CONSISTENT Styles for Storytelling in Midjourney

version 5.2 of my journey gave us tons of great new features but one feature has mostly been overlooked and it's far more powerful than it initially seems High variation mode that doesn't just give you better variations it actually allows you to maintain a consistent style in ways that weren't possible before now that's a huge deal for storytelling so if you're working on books Comics storyboards or anything involving a story you'll want to stick around for while I was digging deeper into what you can do with high variation mode and I realized that you could actually use it to create consistent Styles I did what I always do I did a quick check on YouTube and I bumped into this video released by Future Tech pilot just a few weeks ago he covers the exact same method and topic but from a slightly different angle which is really great and he's got some really awesome content so make sure you go check out his channel later on after watching this video so before we get started let's first make sure that we've got all the right settings I'm going to start off by entering the settings command in order to open up the settings panel and then we're going to do two things first of all we're going to switch on remix mode this is important so that whenever we hit one of the variation buttons the remix prompt window opens up and we're also going to switch to high variation mode now you might be wondering what this High variation mode actually do well let me show you real quick I'm going to enter a very basic command house nothing more and now you'll see the usual buttons that we always have below all of our image grids and the row with the V buttons whenever you hit the V button Now remix mode will make sure that you get the remix prompt and the fact that we've activated High variation mode will ensure that it will be treated in high variation mode if I switch to low variation mode then we'll be using more or less the old version of remix but we want to use the new remix also just as a reminder whenever you have an upskilled image you'll see the buttons very strong and very subtle very strong is high variation mode and very subtle is low variation mode all right so let's get started with our very first example I'm going to move myself to the side here and enter my first prompt imagine and then graphic novel illustration wide angle of deserted City dystopian apocalyptic and let's start okay perfect so these are the typical images that you would expect from this sort of a prompt but now the difference is we're going to use one of these four images in order to maintain our our style and it's kind of hard to describe what I mean by that because we're not necessarily just maintaining the medium so the graphic novel illustration but we're also maintaining the color scheme and also parts of the composition and most of all what this method does it helps us to maintain the overall setting the theme and the atmosphere of the image that we choose so what I'm going to do is I'm going to choose the image in the bottom left corner and create a new scene a new image based on that style I'm basically going to be able to create another shot that's set at the very same point in time in that storyline and I'm going to do it without mentioning dystopian or apocalyptic anywhere in my prompt so let me close this real quick I'm going to hit the V3 button because I want to work with the bottom left one that will open up the remix prompt and then I'm going to enter this prompt two infantry soldiers walking toward building that's all I'm going to do I'm not going to add any stylistic elements submit and what you'll notice is that these new images are stylistically super close to the one that we remixed they did lose a little bit of the graphic novel illustration style or we didn't mention anything regarding the deserted City nor did we mention apocalyptic or dystopian nothing about our prompt actually says that this is what the city should look like so just to illustrate my point let me show you what happens if I enter this prompt individually without remixing these images now look very different now obviously it's still a war zone the style is now photorealistic the mood and the color scheme is also different so what can we do to maintain the graphic novel illustration style that we had in our first image well building on that previous example let's simply reinsert that Medium so I'm going to go back and hit the V3 button on our original image one more time it will open up the remix prompt and this time I'm going to enter the same prompt two infantry soldiers walking toward building except that I'm going to keep the graphic novel illustration at the beginning then I hit submit and as you can see this time it retains the full style much better the color scheme and the mood are fully intact and the composition is roughly still close to the original but now you might be wondering well what's the point about this can't we just prompt these images directly well let's give that a try and let's enter imagine and then graphic novel illustration two infantry soldiers walking into building dystopian apocalyptic and yes you get roughly the same style and you also get far more diverse compositional options but they're not necessarily the same color scheme you're also using up tokens in your prompt in order to describe the mood that could otherwise be used to describe more things about the subject and the environment so it's always a trade-off between these two elements let me show you two examples that illustrate this quite well going to go back to our original image and hit the V3 button again to remix it and this time I'll enter the prompt graphic novel illustration and Cat and Mouse running through kitchen submit what you can see here now is that we've managed to place at least the cat maybe not the mouse we have the cat that's running through a kitchen and it's placed in a very similar setting as in the original image that we remixed the landscape is dystopian apocalyptic and even the color schemes very close now if I tried to prompt this directly in a single prompt without remixing graphic novel illustration Cat and Mouse running through kitchen dystopian apocalyptic then I will get slightly different results sure enough the style is similar however there's much more focus on the cat now and the scene within the kitchen is much more defined so creating the previous example without using remix mode May either require more words or it might not work at all oh and by the way if you're looking to learn how to control my journey and improve your prompting skills check out Master System journey and a whole bunch of other free stuff down in the video description okay so let's look at another example in this example I'm going to start off with this prompt 3D Pixar Animation wide angle of rabbit eating a long carrot inside a forest playful cute