From the course: Midjourney: Tips and Techniques for Creating Images
Using external images with a text prompt - Midjourney Tutorial
From the course: Midjourney: Tips and Techniques for Creating Images
Using external images with a text prompt
- [Educator] In this video, we're going to look at how we can upload an image to Discord, followed by some text in order for Midjourney to create a new image from those sources. Now, a couple things have changed since the last video that you should be aware of. I've switched from the trial plan that I was using earlier in the last few movies to my paid plan that you see here now. And one advantage to that paid plan is I have my dedicated direct messaging with the Midjourney bot, so we don't have to actually follow the entire channel threads of producing imagery. I also have the ability to use private mode and not share my imagery with the public. So when it comes to Midjourney, so far we've seen how we can use the Imagine prompts to enter in some text, but Imagine also works with imagery followed by text. We can also use images as a style reference as well as with the very region feature, which we're going to dive into later in this course. So stay tuned for that. So the first thing that we need to do is give it an image to reference. One way to be able to do that is to go to a web link where it could see an image. Alternatively, we could upload it to Discord. If you have a paid plan, you can upload more content, but even with the free version, you have a few megabytes available to you. First thing first, let's upload something to Discord by clicking on the Plus icon in your message thread. So we can upload a file. Inside your exercise files under the Midjourney images, go into the image folder, and you should find this shot of leslieville2. You're going to select it and you're going to click open. You can see the image here, but just press the Return key, it's going to upload it. It's now uploaded it into Discord. In order to reference this image in an imagine prompt, we need to Right or Control + Click it, and then from that let's copy the link. Okay, we're ready to go. The way that this text prompt has to be structured is first of all, let's like before hit that Forward Slash key followed by imagine prompt. I'll then hit the Colon and we're going to now use that link I just copied. So on my Mac I'm going to press Command + V. That could be Control + V on your PC. Right after is now where we enter the text prompt. In this case I'm going to type in, with dinosaurs running. And I'm going to follow by pressing the Return key. Now, I want you to notice something. You can see that there's this tag at the end of the prompt, --V5. And I've enabled this on my settings which we're going to take a look at in a second, which allows me to use the most up-to-date version of Midjourney at the time of this recording. Depending on when you are using Midjourney, this could change and already be at version five and now we can see the final images. Now there's a couple things to mention here. One is that this image by default gets a weight next to the portion of the text prompt that you give it. So for instance by default Midjourney says, "I'm going to reference this image to a certain degree, and then also take the text prompt into consideration to a certain degree." In some instances you may want Midjourney to use more of the image and have less of a weight when it comes to your text prompt. And this is something that we can help guide Midjourney with if we're unhappy with how our images look. The other thing that we can do is also guide Midjourney to create artwork based on the aspect ratio of our image. So you can see here in this case, it's very close to 16:9. So what we're going to do is rerun this. So I'm going to copy here with the prompts the image link followed by the text prompt again. Simply highlight it and press Command + C. Let's use that forward slash followed by imagine. Got our prompt set up, I'll use the colon and press Command + V. And right after the word, running with dinosaurs, I'm going to add two things to the end of this prompt. So I want to press the Comma key and next we want to use these two small dashes which are next to the zero on your keyboard. Let's type in ar, this actually stands for aspect ratio. And then we're going to type in 16, again colon, nine. We're telling Midjourney in this instance to create a 16:9 aspect ratio of these images. Now I'm just going to press the Space key and we're going to add one more which happens to do image weight. The image has a number associated with how much it will reference it. For version five which I'm using, the lowest end is 0.5 and the highest end is two. Depending on what version of Midjourney you are using, which you can control in your settings and we'll get to, is going to be depending how much of a number you might want to give to your photo. In this case I want to give it almost the highest image weight possible. So I'm going to type in iw, and that value with V5 is two. Again, because I switched my settings to V5, V5 is automatically going to be tackled on at the end. Again, and these are known as commands that we've placed after our image and text. So let me press the Return key. Okay, now it's finished generating the images and let's notice a couple things. One is, we now have the 16:9 aspect ratio for each of our photos which is pretty cool. But based on the highest image weight available in version V5, we can also see it's giving a lot more consideration to the image versus the text part of the prompt. Just compare that to the one-by-one images that I have up here. You'll see variations that in some cases don't use that much of the image at all. Can see some of the light streaks here on the lower ones and the reference of the sign. But in this case, much more literal, in every single photo, we get those lovely light streaks. Not to mention an interpretation of the sign here in Leslieville. In this case I like the first image here because it's on the top left hand corner. So I'm going to press U1 in order to upscale that. And in V5, one really amazing thing is that the upscaling already happened with the creation of these images. So it's ready for you right away. And if you want to now have this on your system, just right click that and save the image. One thing that you could practice right now inside of Discord is being able to upload a different image with different text prompts and try different image weight values. So you can do an iw 0.5 being the lowest or do something in between to see how this is going to affect your final output. And also just note that this is not the only way that you can use an image reference. There is a new style reference parameter in version six that we're going to take a look at, and we can also use image references when we are varying regions, which will cover in a later video. So there you have it, how to create a image using your own reference images along with a text prompt inside of Midjourney.
Practice while you learn with exercise files
Download the files the instructor uses to teach the course. Follow along and learn by watching, listening and practicing.
Contents
-
-
-
-
Using external images with a text prompt6m 59s
-
(Locked)
Blending images2m 18s
-
(Locked)
Getting help and adjusting settings5m 33s
-
(Locked)
Using Midjourney v5 and beyond6m 8s
-
(Locked)
Text separation and weights to drive image creation4m 7s
-
(Locked)
Midjourney parameters for lighting, image size, and more6m 15s
-
(Locked)
Midjourney parameters for video, seamless textures, and more3m 59s
-
(Locked)
Using Style Raw and the Stylize Parameter3m 46s
-
-
-
-
-