How To Use Midjourney - From 0 to PRO [2023]

If you're fascinated by AI's ability to generate amazing pictures and want to get started, this tutorial series is for you. In this multi part video series, I'm going to teach you how to go from an absolute beginner to advanced user. Along the way, I will teach you tips and tricks to create absolutely amazing pictures with Midjourney. In this first video of the series, we will cover the basics of Midjourney. So we will look at how to create the Discord and Midjourney accounts. Then how do you invite Midjourney bot into your own server, even as a free user? Then we will look at the basic prompts. We'll generate our first few images and understand how the image creation process works in Midjourney. Then we will explore all the settings. So that will include all the models and different options that are available within the settings. We will also look at Mid Journey specific codes or parameters and their impact on the image generation. So in this first review we'll look at all of this and a lot more. Throughout this series, we will be heavily using Mid Journey's documentation because that is the best place to find the relevant information. So let's get started. Step one. We need a Discord account. Mid Journey is using Discord to offer their services. So first click on Open Discord in your browser. Let's give it a name. So I'm going to call it Test User. Click Enter. Then you have to verify that you are a human. The verification is done. Now I need to give it my birthday. Do something. I create your own Discord server. So there are different options. I'm going to click on Create my own Discord server and then you can select what it's about. So I'm going to click for me and my family. And let's keep the name like this, right? Start a conversation. I'll just skip it. Now we need to actually provide an email address. Simply provide your email address and then put in a password and click on Claim account. Okay. So it will send you a confirmation email. You need to go to your inbox and click on the confirmation link. So I have verified my email address and now here I am in my server that I created. And you will see the user ID as well. That's your user ID and the username. You can invite other people to your server by clicking this invite link. I will do this later on. Okay, now step number two. We need a mid journey account. So go to midjourney.com I'll have a link in the description. Now here you need to Click on Join the Beta. So if you click on Join the Beta, it will take you to their Discord server. So there are a whole bunch of people in the in their Discord server. So click on Accept Invite. Now you have to verify again that you're a human. So it's asking me to click each ball containing a soccer ball. Okay, click on this icon. It's the Mid Journey server and it's up to you if you want to select your pronoun and location. I'm going to just skip it. Okay, so if you're not familiar with Discord servers, let me walk you through this. So first you will see on the left hand side they have different channels. So for example the announcement channel where they actually announce different features or different changes. So it's a really good place to keep yourself up to dated with what's going on with Midjourney. Then there are recent changes. That's another channel so we are interested in is getting started. Before that you can check out different the rules channel. They have different rules and community guidelines that you need to follow. Next let's check out this getting started. So you will notice that for a free tier you can get up to around 25 queries, which is basically you can give 25 different prompts and then in order to generate more images you need to buy the full membership. So let me show you what it looks like. So they have these four plans. First one is the free trial, then basic plan, standard plan and pro plan. So it can go from free, which gives you only 0.4 hours per lifetime. So that's around 25 queries or images OCM queries, not images. I'll show you like what's the difference? And then for the pro plan it can go up to 30 hours per month, but you have to pay around $60. So let's go back to the Discord channel. So how do we use this? So to create an image you need to go to the newbie part channels. So on the left hand side you will see there are multiple newbies channels. So you go to one of them, then you type slash imagine and whatever you want. So it's a text to image generator. In order to generate image an image you first need to type slash imagine and then your prompt right? And the bot will create four images for you. So one of these commands, the slash imagine that is going to be counted as one query. So before heading over to the newbies channel, there are a couple of other channels that I think are worth exploring. So for example discussion different people are talking about different things. Then there are people who are sharing thoughts about the prompts in the prompt channel channel. And then they have daily themes. So people create images based on daily themes as well. Okay, so let's create our first image. For that we go to this newbies 108. You can click any one of them right now here you will notice that you see a lot of images being generated. These are different users on the feed here who are just running prompts and these images are being generated. So you need to look at this mid journey bottom right here is a prompt from a user. For example, that's my LP is running a prompt, right? And the image is being generated in front of everybody else. So for the free tier, the image that you generate is going to be shown in this channel. So everybody can see it. There is a way around it. I'm going to show you in the next step. All right, now in order to generate an image, so first you need to come here to message, slash, type, slash, imagine. Then this will pop up, just click enter and you need to give it a prompt. So the prompt is basically one or more sequence of sentences that you want to create. So anything that you can imagine. So let's say a simple prompt I'm going to try is a wolf wearing sunglasses. That's my prompt. Okay, now you type in, you click enter. And since I'm running this for the first time, so it