5 WAYS To Get Started With AI ART [Beginner's Guide 2023]

so you just heard about AI generated art an AI tool that can generate an image or a piece of art using only just text and you want to try it right away then you are at the right place in this video I am going to walk you through the most popular options and ways you can generate air as briefly as I can whether you want a free paid online or local option to generate realistic stylistic or anime images this video will inform you about all of them however I will not be digging deep into their pros and cons so perhaps you should watch some of my older videos about air if you are interested now let's get started currently you can generate your AI images either locally on your computer or online the online option ranges from Discord websites phone apps to Google collab notebooks but if you just want to test it out as soon as you can without the hassle of installing or learning anything at all the fastest way to get started for free is either using open ai's Dolly 2 which is on a website or a mid Journeys Discord bot on Discord all you really need is an Internet connect action and it doesn't really matter if you are on a phone a laptop or a desktop for Dolly 2 you can sign up right away on their website and have 50 free credits with 25 additional credits every month where one generation costs one credit and you get four result images from a single text prompt for mid-journey you can join their Discord server and go to any of the newbies text channels type in slash imagine followed by your text prompt and it will generate it right away you will have 25 jobs which also includes modifying your resulting image with the upscaling or generating variation function however you will not get any additional free drops once you have used up all the 25 jobs the reason that I suggest that these two is because they have one of the highest qualities and can be accessed really easily through a website or Discord which makes it available for mobile users too however this is just the fastest way to get started but since I already mentioned some online services let's actually start by going over the ways to run them online running these text to image AIS online means you barely have to set up or obtain anything yourself and are using a service that is being provided by someone so naturally most of the time you have to pay money in return just like normal goods and services for online websites one of the most popular choices right now would be dream Studio which is made by one of the big three text to image companies called the stability AI stability AI is the only one that provides open source AI models called stable diffusion where everyone can download and run it themselves for free which I'll talk about later into the video in the local option but the web service dream studio is mostly targeted for the people that do not have the high-end Hardware to run the models locally the user credit system which is roughly ten dollars for one thousand credit and 0.2 credit can generate one image so generating one image roughly costs 0.2 cents because depending on the image size you want to generate the cost may increase or decrease another popular website would be Dolly too which is the one that I mentioned earlier the main difference between them is that they provide different models so using the same text prompt will give you different results since the results generated from a model depends on the data that was used to train it but Dolly 2 is slightly more expensive where 115 credits cost you 15 which is around 3.25 cents per image however the image size is fixed and it is limited to generate 4 images at once so I would say it is much less flexible and actually today's sponsor also lets you get cheaper rates for Dolly too more about them in the end and for the weaves that want to generate illustrations or anime related art websites such as novel.ai or hilara.ai would be the best because the models that they provide were fine-tuned to generate anime stuff much better than the general models Dolly 2 and stable diffusion has I recommend these two because they both use aspect ratio bucketing which allows good high resolution image generation and coherency so good results will be guaranteed more often polara gives 100 free credits upon sign up and you also get free credits for daily login their payment plan is a bit unique but roughly 0.7 cents per image or as low as 0 point four cents per image for the highest tier if you are a free user you can only Generate random images but it's a good demo for you to see how to generate good results since prompting correctly is really important it performs really well on high resolution too so you should check them out novel.ai has no free credits for image generation so you basically have to subscribe to at least a plan to generate images for their best plan it roughly costs 1.25 cents per image for image size larger than 640 times 640 with more than 28 simple steps but anything lower than that would be completely free if you buy the plan on the other hand running online also includes using the messaging platform Discord mid-journey is one of the text to image AI services that is only available through Discord and like I mentioned earlier you get 25 free drops to generate anything until you have to pay for more jobs bit Journey works slightly differently because instead of buying credits you buy GPU minutes so rule of thumb is that one GPU minute can perform one job that includes generating 4 image is upscaling an image and generating variations of an image so the price would be around 5 to 2.5 cents per job depending on the tiers you get the higher ones even offer Unlimited image generation for smaller images people choose to use mid-journey because the style that it generates has this iconic mid-journey touch on top of that mid-journey focuses on producing beautiful images so you are pretty much guaranteed to have a lot of fun generating since anything you type in will create some sort of aesthetically pleasing art and would not require a lot of prompt engineering and for the weebs nity journey will be your top choice for generating on Discord needy journey is pretty much mid-journey but anime I have a video about it you can check it out if you want to but yeah it's just very good at generating anything anime related and Anime keywords would work really well on there too since it's pretty much ran by mid Journey the pricing is pretty much the same if you don't have access to a laptop or insist on trying it on your phone because you don't want to use mobile webs which is totally understandable there are a lot of existing apps that can generate AI art for you too but they usually run on a policy which is an ad video for a generated image if you want to do it for free some pretty well established apps such as IB or wonder AI are all available on both App Store and Google play with payment plans that costs five dollars per week that has a three days free trial and a standalone AI Avatar package where you can input selfies and generate your own face with different angles costumes or core effects which costs around eight dollars but eight dollars for a handful of images just for an AI Avatar is really expensive and since AI Avatar is just using a method called dream boost today sponsor open r dot AI can also do this with a much cheaper price or you can learn to DIY if you have some good Hardware which would be completely free with a lot more customization and to be completely honest with you the models that they have for these mobile apps are pretty mediocre it is pretty hard to generate