CONSISTENT Styles for Storytelling in Midjourney
Vložit
- čas přidán 15. 06. 2024
- 📙 Midjourney COURSE mastersofmidjourney.com Beginners to Advanced
🚀 FREE Midjourney Cheat Sheet tokenizedhq.com/freebies/mj-c...
🔗 FREE Supporting Material tokenizedhq.com/freebies/vide... Folder: 2023-07-22 - Consistent Styles
🔗 FREE Promptalot Extension promptalot.com/extension
Check out @FutureTechPilot for more great content and office hour recaps for Midjourney.
-
📰 AI Newsletter 👉 Your Inbox tokenizedhq.com/newsletters/ai
🐤 Follow me on Twitter / chrisheidorn
💼 Follow me on LinkedIn / christianheidorn
📸 Follow me on Instagram / christianheidorn
💬 Join the Tokenized AI Discord tokenizedhq.com/invite/discord
-
📺 Recommended Related Videos & Playlists 📺
Commands, Syntax & Parameters: • Commands, Syntax & Par...
Prompting Methods & Experiments: • Prompting Methods & Ex...
Character Design Series: • Character Design with ...
Storytelling: • Storytelling with Midj...
Watch Me Prompt: • Watch Me Prompt in Mid...
This Brand Does Not Exist: • This Brand Does Not Exist
Midjourney for Merch: • Make Money with Midjou...
-
Last week, I was diving deep into some experiments with remixing in both Low and High Variation Mode.
I discovered some really fascinating workflows that I can now implement in a lot of the advisory work that I do for my clients.
But one thing really stood out for me and surprisingly enough, barely anyone has talked about this.
Yes, it's fairly obvious that High Variation Mode (aka "Vary Strong") has given us a tremendous amount of flexibility in creating variations of our scenes.
But I've also come to realize that it's an incredibly powerful tool for maintaining stylistic consistency if you're working on a bigger storyline.
You can now create stylistic image templates that then serve as source material for endless derivatives of pretty much anything you need.
So let's say you're creating multiple shots that all take place at the same point in time and location in a story.
Using High Variation Mode you can create each one of those shots, embed different characters in them, and still keep the visual style consistent.
This is huge for anyone who's working on books, comics, or storyboards, and I'm super excited to be able to share this with you today.
-
⏰ Timestamps ⏰
00:00 High Variation Mode's Biggest Secret
00:58 Activating High Variation Mode
02:05 Creating the Style Template
03:15 Creating Derivative Scenes
06:35 Example 2: Style Template
07:58 Example 2: Derivative Scenes
09:54 A Better Approach
12:17 Benefits & Limitations of Technique
#midjourney #midjourneyart #midjourneyv5.2 #remix #consistentstyle #aiart - Věda a technologie
Midjourney should come with an option where you can "lock" certain aspects in a project. Style, colorpalette etc. and leave everyhing else open with the bulk data it uses.
YES! That's what I though after my first generation in MJ. "LOCK" characters, "LOCK" style and create prompt of environments only. Or movements. That would be awesome!
Should be implemented right away.
Midjourney could make all of our lives better if it would add two arguments --cc (for consistent characters) and --cs (for consistent scene).
Not as simple as you think.
Excellent workflow to maintain consistency.
Thanks Christian.
+1000. Thank you. Future tech pilot rocks too. 😄
I never thought of this! Very helpful overall, thank you
I've been working with an author of children's books (we have 2 in print now!) and although I'd been using some ways of consistency, e.g. characters, the style and environment wasn't always quite as coherent as I'd like. This will make it sooooooo much easier and better :) Thanxalot !
My pleasure
I want to do this. Are there any other methods or videos you would recommend for getting the right scenes and consistency in the styles?
Thank you for posting and sharing this. Very helpful and informative.
Very helpful, concise video. Thank you, Christian!
Thank you for chaptering!
Thanks for the great tips!
how did You get those "copy prompt" buttons etc?
It's from a Chrome extension I built. Google "Promptalot Helper" and you should find it. It looks a bit different now because I added a bunch of features in the past few days.
Thank you❤❤ GREAT VIDEO!
Finally ! I understand the trade off between Remix and prompting with seed and image links!!! Thanks. I was just not getting remix mode.
Awesome 🙌🏻
really great work you show here,
I happened to do the same thing before someone made the videos on consistency.
I was very surprised by the quality and it works really well
You can get some nice consistent environments, nice work. I also use image prompts by choosing the one you like best and then applying it with a lower image weight and you can get different camera angles with the same environment and style. It's nice when you have a lot of scenes in the same area - and the new photoshop generative fill can also do wonders with changing details and keeping the same style.
