Getting Started With Playground AI + Stable Diffusion

hello everyone and welcome to playground AI the text to image web UI that will Unleash Your creativity text image AI is a type of artificial intelligence technology that generates images from text to get started is quite simple head on over to playgroundai.com you'll be directed to the home page which is the community Gallery here you could sift through various images here to either get inspiration learn about the prompts but we'll come back to this later to get started you simply have to click on the top right icon here click on login and sign in with a Google account once you're logged in simply click on your profile to give you more options on the site I encourage you to check out the FAQ section if you're new to playground AI there's also a direct link to join the Discord server and a quick access option to request support now let's go ahead and click create and start creating some images what you see here is the user interface and I'll explain all these settings as we create on the left here is where you would enter a descriptive text of the image you want known as a prompt personally I like to start off with a simple format the kind of image you want it could be a painting a drawing a sketch and what is your main subject or theme you may also want to consider your environment so to start off we're going to try something very simple here and I'm going to enter a cute and adorable chipmunk eating acorns now before we generate the images let's go to the right side here under model I'm currently using stable diffusion 1.5 think of this as the base model to create your image there are also options for stable diffusion 2.1 but if you're starting out I encourage you to start with 1.5 for now there's also an option for Dolly 2 for an additional cost for now let's select stable diffusion 1.5 just below we have our image Dimensions we'll leave it at 512 by 512 for now and the rest of the settings will come back to as we progress with the image but the one thing I do want you to check on is the private session otherwise all the images you produce will be showing on the public home page number of images let's go for four all that's left to do is to click on generate and here are the rendered images based on the text that we inputted since we're here I want to show you the column slider here if you slide that to the right you can select how many images that you want per row as for the results we have a cute and adorable herbal chipmunk eating acorns while kind of eating acorns we have some on the ground here this one seems to be having a beverage as you can see the images are fairly coherent and very photorealistic but maybe you wanted something a little different like a cartoon style one of the benefits of playground AI is that we have some built-in filters here that can change the look and style of your image if we click on the drop down here it will reveal several filters that you can choose from for this example we're going to click on play tune and simply hit generate once again now as this runs you're going to see that there's some additional text within the image and that's what the filter does it does some of the work for you and you could simply focus on your ideal and build from there we'll get into more complex prompting in future videos now it's quite obvious the results of this image is very different from the photorealistic images that we got before the results here are more like a Pixar Style cartoon one thing to take note of text to image AI is very random and not really consistent so you will get varying results and even results that just don't come out just right like I think they were attempting to put glasses on this one that didn't work out we could simply delete this image from here by going to the bottom right to delete the image this one's kind of cute now as cute as these are I'm gonna generate four more images because I'm looking for an image that I can work with that has a composition that I could build upon the other thing I want to do is change the aspect ratio to 512 by 768 so I'm simply going to generate four more images here here are the first two images as you see they're not that great we actually have acorns coming out of the side of this one and then we have a human like figure here that's going to happen every now and then if we look at the other two here much better results but it definitely needs some work you'll also notice that the proportions are off the hands may be a little janky the arms may be a little long just know that's very normal but the one thing that can help this is the use of negative prompts if we go to the left here you'll see this option remove from image we can click on that and then we can enter in things we don't want in the image for example deformed limbs double heads long necks extra arms extra legs you get my point I'll leave a link in the description below to a Google doc where you can basically copy and paste common negative prompts that I I personally use keep in mind this will not prevent the deformities but it will help minimize them there's a couple things we can do here we can experiment with the advanced options here we click on here we see an option for sampler I won't go too in depth with this right now but the Samplers I like to use is either Euler Euler ancestral dpm2 or dpm2 ancestral these four are kind of my go-to's some are a bit more artistic some do better with photorealism I encourage you to experiment and see what works for you for this I'm going to switch it to Euler we'll stay on the play tune filter but I quite like the composition of this image this is sort of the image that I had in my head so what I'd like to do here is create some variations now there's two ways to do this we go to the top left here you'll see an option for create variations if we click on that it's going to generate four more images based on this image you see that