I'm going to use style raw simply because in all of the previous tests that I did I noticed that all of the carrots tended to have a shape similar to an egg and I'm guessing that's because a lot of the training data may contain rabbits in a context of the Easter bunny anyway so using style raw and saying long carrot is kind of my attempt to try to get an actual carrot and not an edge hit carrot so let's have a look how these turn out now and sure enough just have a look I've got all these orange eggs maybe even some orange cherry tomatoes here and something in the bottom right corner that looks kind of like a it's it's a really weird card but all of this doesn't really matter because it's not about the carrot or the rabbit it's about the style so I'm going to work off of the bottom left one again and pay attention to the overall setting the colors and maybe even some of the objects that you see in this image because that's going to be the reference point that we'll use in order to determine how well the style was actually replicated alright so we're going to hit the V3 button to open the remix prompt I'll enter my new prompt and you'll see that I kept the medium at the beginning of the prompt and I'm doing this because otherwise the results would look slightly too realistic but then I enter my prom two Badgers hiding behind a tree then I hit submit and sure enough it maintained the setting the colors and just basically the overall style we've got these two really really cute Badgers let's compare these results to what a regular straight prompt would look like I'm gonna enter my prompt imagine and then 3D Pixar Animation wide angle of two badges hiding behind a tree inside a forest playful cute now obviously this is a much longer prompt so let's have a look and sure enough this also worked pretty well however if you were to compare both images side by side you would see that these images have a slightly Greener slightly colder tone the environment is also a little bit different with slightly different vegetation one thing I will add though is that the characters look much more like badgers so why is that let me quickly show you another prompt with a different character I'm going to go back and hit remix again and then I'll enter 3D Pixar Animation gazelle jumping across Meadow submit what you can see in these images is that yes if you're not paying too much attention you might think that this is actually a gazelle and the overall style is very close to the remixed version you've got the same color scheme a similar setting and the style is overall great but if you pay close attention the face of our gazelle seems to have somewhat merged with a rabbit and the gazelle also has some very rabbit-like ears the very same thing happened with our original Badgers too so how do you avoid this well first of all let's enter a entirely new prompt imagine that 3D Pixar Animation wide angle of forest environment playful cute and I've also added a negative prompt no animals and creatures just to avoid seeing any sort of creatures or animals in the image so let me confirm this prompt and let's have a look at what they look like right so this creates four different images that basically contain the scene that we want to work with all all four of them have mostly the same style the only thing that we need to do is we need to pick something as our base image and bass style I'm personally going to go with the top left one this time so I'll go back and hit the V1 button I'll enter my new prompt which is 3D Pixar Animation two Badgers hiding behind the tree submit and you'll notice two things right now first of all the badges look much better than in our original remixed version there's also slightly less focus on the characters overalls now that's something you may or may not like but the overall setting the style the colors they're all roughly the same now granted the ones here on the bottom left they look kind of weird but we can do this with all sorts of other things so let me hit the remix button one more time V1 and this time I'll enter 3D picture animation two badges hugging each other submit and that gives us these really really cute images of two badges hugging each other and they're mostly in the same kind of environment it just looks like they might be positioned somewhere else within that Forest let me show you one more going to hit V1 again on our environment and then I'll enter 3D Pixar Animation Trio of dancing Badgers submit now I'm not entirely sure whether we'll actually get three badges but we'll see doesn't always work and yes actually these look pretty good this is turbo mode by the way this shows you just how quick turbo mode is it's also super expensive anyway so let's have a look and yeah we've got mostly three badges maybe some of them might be closer to the two raccoons I don't know but we got them dancing in an environment that looks exactly like our source image and it's all playful and cute so just to summarize what's the key benefit of using this method to maintain a consistent style well what you get is consistency in color grading color scheme in the setting that you're defining in your original image and of course the median so illustration style or animation or 3D models that's this kind of consistency that you're going to get if you reinforce it with the medium in your remix prompt as well now the disadvantages of this method is that there's going to be a far narrower scope for your compositions so whereas with a regular prompt you may see lots more variation if you remix these then you're going to have a composition that is roughly in line with the image that you're remixing so a good approach to mix that up a little bit would be to Simply create a new base style image that is roughly in line with the composition that you would like to have and then remix that one the only difficulty is that it might prove to be a little bit challenging to ensure that that original style image is in line with the other style image that you created but that's something that obviously is a process of a trial there either way I think this opens up so many pretty cool options for storytelling and I'm really curious to see what you guys do with this anyway that's it for today make sure you check out the video description for a whole bunch of free stuff as well as a course on how to control majority I hope you have an awesome day take care and keep on learning foreign [Music]

Channel: Tokenized AI by Christian Heidorn

Views: 43,687

Keywords: midjourney, midjourney v5, midjourney version 5, midjourney tutorial, midjourney prompts, midjourney workflow, midjourney learn, midjourney 5.2, midjourney remix, midjourney consistent style, midjourney storytelling, midjourney high variation mode, midjourney low variation mode, midjourney vary strong, midjourney high vary subtle

Id: lCFzMnBDqEc

Length: 14min 5sec (845 seconds)

Published: Sat Jul 22 2023