Ultimate Midjourney Guide: Beginner to Advanced

I've spent an outrageous amount of time experimenting with mid-journey and I want to show you everything to take you from setting it up for the first time to generating images like these plus how to make different characters using your face hyper realistic photography creating logos understanding parameters in-depth prompting techniques everything you need to know and more before I jump into the setup here's a quick example of just how easy it is to get started last night I was messing around with just single word prompts you just type a concept and see what it comes up with did a bunch of these microcosm was probably my favorite or you could add something simple like RGB it'll make it colorful or bump it up to three word prompts you can also prompt with Emojis which is pretty fun another just fun tip is it's great at replicating Celebrities especially version five you probably saw a dope Francis go viral recently you get the point it's really easy to make cool stuff but we want to have more control so let's start from the beginning the only way to access mid-journey right now is through Discord so if you don't have a Discord account just create one it's free and easy then go to midjourney.com click sign in then authorize the Discord bot the free trial is currently disabled and may or may not be available as you're watching this video they have a few different pricing options the ten dollars per month is good for most people now I assume you're not in as many Discord servers as I am but this black and white one is the mid-journey server over on the left you'll see these newcomer rooms you can type your prompts in the Newbie chats but everything will be public and it's hard to keep track of so I recommend starting a new Discord server and inviting the bot if you want to click the plus button on the bottom left then click create my own and name your server then come back to the mid-journey server find the mid-journey bot over on the right click on it then click add to server now this is your own workspace where you can stay organized and create and private and if you want to get fancy like mine you can right click over here and create new categories and channels there are two commands you'll use to interact with the bot they each start with a forward slash the first is settings you don't need to pay attention to most of these except turn remix mode on this will help when modifying versions of the images you get and then right here is mid Journey version 4 and version 5. they do different things better but everything I'm going to talk about will apply to both then the only other command you need to know is imagine that will pop this up where you enter your prompt now just type any anything at all that you want to see I'll start with something simple easy as that it'll start generating just give it a little time to finish then you'll see some images that are pretty cool it'll provide you with four variations each time the U stands for upscale this bottom left is number three if we click u3 it will upscale which gives us a high quality version then the V is for variation let's say we like this one but it's not quite there we'll click V2 and this is where that remix comes into play we can use this as our reference image and adjust the prompt and the blue button just re-rolls it completely giving you four new images from The Prompt without referencing any of the previous images to get more specific results you can add tags to The Prompt separated by commas one of the most impactful choices you'll make is style so I'll use the exact same prompt as before fox in a mountain landscape then add a style tag at the end or instead of choosing a style you can write a specific artist's name and it will replicate their Style [Music] or you can try more unusual mediums it will recognize some pretty obscure stuff [Music] the first two incredible free resources here the first is mid Library dot IO this is an extremely comprehensive reference guide it shows you mid-journey Generations from different artists and artistic techniques it's really easy to navigate it also has amazing in-depth blog posts next is this GitHub which has a massive repository of ideas and testing it doesn't have as aesthetic of an interface but it's still really easy to navigate the style guide has reference Generations from different styles artists perspectives lighting colors materials camera lenses then beyond all the Styles there's things like testing and cross analyzing every minute detail like waiting keywords quality chaos and stylize cross analyzes really these are both invaluable resources can't recommend them enough you can learn a ton from all these comparisons and just get inspiration but moving on to parameters so they all start with a dash dash the most straightforward parameter is aspect ratio that's the width to height ratio they default to squares which are a one-to-one aspect ratio but you can change that to essentially anything you want some common ones would be 16 9 that's the ratio for TV and computer monitors YouTube thumbnails flipping that to 916 would replicate a picture from a smartphone like if you're mimicking a selfie three two and two three is common for photography you get the idea just type dash dash AR space then the aspect ratio you want the next parameter option is stylize using S and then a number between 0 and a thousand this mimics how strongly mid Journey's default aesthetic style is applied but low stylization values produce images that closely match the prompt but are less artistic then High stylization may be less connected to the prompt but very artistic there's a few more parameters it may seem like a lot at first but once we go through some examples it'll all make sense so just stick with me Q is for Quality you can choose 0.25 0.5 or 1. this affects how much detail it adds the higher numbers will also take longer to generate it's not affecting the resolution of the image it's how much detail is within it chaos uses numbers between 0 and 100. this changes how varied the results will be to higher values produce more unusual and unexpected variations which would make all four generations done in very different styles you'll use seed to reference a previous image you generated to get the seed you need to react with the envelope emoji and we'll DM you the seed then you'll copy that and paste it into the parameter so that will use that image as a reference and produce new generations that are similar to it the no is for negative prompting so no plants will try to generate an image without plants and this can be hit and miss sometimes then related is prompt awaits and multi-prompting adding colon colon to a prompt tells mid-journey to consider each part of the prompt separately so if you just type hot dog you'll get images of a hot dog but if you add the two colons it will treat them as two separate thoughts and you'll get images of a dog that is hot then adding a number after the colons will add relative importance to that part of the prompt so if we were to add a 2 here the word hot is now twice as important as dog so the result is a dog that is very hot there's one more parameter that only works on version 5 that's tile it produces