is asking me to accept the terms of services. So simply click on this and then the bot will start generating your image. Okay, now here like you need to actually come down to look at your own username. But this is a process of it generating images. So you see the progress. You will notice that it's really hard to figure out. You have to scroll up and down to see your own image. Right? So there is an easier way whenever the image generation is completed, right? So it will show you in the bottom or you can simply go here to this inbox icon and and then go to mentions. So it will show all the messages where your, your username is mentioned. So these are our images. I think they look pretty good just for a very simple prompt. And midjourney is actually known for generating great looking images from simple text. Okay, so let's go over some other options below the four generated images. You see this U1, U2, U3, U4 and then V1 V2 V3 V4. So the first one is if you want, if you like one of the images and you want to get an upscaled version of it like a high resolution image, right? So you will click on these U options. U stands for upscale. And the way it's arranges, the first One is marked U1. Second one, sorry, the first one is marked 1. This is 2, 3, 4. So you will select the corresponding U. For example, if I want to upscale the last image, so I'm going to click on U4. The next option is this V1. These stands for variations. So let's assume if you like one of the image and you want to generate four different images with a very similar style so you can click on whichever image you like. So for example, in this case I'm going to. Let's say we want to generate images similar to this image too. So you can click on V2, right? And the bot will start a new job. Okay, one more thing. You will see this icon. This is just to reroll. So for example, you want to rerun the same prompt so you can simply click on this icon and it will regenerate your images. All right, so usually the way I here are like four different variations based on the second image that we selected. Now just to show you a practical example of upscaling, so let's say I want to upscale this second image now. So I'm going to click on YouTube. So let's try to find our upscaled image. Here we have it now if you click on it and click open in browser, you will see the upscale version which is a much higher resolution image. Let's look at what are these options in the bottom. So first these emojis. This is basically the feedback that you can give to the Mid Journey bot. So like if you liked it, so you can tell it that you love the image and so on and so forth whether you don't like it. That way they can get a feedback from the users to improve their algorithms. Now this option make variations if you click it. So it will simply create four more images based on this image. Now we have Light Upscale Redo and Beta Upscale Redo. These are two different types of algorithms they're using for upscaling this image even further. So beta is, I think they are testing the latest algorithms that they have. Now one thing you noticed is you have to really scroll up and down to see all your generated images. One way I showed you was clicking checking your inbox. But that stays this there for only seven days. So like after seven days you're not going to see these messages right. Now you can check them out on the Mid Journey website as well. So Click on this web link. It asks you that you're leaving the channel. So say yes. All right, now it brings you back to the Mid Journey website. Okay? Now here, if you click on sign in, it asks you to authorize Mid Journey Bot to access your account. So authorize it. And now it brings you to this account if you click on home. These are the images that we have generated so far, all right? So you can see them. There's also a community showcase. So if you click on it, these are different images that are generated by the community or different users in Discord. So they pick some of them and show them here. So one thing is, if you click on them or like bring your mouse over it, you will see a part of the prompt, right? So you need to be a paid user to see the whole prompt. So it's, it's. It's a really nice feature to get inspiration from other users. Now let's go back to our Discord server. So if you generate a lot of images, it's really hard to keep track of where your images are being generated. And you get can like you'll have to go and scroll up and down. But there is an easier way to avoid all this clutter. And that is you can actually invite this Mid Journey bot to your own private server. So in step one, if you recall, we created this server which is completely private. There is nobody else in it, right? And you can invite the Mid Journey bot to your server. And that way you will be able to simply generate images in private. All right? Now in order to do that, you need to first come here. This show member list click on will show you a list of all the people who are in this channel at the moment. Here you will see this Mid Journey bot. So right click on it. Actually left click on it. Then there's this message saying add to server. Click on this. You will need to select the server. So go choose your server here, then click Continue. Now there are these different options. I'm going to accept everything and simply authorize the bot. Then you need to verify that you are human again. Okay, so the bot is now added to our server. Okay, so head over to your own server. Now in the users list, you see two people. So that is the Mid Journey bot and your user account. Now let's see if it actually works in our server or not. So simply type imagine. And now I want to create another image. So this time a prompt I'm using is a robot robotic samurai, right? So click Enter and let's wait for the Midjourney Bot to generate our image. So it started generating the image and it seems to be working. So, so that's nice. Now you have the Mid Journey bot and there are no other uses and you don't really have to scroll up and down to look for your images. So this is great. Okay. And you have exactly the same options as before. So you can upscale a certain image or rerun the prompt again or generate different variations. So let's simply look at