Using generative AI images with other AI apps - Midjourney Tutorial

- [Instructor] In this video, we're going to look at additional generative AI tools that we can use, along with Midjourney, in order to bring our stills further to life with design, not to mention video. Our journey is going to start here with openai.com and a product called DALL-E 2, which you can get to by going to the following website, and once you get here and actually sign up for an account, you can choose to Try DALL-E. And then, from here you'll notice that it does give you the ability to, first of all, course, create generative imagery. But if we click on the three-dot icon, there's something called Outpainting, which can be very helpful with certain images that you create in Midjourney. So let's try Outpainting, and the first thing that we're going to want to do is upload a piece of Midjourney imagery here. Let's click on the Upload an image button here at the bottom, and inside your Midjourney images folder, we should see a folder for Midjourney with Other Generative Art Apps. We can open up that, and this photo here, which is all of my thoughts out in the Pacific Ocean, is what we'll be using. I'll press the space bar so that you can see it on my Mac. And let's click on the Upload button. So what this app is going to allow us to do is actually extend the dimensions of the picture by painting in areas where there is nothing. In addition, we can actually choose to erase things in the photo and replace it with other content. Let's click on the checkbox to place the photo, and you're going to see a frame box that you can move around, this blue frame box, which will allow us to now generate a new frame. I'm going to allow for a little bit of overlap here, and inside the edit field, I'm now going to type something, which is More mountains, night sky, constellations and northern lights. And with that entered and the generation frame identified, the area, I'm going to choose Generate. DALL-E gets to work, and now we can see the creation of this new generative frame. Now the cool thing is it's not just this one image that was produced. In fact, you'll see these four dot icons here, and if we use this arrow, it will show us some different variations of what it created. I'm kind of really liking this third one here. I'll choose Accept. Now the great part is we can generate an additional frame. We can just drag this box to another area here, and in fact, using the same prompt, I'll generate another frame. DALL-E gets to work, and let's see what it comes up with. Again, we can use these arrows to see the different creations it came up with. Kind of like the first one. I'm going to accept that. And now, I'm going to create some additional generative frames up here at the top. Okay, here's the end result of the image that I ended up with, and as you can see here, once you've created this, you can click download in order to select it. And here's the image, much larger than the 1024-by-1024 frame that I handed over to DALL-E. So definitely something to check out, Outpainting. If you sign up for an openAI account, you get a certain amount of credits for free, and after that you can purchase credits on demand. Another thing to keep in mind is that, if you're a ChatGPT Plus customer, you have access to DALL-E 3 directly in the chat when you're using version 4. To see this in action, simply just start a new chat, and I'll head over to the Midjourney images, the generative art app folder, and let's select the image of the ghost. So I'll select that, open it up. It's going to upload to ChatGPT. It's going to see that I've uploaded an image, and how can I assist you with it? I'm going to ask ChatGPT, I'd like you to reference this image and create a new one with the exact same ghost and frosted window. Can you also extend the new image by 300 pixels on each side showing the interior of a house? I'll follow this by sending this message. ChatGPT is now creating a new image, and let's see what it comes up with. Here we have a new image with a ghost that is definitely not the exact same one in our above image, but it does create a new image from scratch when using ChatGPT and doesn't just out paint from the original image. Shortly, OpenAI is going to make DALL-E 3 available in the same way that DALL-E 2 is, where you can out paint existing images. In addition to text-to-image apps, another thing to watch out for is text-to-video apps or image-to-video apps. One of those that you can use along with your Midjourney images is Pika Labs. You can work with it in Midjourney. However, they've just recently released a web application. You'd have to sign up for the beta. However, this is quickly changing and may not be available for free at the time of this recording. In order to use Pika Lab's login, and in this case, we want to upload an image, so I'll click on the icon here at the bottom, heading to my Midjourney folder here. And let's select the Paper Quilling image with the lion. I'm going to upload that. And now we have an option here. We can actually describe the story that we would like to see with a text prompt. And one thing to note is that we don't even have to describe the story. We can just define the movement and how fast that movement occurs within this scene. In order to accomplish this, we can go to the controls down here. Let's go to video options first and define the frames per second. We can set that to 24. I'll then click on the camera icon where we can define movement. In this case, I just want to move or zoom in on the image, and we can set the Strength of motion somewhere between 0 and 4. Next, we have additional parameters that we can add here. One is how consistent we want this to be with the text and even define negative prompts, like we did in an earlier movie, with parameters inside of Midjourney. So if you don't want your video to be ugly, comma, bad, comma, jittery, glitchy, you can enter that and then choose a consistency style with the text. I'll set this to be a very low amount. Once that's all defined, we've got the image up here, let's just click on this icon here, the star, and Pika Labs will get to work. Video has finished generating, and here we can see the image input. A nice slow camera movement. Our birds are moving in the background and some subtle movement there with our lion in the foreground. If we'd like, we can expand this image. I can also do other things with this image. I can add four seconds, Upscale, Edit, or even Reprompt or Retry if I'm unhappy with the results. So definitely take a look at Pika Labs where you can take your still images and create video from them. Last but not least, let's take a look at an app called D-ID, which allows you to work with AI-generated presenters. Now in addition to this, we can actually upload our own presenters, which have to be a still image. The general workflow when you go to D-ID, which you can try out for free, get a certain amount of credits, is to first of all, click the Add button, once you've decided to create that video, and already have your image ready. This is the image I would like to try. I'll upload that. We can see here that it's taken my image. And then, I guess you have to type scripts in here. So I'm going to type in, Hi, I am representing Midjourney. Now the price of credits is based on how much you actually type in terms of your script, but once this is done, you can click to generate a video. It will tell you the amount of credits that's required, and let's choose Generate again. And when the video is done, we'll be able to preview it in our video library. Okay, here's our video, and I'm just going to play it. - Hi, I am representing Midjourney. Hi, I am representing Midjourney. - [Instructor] And we can see there that it's actually added some animation there to her lips, not to mention her eyes and her face. Of course, we can also use, instead of their own built-in AI voices, our own voice for video creation and presentation. So there you have it: three additional generative AI apps you can use in conjunction with Midjourney.