some good looking images and you can probably get something much better if you use mid Journey on your phone through Discord at a much cheaper rate and you'll always get some really good AI art but there is also another online method that was a lot more popular back in the days and it pretty much raised text to image AIS to where it is today you might have heard of it but it's called Google collab notebooks Google collab is basically an online code running service provided by Google and includes a GPU that you can use for free with a certain amount of computation limit well you can overcome that limit and get better gpus by paying but that is definitely not ideal when your budget is literally zero dollars and technically you will have infinite free usage and it's just that you need to keep on creating new Google accounts to get new GPU time while waiting for your other accounts to get the free GPU time back for legal purposes that was a joke buy Google collab subscriptions instead anyways there are a lot of Google collab notebooks you can choose from to run a lot of different things ranging from generating images to fine tuning AI models creating embeddings or creating latent space interpolations or even dream Booth AKA AI Avatar some of these terms may be new to some of you but you will eventually learn about them when you are deep into the rabbit hole Google collab is pretty much the alternative of getting all the functions a local PC users can do since it is a service that acts as a cloud computational device but not a polished text to image service the downside for running Google collabs is that it's much slower than other services or running on your local machine because the free TRS GPU is not as powerful and you need to set up the environment every single time however the gpus that they provide have very large vram so some people use collab to train or to run some stuff because even though their local GPU might be slightly powerful the vram that the free GPU has is much larger the hard part for running Google collab is to find a notebook that has the function that you want whether it is text to image or some funky functions for an AI model there are tens of thousands of notebooks out there because anyone can create or modify a Google collab notebook and share it to anyone there are lists of Google column notebooks that people curated so I will leave that down in the description disco diffusion is kind of ancient now can't believe there's a date I would say that it was a Google collab based techsoup image project and it was the thing and based pretty much off people editing and sharing Google collabs but if you have a powerful GPU Google collab would never be a good option for you for running AI image generators running it locally on your PC would be your best option if you have a Windows machine with an Nvidia graphics card that has eight or more vram running it on your PC is pretty much the gateway to the Deep rabbit hole that is text to image AI the current best or I would say the popular way to run it on your local PC is using this GitHub repository that is made by automatic111 called stable diffusion UI it is a web UI that is supported by a very large community so whenever there's a new function that research proposes or a new model that got released this stable diffusion web UI which I just commonly referred to as automatic would support it in a matter of hours compared to online web services or Discord Bots running it locally provides you with an option of running any AI model that you can possibly obtain and pretty much completely for free so it's just the dream Studio but free and runs locally with you having to go out there and collect AI models like Pokemon cards and sometimes you might have to set sail the upside about this is that you can download any models that someone posted online and run it locally on your computer this includes stable diffusion models whether if it's an AI model that fine-tuned on generating a specific celebrity or generating in a very specific R style you're pretty much free to do or customize anything and use textual inversion apply hyper Network or fine tune with stream Booth to get the same effects like AI Avatar web services just don't have the power and flexibility to do all of these at the same time running it locally is like getting a whole workshop with all the tools available to you and you can make whatever you want by yourself and apply whatever skills or style that you learned while using online services to generate images is like getting a core printer to make something for you and any work that they create will be made out of wood and in their style which is pretty restrictive and you have to pay to set up automatic on your PC is pretty straightforward and you just need to download python download git download the codes and download a stable to Future model then run this batch file and you'll be all set if it's your first time you'll find it to be pretty difficult to generate some nice looking images it's harder to generate beautiful images not because the models are less good than mid Journey but because you have more control so this is where prompt engineering comes into the picture if you want to learn or see how others generate you can go to Discord or Reddit to check out discussions made by others and I suggest to check out the official stable diffusion Discord and they're ready for some very high quality posts or you can check out today's sponsor openart.ai which is a website that contains a huge library of air that people have previously generated along with the text prompt that was used categorized by the AI gen generator that it was made from with the addition of the creation section that lets you run Dolly 2 and Sable diffusion models online with a much cheaper price per image compared to their official website oh they actually just added new models in so you have more choices to pick from now other than stable diffusion and Dolly too for Dolly 2 if you buy the 15 package or higher on open art it's a 0.25 cents cheaper per image and the 30 package is straight up 1.25 cents cheaper if you think it says don't worry they just use the official API and run it on a loss also openr has this photo booth function where it is exactly the AI Avatar function that the photo apps have which provides you a way to use dreambooth as a service online while AI avatar on phone is limited to humans or animals and bad models open arts photo booth can also learn art style unless you choose from a range of different base models this will guarantee the quality and can be great if you have a specific output image style in mind the pricing is better than eif Parts on phone apps too and you can also buy in bulk and generate up to 400 images so check out openart.ai if you're interested with the link down in the description and thank you for watching a big shout out to Andrew lascellius Chris LeDoux and many others that support me through patreon or YouTube follow my Twitter if you haven't and I'll see you all in the next one

Channel: bycloud

Views: 36,418

Keywords: bycloud, bycloudai, how to run ai art, how to get started with ai art, beginner ai art, ai art beginner, how to ai art, how to use ai art, how to make ai art, how to make ai art for free, how to generate ai art, ai art, generate ai art, how to generate anime ai art, how to generate ai images for free, how to generate ai images, introduction to ai art, how to generate ai anime art, ai art how to, ai art price comparison, ai art service comparison, ai art services

Id: zxnXpSfoXhQ

Length: 13min 11sec (791 seconds)

Published: Fri Feb 10 2023