tileable images that are perfectly seamless so you can use them as patterns this is pretty amazing actually I know that's a bunch of information I'll have a short cheat sheet down in the description description to help remember it all a lot of the time you won't even use parameters outside of aspect ratio or you'll only use one or two at a time they're important to know also in a minute I'm going to show a tool that does most of the work for you but let's do some practice first then we'll move on to prompting with your own images hyper realistic photography is one area where mid-journey really shines especially version five I want to go through some photography examples so a really easy tip is to just give it the name of a photographer that shoots in the style you're looking for like an American Soldier photographed by Robert Frank or a lion Photograph by Paul nicklin it already knows what to do with that I added the aspect ratios of two three and three two because those are the default for DSLR cameras and also it gives it more room for interesting compositions when it's not a square but if we want to get more precise and customizable this is a good format and you can just delete any of these fields you don't feel like using but it's a good base to work from one of the most impactful Fields there is just naming the type of Photography you're looking for so here's some examples using the same prompt and only changing that field lighting has a huge impact on the mood you can go down the list and experiment with these each field will impact the outcome in different ways it's fun to mess around with the more unique types of Photography like macro or underwater or satellite as you do this and make adjustments it helps to try to think like a photographer think about what emotion you're trying to capture how the pose angle and lighting will affect the mood you can really deep dive here and create some incredible stuff Nick St Pierre's Twitter is a great resource to learn more a couple of the photos he's generated have gotten some press and he has in-depth tutorials on his process and various experiments there's also this amazing color guide in the mid Journey Discord that helps invoke colors into specific areas like eyes or clothing without getting those colors in other places which is a problem you'll run into but let's go to kind of the opposite end of the spectrum and create a logo the most important note here is that mid journey is notoriously bad at creating text so if you want text on your logo you'll have to add it later in Photoshop or canva or photo P whatever you use you can just type in create a logo for this type of company and see what it comes up with then just regenerate until something good pops up these are all just from The Prompt artificial intelligence logo there's some solid ones in there but a good way to give it more guidance is to use the name of a logo designer you like Milton Glazer usually comes back with some great stuff here's a few other designers I liked you can also guide it with types of logos if you have something specific in mind it's pretty crazy how easy it is to create a great logo in here if it needs some touch-ups you can use vectorizer.ai for free that will turn it into a vector file that you can edit in a vector editor like illustrator you can also inform mid-journey with an image to work from to replicate a face or environment so I'll start with an image of my face first you need to get a link to the image the easiest way is to just send it in Discord so just click Plus on the left side of the text box select the image on your computer and send it then you'll right click the image and click copy link then paste that URL at the beginning of your next prompt and it will influence the jobs composition style and colors there's Endless Possibilities here these are all just styles of anime here's some other cartoon styles then I tried it out with different time periods one I thought was just amazing was pleistocene these don't look as close to your face as they would if you trained a stable diffusion model but it's still really fun it will come up with different things depending on how you phrase it and which version you use this is in Tim Burton's Style versus as a Tim Burton character here are some more unique ones I tried out so yeah a ton of fun and by no means the only thing you can use image prompting for here's a photo I took that I want to make look like an oil painting there is one more parameter that's Image Weight you use IW and that will allow you to control how closely it tries to mimic the image you uploaded because image weight at two it looks great then as I lower that it gives mid Journey a lot more artistic license that's the only text I gave it was oil painting so it'll start to add random objects and people because it's drawing from this whole Bank of oil paintings you can also combine two or more images you can use the command blend to do this as well but if you want to add instructions stick with imagine then you can write in your text and parameters this definitely takes some experimentation to get the hang of here's me combining my dog with a photo I took of a moose there's tons of possibilities here so now that we've practiced and you have some understanding I'll show you a couple resources that can do a lot of this work for you prompt.newshot.com is great you just type in your main idea then click to Styles lighting depth of field element you can adjust the parameters upload an inspiration image if you want add negative prompts here then it will generate the entire prompt for you and you just copy and paste it helps a ton just to save time and also to get ideas a new feature they rolled out is the describe command you can upload your own image then it will give you four prompt ideas back using what it detects in the image and you can edit those prompts as you submit them a lot of times these will turn out better than the original also it's just a great way to get new prompt ideas it's also pretty fun to upload a picture of yourself and see what it comes back with let's see if it can make a better thumbnail than me not bad Twitter's a great place to get prompt ideas here's a few of my favorite Twitter follows I create life posts some prompting challenges to participate in and just all around great stuff Chet Bliss has been posting a ton of unique ideas these paper dolls by Anonymous are amazing Linus eckenstam posts good mid-journey and all-around AI content another great method is just to scroll through the explore tab on the mid Journey website this is a really cool one I saw if you scroll down there will be related images that shows the full prompt here and the parameters down below you can click copy full command and it will copy the entire prompt and parameters then you just paste it into Discord and generate you could also try saving the image then using the describe command then use that as inspiration for your own prompts let me know in the comments if you want me to do more mid-journey videos to Deep dive into specific Styles or Advanced prompt structures and I'll see in the next one

Channel: Futurepedia

Views: 249,683

Keywords:

Id: q1FIh6uav68

Length: 11min 54sec (714 seconds)

Published: Sat Apr 08 2023