it's slid into this area called image to image now it's going to take this image's likeness and create similar variations now you could clearly see that each image has similar composition but changes in the smaller details we look at these two images here we see even this one has like a color here he's actually holding something this time that's what I'm trying to do here let's quickly talk about image strength if I slide this to the left and make the number lower we'll make this 10. the variations will be more creative you're basically telling the AI keep the likeness of the composition but be a bit more creative with the image as we look at the rendered images again you see how slightly different they are now it's now developing more of a realistic tail we look at these images here even better here previously it had this kind of almost like a shell on its back but I'm really after something like this where the Chipmunk is holding the acorns or nuts or whatever it is now if I bring the image strength higher to let's say 65 it's going to keep more of the likeness of this original image here and it's only going to make slight changes as you can see they're very similar to this original image slight differences we have that shell-like back here and slight differences in some of the details whether it's the eye color between these two this one the Chipmunk is holding the nuts I'm actually going to come back to this one and use it for image to image now before we do that beside the create variations icon is your download button you could simply click that to download the image this will expand the image to full size so that you can view it properly and this rate image feature is really for analytics for the developer to know whether you like the results or not I'm going to click on love here because I like it and send them some feedback the actions icon here will bring up similar controls download image to image view full screen and delete we also have face restoration which is great for portraits and to upscale the image if you recall we had Dimensions 512 by 768. if we were to upscale this four times it would be 2048 by 3072 at 72 DPI let's go ahead and bring this into image to image I just want to make a small adjustment to this we're going to bring this back to about 40 and I want to quickly introduce you to in painting it's grayed out at the moment because we have the filter on we just need to turn it off at this point it's not absolutely necessary to have the filter on since we have an image as a reference now you see that the option is available we're going to click on that and I just want to see if I can improve the hands here a little bit and get the nuts a little more uniform down below here you're going to see some simple options for the masking brush this simply increases or decreases the size and then there's an eraser here so you simply just have to paint over this area and what the mask is for is you're telling the AI I just want to change this section of the image leave everything else now if I made a mistake I could simply undo it and do it again or maybe I want to you know get a different type of tail here so I'm going to mask this area for potentially more tail and then we're going to click done you'll also notice on the right that we have a number populated here under seat a seed number sequence is related to this image every image that you generate will be assigned a seed number I'm going to remove the image and just show you a couple more of these filters with the same prompt if we select poly mode for example let's make sure that we have random images here and generate four Images we're gonna get a completely different style and look I like to think of the filters like built-in presets as you can see totally different results from what we were getting with the previous filter much more toy like very cute and adorable I quite like this result actually in the next video we'll cover some basic prompting from scratch without using filters to get the results that you want let me know in the comments below if that's something you're looking forward to in the meantime I do want to show you your profile Gallery so if you click on your icon here click on profile here you will see your personal gallery now you can sort it by your newest your oldest or your most light there's also a section here where you can put in your website in my case I've got my YouTube channel here but the great thing about your personal gallery or even the home page Gallery is you can click on these images and see the prompts that were used the negative prompts that were used you can copy The Prompt Here remix is an option if I click on this it'll populate all the prompt areas here the settings that were used and produce similar results now notice the seed number here we can also randomize it to get various results but again with the same settings just below we have our seed number the settings that were used how many likes the image got download button a copy link button that you can use to share on social media if we click on the three dots here you can make this private or public or remove the image you'll also see similar images based on your prompt that other people have done now that you're more familiar with playground AI I'd love to see some of your work in our Discord server so make sure you go ahead to your profile icon click on join Discord and share your AI generated images with the community also let us know in the comments below what you want to learn next until we create again my friends I'll meet you at the playground

Channel: Playground AI

Views: 231,451

Keywords: Getting Started With Playground AI, playground ai, playgroundai, playgroundai.com, text to image ai, text to image, text to image generator, artificial intelligence, ai, stable diffusion, stable diffusion ai, ai image generator

Id: wITLbrDGy-0

Length: 14min 13sec (853 seconds)

Published: Thu Jan 26 2023