those. These are pretty nice images, but based on a very simple prompt that we provided. Okay, next I want to show you what other things we can do with the Mid Journey Bot. So if you simply type slash, you will see this section called Mid Journey Bot and then ask blend fast, help, imagine info, so on and so forth. So there are like a whole bunch of options which we are going to go over this during this series of videos. So these are the options relevant to Midjourney Bot. The second section is the built in features for the server that you just created. So you can actually kick out or ban users if you invite more users. Right, but we are interested mostly in the Mid Journey bot. So. So let's say I want to look at the other settings, right? So slash settings. Now you see several different options here. So for example MJ version 1, version 2, 3, 4. So these are different version models that you had. The current version is version 4 and this gives you the best results. Now in order to understand all these options, I would recommend you to check out the documentations. So simply go to the Mid Journey website, click on Get Started, then come down to User guide. So here they have provided explanation. According to them, Midjourney routinely releases new model versions to improve efficiency, coherency and quality. The latest model is by default. So if you generate any image, this is going to be with the latest model. And if you want to select a specific version of the model, then you need to type in dash dash version or dash dash V. The current model is version four. And even version four has these different different models, A, B and C. And the previous models were, and they will give you different results because they are trained on different data sets with different styles. So like Even here, this 4C is the latest one. And the result compared to 4A are quite drastically different. Now I want to show you how far Midjourney has come and how much it has improved from let's say version one to version four. So in order to use a specific version, let's say we start with version one. Okay, so you click on version one now your prompt is going to be using this one version one. And you will see this dash dash v1 as a suffix to your prompt, right? So let's say imagine we type in the same prompt that we had before, right? And let's run it now. You will see apart from the prompt that we provided, it added this suffix dash dash v1. Okay, so here are the results of using v1. So they are really cartoonish compared to what you can get with V4. V4 gives you a much better results in order to show you how quickly we have made this progress. So V1 was actually released in February of last year. So that's the second month 2022 and till April it was the default model. Then they released V2 and then V3 and V4 more recently. Let me for the sake of completeness, let's check the same model or the same prompt with V3. So let's click on V3 and I am going to type in slash. Imagine our prompt, Click enter now v3. This indicates we are using version 3. Okay, so here are the results for v3. They are also drastically different from what we have today. So it can give you an idea of how much progress has been made. So apart from these four models, there is this Nichi mode. So. So let's see what it is. So if you scroll down, there's this NII model. So NII Model is a collaboration between Midjourney and Spill Brush, tuned to produce anime and illustrative styles. So this is basically a model that was fine tuned towards producing anime and you can use this on your work. If you are working on more anime style images, the next two options are MJ Test and MJ Test Photos. These models are basically test models. Sometimes Mid Journey want to experiment with new models and they simply release it. So those are basically the test models. So if you look at it, Test model Occasionally new models are released temporarily for community testing and feedback. And there are currently two models you can access this Dash S, Dash ST or Dash Dash Test B. The next three options, half quality, base quality and high quality, they are related to the quality of your output images. So if you recall from this subscription plan comparison, you get a certain amount of time on gpu, right? So if you want higher quality images, it's going to be using more time and the cost is going to be higher, right? So that's why it's like high quality two times the cost if you want low quality images. So it's simply the details are going to be less compared to the High quality images. So these are different options you are going to play around with. If you need really high quality, just keep in mind that it's going to cost you a lot more. And if you let's say click this high quality and you will see this other new option that is going to appear dash dash Q with your prompt, I'm going to go this in more detail when we're talking about Mid Journey specific codes. So for the time being I will simply select base quality which is the default value. But you can check out the documentation. Actually let's go and check out the documentation relevant to this quality. So you will have to come up to the parameters and then click Quality. All right. So according to the documentation, the quality parameter changes how much time is spent generating an image. High quality settings takes longer to process and produce more details. Higher values also mean more GPU minutes are used per job. The quality setting do not impact the resolution. So for example, here is a comparison of different setting of quality. So if you set it to 0.25 you will get the quickest results, least details results. So you get the quickest. But this is going to be the least these days. And it is four times faster, which means 1/4 of the GPU minutes. Then the second option which is quality 0.5. I think this is the default value. Sorry the. So the default is value is one. Okay, so half quality is 0.5. That is two times faster compared to 0.25. Right. And then this is the base quality which is the default value of 1. So anything higher than this you will incur more charges. But the results are going to be better quality images. And all four models, actually all five models supports different quality parameters that you