Yep, that helps too, but kinda depends on what you're trying to create.
Great video! Thank you very much. Another challenge is to put two or more different characters in the same scene. For example, two different animals doing specific activity or action.
I want to solve this problem too，I want a picture of an goat and a wolf ,but it doesn't work
@@yoyolittle3690 The only way currently, is to create these characters separately in a white background. Then add them all together in any scene.
It's a great tip. I'd say this has also the advantage that you can "split" your composition into steps: you first compose the environment and then place the actors, or you first focus on the actors and then do the environment. That allows for some flexibility, which was previously possible with weights but usually was very tedious to achieve. Basically this is a great technique for getting more predictable shots, and the step-by-step prompting is just closer to how people think. :)
Yup, it also avoids having to assemble stuff in an image editor. It has its limitations, like everything, but it's a great technique.
thank you!!!
Cheers dude saved me some searching!
Pleasure
This consistency worfklow on remix mode works well also on the zoom , and pan features. Great when u want to just have a little bit more image resolution and still get consistency in the expanded image. Thanks for the video. Hope to see more awsome videos from you..
Thanks ❤
Really useful advice with using V for background first and only then including characters on that new background. Otherwise you get hybrids, which is also cool for certain projects.
That's is interesting. May help me with my comic strip because I could not get a consistent style. Thanks!
Brilliant tips here thanks! I didn’t know about this added feature to remix mode. The issue I’ve had though is getting the characters similar through out a story.
So in your case when you had two badgers you really liked how do you then have them same look in a different scene? The video seemed to be more about keeping a similar background more so but I don’t think the badger style was similar in some of the images
Thanks Chris ! As always, an extremely clear video, and on a topic of interest !
I was wondering if we can combine this new technique, with the techniques proposed to get consistent character (image prompt with your character when you remix your base image ?)
You can combine anything and everything. But it doesn't mean that the workflow will always work in every situation.
Great❤🎉
Very interesting. How would you take those characters now and insert them into a different scene??
Is there an update to this? More. consistent storytelling in midjourney. Thanks!
This is great, i really appreciate the content. When story telling, what if you want to be consistent with characters but the setting changes. ?? Is there a video on this? Thanks in advance
Start with your character and then pan and zoom.
Thanks for the videos!
little tip if I may: try to find a way to show the added value that your other separate courses have. I have noticed the little "ad" throughout the video, but it didnt convince me with anything extra compared to what im already seeing in this YT video. Keep up the good work Sir!
Appreciate the feedback. I do address this on the sales page but perhaps just not as much as you'd like.
That being said, my background is in Marketing and Sales and it's absolutely normal that the vast majority wouldn't necessarily be convinced. In fact, a conversion rate of only 1% is already considered very good in online marketing. In my case, it's a fair bit higher than that, so the offer must be reasonably compelling.n🙂
I get where you're coming from but based on my experience, you can't actually convince anyone of anything they don't already want to do. People are either already looking for something and are thus predisposed, and then you just need to address their concerns. Or they just aren't and then it doesn't really matter what you say.
The reason I put in the ads is simply because it's a numbers game and you need to get your product in front of as many eyeballs as possible.
I understand. I was just trying to be a good viewer and give you my honest feedback. Interesting perspective on makreting.
And I appreciate it and that's why I thanked you for it 🙂
Contrary to popular belief, nowadays marketing and sales are 80% data-driven disciplines. Numbers don't lie. Everything else is anecdotal evidence. You may already be familiar with this, but if not, I recommend checking out the "law of large numbers".
@@TokenizedAIAnd the rest 20% is the personal style? Yes, I read about the law of large numbers and I agree. Now that we have actually so much data on those eyeballs looking, there no room for personal interpretation on how online marketing works.
Not sure I'd say personal style. There are many tried and true principles in marketing and they are immutable laws. Branding is also super important for certain products.
The marketing (and especially advertising) industry is saturated with creative people. I love creative people and their work. However, far too often their focus is on the creative element and not on how something translates into sales. Historically, it's been incredibly difficult to quantify successful marketing. So they come up with fuzzy terms to justify things. The number of marketers who actually understand direct response marketing (and this goes back to the Ogilvy days) and math is miniscule. Far too many people do "marketing" but haven't got a clue about how sales works. They are inherently linked. Marketing is an enabler for sales. But sales is always a numbers game.
If I may offer an analogy for the point I was originally trying to make. I can use a fishing rod trying to catch a single fish, desperately trying to hook it with my "bait". Or I can use a fishing net in a stream where lots of fish are already swimming. Both strategies cost me the same amount of time, but one will yield far more results.
