 Hello everybody and welcome back to another Ethan journal video in today's video We are going to be doing a part two to my creating images with AI On a Mac. I did this video back in October as a part one and that video really blew up It's like one of the most popular videos on my channel right now And I thought I'd kind of do a follow-up now that I've learned a lot of stuff from the comments I got on that video between YouTube and rumble And I've learned a lot I've learned a lot about diffusion B and how to run the whole diffusion AI in image engine So I think let's take a look again at this and see if we can improve and get some better images And just like last time I will be posting these images in the description. So the further do let's get started All right everybody. So here we are in the diffusion B app as you can tell it has changed a lot They have a lot of new settings Such as text image image to image and outpainting Along with the history so you can see the images that you've created in the past. They also have prompt ideas Styles you can actually like choose the kind of image you want It already pre-fills that in there to like help the AI know what you're looking for It also has a lot more options now. So they have higher resolution images And they've essentially created a lot of this more now. I should also add In here that these this engine that runs this is created by a company That I forget the name of but I'll post them here on the screen But I also learned that they have this essentially this entire app on their website So if you don't have a Mac, but you still want to try Experimenting with image engine diffusion you can still experiment with it on their website completely for free I know that's a really nice thing about diffusion unlike open AI's dolly It's completely free to use and this app is available on Mac. I didn't pay for it at all It's just here for me to use and it's standalone. It doesn't need to be on the internet Which means I'm assuming it doesn't have as any tracking or analytics and I Declined any of the ones during setup, but it's a really decent app and I really enjoy it. I'm not being paid By this company in any way. It's just a really cool app that I found on github that I thought would make cool videos So let's get started. Let's go with the same one that we did last time and see if we can get anything So I'm gonna do a cabin on On the edge of a lake during The night and now I've done this and then what I'm also going to do is I'm gonna go here into styles and I'm gonna choose some some styles So first off what type of image you want. We want to do a drawing style. I want to do a visual style I'm gonna probably go with this one as like a sketch because I think that'd be kind of a cool art style for this too And I'm gonna have it do that one as Like a colored pencil sketch. Maybe that'll kind of look neat I'm not I wanted to be HD for sure because it'll probably help with resolution along with Maybe making it look some cool Effects in here as well, but I'm just gonna leave it at that for now We can experiment with this later on but I'm also going to do this simply at the 512 by 512 resolution And let's see what comes out again like last time I do have to stop the video because this app uses pretty much my entire resolution. So I'll see you guys all I'll see you guys all when the image generation is done. All right everybody. I am back It has literally been 15 or 30 minutes I don't know I was helping my parents set up with Christmas while I was waiting for this Image to generate but you guys are probably gonna be pretty happy because it actually turned out pretty gosh darn well Ta-da, it actually looks like a painting or more like a sketch. Um, I Actually think it turned out pretty well. It's got a lot of details. It even has like the reflection in the water It's really really nice So I'll go ahead and save this And we will get started on the next image All right. Well now we know that worked pretty gosh darn well. So let's come up with our next image I want to similarize it to the last episode we had so let's do another one similar to the other one, so let's do a camp site in the middle of a forest At night and then similar to last time. Let's go into our styles here For this one, let's do something a little bit different. Let's do a detailed and intricate And that's not doing more there, but let's try to make this one Let's see. Are there any really good ones in here that we could do? We could do a geometric We could do a comic book digital art graphic novel a mosaic. That'd be kind of cool We could do a realistic one. We could probably try that We could do a CGI make it look computer-animated. Do you want to try doing something like that? Let's do it Let's just do a CGI And then we'll go down here. We'll get the information. We'll do high definition And then there's anything here to make it look really good like we could do a cinematic like a nice look of there So let's add in those. We have a lot of a lot of styles in there now What do we want to do for lighting? Do you want to try something there? Let's see. I Just can't do beautiful lighting. All right, so that's gonna be our prompt and let's just see after this is done what ends up coming out of this random contraption of search terms or Image generation terms, so I'll be right back All right, I am back and this one turned out pretty similar. It actually looks pretty nice So I'll show you guys it right now here. It is It's similar design to the last one. Here's the image It'll probably be up right above me of the actual image of the last video with a similar prompt Of course a little bit different, but it turned out pretty well. I actually think that This CGI look it kind of made it look like it's in a fisheye lens But the image turned