terrible even Med Journey performs much better given slightly more believable results not perfect but more believable at least these ones do not have such an apparent cartoonish and artistic style and I know that I can do so much more with mid Journey if I learned proper prompt engineering for it but I think this kind of loses the point the AI can be really good at doing some complicated and properly described things but it is only as good as it is usable why do you think chbt is the number one Tex generation AI in the world because it is easy to use you're just chatting with a writer behind the screen like you would do with any normal person you don't have to write weird prompts like you do with mid Journey the same goes for Dolly 3 I believe it will become the number one image generation platform when it properly launches as a standalone website even from what it can do now from the chat is super impressive I think I could confidently say that the chat version of Dolly 3 is the best version that a regular person might want to use think of all the people who just want a quick and simple image generation the teachers creating illustrations for students the artists looking for inspiration the entrepreneurs trying to capitalize on AI no one wants to learn prompt engineering and dive deep into how it all works everyone wants to just type in the request and get a perfect result every time or at least with minor edits and that is exactly what Dolly 3 is aiming for with each update it will get more and more features and we'll learn to do more and more stuff and as you see it is already super good at changing parts of the image that it can detect imagine what will happen when it will be a standalone website with a proper Chad and ability to select and erase stuff like in Dolly to and do you know what the best part is it is included in Chad gbt Plus subscription you just pay 20 bucks once a month and you get access to this superb image generation and the leading text generation platform is there a better deal than that Dolly 2 uses a token system that is convoluted and inconvenient and I'm not even talking about the terrible results that it gives mid journey is pricey and needs proper knowledge to use to full advantage so I think if you are a general audience and you just want to generate a few images a few times a week without burden yourself with studying the specifics of the prompt engineering you will not find anything better than Dolly 3 and I will continue testing it and pushing it to its limits