can give. And in order to add a quality parameter you will simply use Q or dash dash quality. Next we will look at the style. So low style, style medium, style high and style very high. So again let's simply go to the recommendation, click on stylize and according to them, this Mid Journey bot has been trained to produce images that favor artistic color composition and forms the style parameter influence how strongly this training is applied. So low stylization values produce images that closely match the prompt but are less artistic. So if you keep this the value of style to low value, then it will follow your prompt more closely. But if you keep a higher value, it will generate more artistic images and you can set the value of style between 0 to 1000. So this option is available only in V3 and V4 and V3. Right. So those two versions and then the test models in native Niji doesn't have that. So for example here are different the prompt is illustrated figs. So if you keep the value of style size to 50 so it will resemble more like a fig. But if you increase the values then it adds more artistic styles in there. Right? So depending on your situation, you can play around with it. Okay, the next options are upscaling. So if you want to increase the size of your image after generation. So if you recall we had this U1, U2, U3, U4. There are three different options. Regular, light and beta. And the best possible place to learn about all these options is the documentations. So we will go again and we'll click on upscaler. Now initially Mid Journey produces all images in 512 by 512 pixels, right? But you can use Mid Journey Upscaler on any grid images to increase the size and add additional details. There are multiple upscale models available for upscaling an image. So let's look at this. So the default sizes, like if you don't define anything, it will be 512 by 512. And there are these other model sizes or image sizes. So if you look at this table and the default upscaler will double the size. So it will go from 512 to 1024 by 1024. Now even the light scalar keeps that size, but the beta upscale actually takes the size to four times. Now let's see what each one of them does. So the default or the regular upscaler, it increases image size while smoothing and refining details. So some small elements may change between the initial grid image and finish upscale. Right. So you will see there could be like some variation from the up between the upscaled image versus the initially generated image. When it comes to the lighter upscale, it will add moderate amount of details and textures and they recommend to use it for faces and smooth surfaces. The detailed upscaler as the name suggests, it will add additional finer details. And this could be used with the this could be upscaled again to the max. But the max option is not available for all newer models. Now the last one is the beta scalar. So this actually creates the biggest images and without adding many additional details. And they also recommended to use it firm smooth surfaces and faces. In order to use the beta upscaler, you can use dash dash up beta as a parameter. There's an upscaler for anime as well. We will go over the Rizmaster next. The public mode is basically all your images are being shown in the Discord servers. If you are A Pings member. So other than the free plan, you can enable Stealth mode and your images are not going to be shared with the public. So the next option is the Remix mode. It allows you to change your prompt while you're generating more like different variations of your image. So let's look at what the documentation says easily. Remix mode to change prompts, parameters, model variations, model versions, or aspect ratios between variations. Remix will take the general composition of your starting image and use it as a part of the new job. Remixing can help change the settings or lighting of the image, evolve subjects and achieve tricky composition. So like the way it works is you generate an image, then you turn on the Remix mode, you make a variation, and then it enables you to add more details to your prompt. So let me walk you through the process. In order to get this Remix mode, you need to type slash prefer to. There are different options and we want to select Prefer Remix. If you run it, it says the message says Remix mode turned on. Click the Variation button. Clicking the Variation button will now give you a chance to edit your prompt. Right, so let's say I go back and I want to create four more variations of this last image. So now if I click on V4 now, it instead of running it directly, it gives me the ability to change my prompt. So let's say I want to add back lighting if I click submit. So it will start generating four more images with this additional prompt that I provided. All right, so here are the images. You can see there is a backlighting effect added to the image if you want to turn the Remix mode off. So you have to type in Prefer Remix and let's run it now. It says Remix mode turned off. You can always turn this on by Prefer Remix again. Okay, the next setting is the Fast mode. So it simply defines whether you want to use the GPU for generation or not. So if you don't use the gpu, they don't cost you anything. But the wait time of generation can be pretty significant. So let's say if you come here to the documentation. So subscribers to the Standard and Pro plans can create an unlimited number of images each month in the Relax mode, right? So Relax mode doesn't cost any of our any of your GPU time, but jobs will be placed into a queue, a queue based on how much you have used the system, right? And it can go from zero to 10 minutes per job. And in order to turn on and off, all you have to do is simply type in the slash fast command or slash Relax. So let's say if you want to deactivate the FAST command, I. I don't think I will be able to do it. Let me check. Yeah, so you need to be actually paid member to use the fast or Relax relax command. So on this free plan I can't do that. Okay, so this is an overview of all the options that are available in Settings. Next we will look at some mid journey specific codes that you can use to change the behavior of the generated images. So stay tuned for that.