out pretty pretty decently as well So let's come up with some with some other prompt ideas All right, everybody, so I took some time to try to think about a decent idea That I could use to come up with something and I finally have it I want to try using the aerial view like an over like bird's eye view type thing I want it to show me like an aerial view of like a downtown. So a Future futuristic Downtown Street and that's all I'm gonna put Because then I can go into here in the styles and come up something so first off I'm not gonna do any of those but I do want it to look real. So I want it to be Is there anything in here for modern field view in HD and Make that look cool and I'll make it With dynamic lighting because I think I'd be kind of right there. I don't need an artist specific and Should I you make it look like an actual engine because I could make it look like but I want it to look real It's my thing. So maybe if I go into here and I find something like So photo realistic and never mind I'll just do realistic here and I think that'll look pretty good And now what I'm gonna do is I'm going to attempt to upscale it as well So let's see if I can make the image, but I want to make it like landscape kind of so I'm gonna do that Well also making it like 760 so it'll be it'll be like I'm like a landscape tile style image instead of being the square So this one is saved. I'm gonna have this one generate and I'll be right back when that one is done Okay, everybody we have come across our first failed image It's kind of sad. I was really excited for this image. I don't know what went wrong We might have messed up somewhere in the thing, but I really don't see our claimant wrong Again, it's AI generated. So I'm actually gonna try this again, but just so you guys can see it This is how it turned out It's just a garbled mess I'm kind of upset with that, but it's fine. It's honestly AI you can only get a couple of really good ones Well before you're gonna get a bad one, and that's how AI Image generation works. It's a computer trying to understand what you're saying and putting that into something We have to accept that it's not always going to be correct And that's something I've learned a lot. I've done a lot of these on the side Both with this and using the GPT-3 model from open AI, which is a text generation one They're both really really good and pretty much anything is. I've also experimented with mid-journey and And Dolly, I've experimented with pretty much the main three ones that are out there right now And all three of them really have really good artwork that can come out of them I'd probably say mid-journey has been the best one I've been able to find Proud with that one is it's on discord, which not everyone can access and it does cost ten dollars a month to use If you want to use more than the trial period of images generated But overall even though this one here did turn out not like we wanted it to I think it's pretty neat to see that we did get a mistake But what I'm going to do here is I'm just going to click ahead and do generate again And maybe we'll get something better this time if not we can change our prompt because I do want to get this image out before Before the video is done. So I'm going to click generate over here, and I'll see you guys when this one is finished up All right, everybody, we actually have a decent result this time it's still a little bit glitchy But maybe there's something wrong with my prompt But I will accept this one as a decent design definitely looks AI generated like the rest of these images have But I think it turned out pretty nicely here it is and Yeah, it's a pretty smaller than the rest but you guys can see these with the link in the description But I think it turned out really nice. You can actually see like the aspects of it being a city Here's an upscaled version of it, but you can tell it is trying to create a city If you on the smaller form it definitely looks a lot better, but here is what it looks like Looking at it a little bit more. It's it's not good. I'll put it. It's not really good But it's still a really cool idea It definitely got the it but it really did do is it did get all of the things you're looking for it Does look futuristic it does in an aerial view it is high definition somewhat I really can't confirm it to have the dynamic lighting and it looks somewhat realistic Now that's where I'm gonna wrap this video up But what did I learn through this process? Well, I definitely learned that a I am a generation is getting better since I Made late my last video Diffusion the entire diffusion AI engine has been upgraded a lot and so has this app It's updated the model quite a few times and it's increased its file size, but I buy good 3 gigabytes, which I'm assuming that's the different models It uses along with the system itself is what it's based off of but overall I think it has turned out pretty well. I'm excited to see where this goes and unlike crypto I think that I am a generation might be something that's gonna take off or at least stay here and not gonna just taper off into Non-existence other than that I would like to say thank you all for watching today's video Be sure to like and subscribe if you enjoyed it I really have nothing else to say other than that when I'm recording this I want to say Merry Christmas to you all because Christmas isn't a couple of days Well a couple of weeks at this point But whenever this video ends up going up when I'm done editing it might be closer to Christmas than it is right now But yeah, alright, so yeah, I hope you guys learned something in this video Be sure to check out diffusion B if you have a Mac. It's pretty cool program But yeah, thank you all for watching. Have a great rest of your day. God bless you and goodbye