Sales copy is important. And addressing concerns such as the one you mentioned is very important too. But in the end, in 95% of the cases that are truly relevant, it's not my copy that convinces the buyer. It's social proof. So, buyers of my course (peers) who were kind enough to leave a testimonial. A happy customer is far more powerful than any "feature" or "benefit" in a sales pitch.
I could go into lots of detail on why my CZcams channel isn't necessarily a replacement for a well-structured course. But the point is, someone who feels that they can already get everything they need for free, isn't looking to buy a course to begin with. There's no obvious "pain" that I could ever fix for them.
My CZcams channel is a place where I share my findings in use case driven style. I record whatever I want and I only update the stuff I want to update. I give away A LOT for free but that's just my way of paying it forward and establishing trust. It also makes sure that people actually see what I have to offer. But my channel doesn't give you holistic training in all aspects of MJ. Heck, it's not even clear what is still relevant or in which order you should watch something. Autodidactics don't necessarily need my course. But many people will do much better with it.
Hello Christian. Thanks for useful video.
Is there any difference in what order to specify certain descriptions in prompts? For example, if we take the following order: "first designate the palette and content of the background, then the style of the author, and finally, the description of the main character" - will Midjorney cease to consider this character the main object?
Order matters, mainly because words at the beginning of your prompt have more weight than those at the end. There are best practices out there but it always depends on your specific objective.
Generally speaking, having A structure is better than having none at all. Most people have none.
@@TokenizedAI Ok, thank you.
How to get consistent styles in stable diffusion using automatic 1111 ?
I don't know
The image at O:17 is too good. How to create that one?
You can use now blend
You could always use blend. Not sure what you mean.
I'm very new to this! But when I type /settings, nothing happens....
Does anyone know what website lets you upload multiple images and then just click a button to get a blending animation between each image in one video? I saw a youtuber show it a few months ago and can't find it.
Sorry, I don't know.
What does the 'Add to Promptalot' button do? How do I get it and the 'Copy ...' buttons? I did a search on CZcams and nothing came up.
It facilitates adding a prompt to your prompt repository on the Promptalot website. Mind you, I'm still building it.
Thanks Chris. That's very helpful. What difference, (if any), would you expect if you upsized your initial style image and used the "Vary strong" instead of the "V" options. I'm really confused about the difference between these two options.
There is no difference. The V buttons change behaviour based on your settings. But they wouldn't do anything differently than what you described. It just saves you the hassle of having to upscale every time. You would only upscale if you want to do a "Vary Subtle" once-off without having to switch the entire settings again.
@@TokenizedAI Gotcha. Many thanks. I love MJ but I find them next to useless at providing adequate manuals on how to get the most out of the program. Thankfully, good folks like yourself fill that gap. They should be paying you a consultancy fee.
@@TokenizedAI To be honest, I'd completely given up on using the "V" buttons because the results were consistently disappointing. I guess I have to learn to trust it, now they've updated things in V5.2.
@@morpheus2573 Their Prompt FAQ channel on Discord is useful but a mess to navigate.
What I dislike is that, as you say, their documentation is lacking and anything remotely up to date is confined to their Discord. And then many of them complain when CZcamsrs share insights that aren't 100% in line with what they believe to be correct. Instead of thinking about what they might want to do differently, they blame the content creators.
They have a very strange relationship with the people on social media that contribute so much to their business with free PR. Not really sure why, to be honest. And they need to be careful not to alienate them.
From what I gather, I'm considered "acceptable" or "not utter nonsense" among the CZcamsrs, which almost feels like a "knighting" at this stage 🤣
@@TokenizedAI I hear you. My experience with them on Discord chat is that there are all of these unpaid (well meaning) amateurs in charge of customer service! If you dare to venture an opinion they are adamant that you have to do something a certain way, or that a widely held opinion is totally incorrect, but they don't bother to update their information sheets to reflect those opinions. So the "wrong" information continues to circulate and become common use practice among people who are struggling to ride this bucking bronco. Very confusing and dismissive operating procedure, to say the least.
Hi, is there also an AI for images, that you can install and use completely offline?
Stable Diffusion. Good luck!
I kind of feel it is always an "almost", I can amost create my comic with consistant characters and stiyles...I've tried faceswap and other things but when it comes to style transfer and characters I belive we are stil lacking an specific feature for it.
How do you get the buttons to copy the prompt?
It's from a Chrome extension I built. Google "Promptalot Helper" and you should find it. It looks a bit different now because I added a bunch of features in the past few days.
@@TokenizedAI free 4 ever and add a translator
I'm afraid I don't understand what you're trying to tell me.
Can someone renew my midjourney subscription, my card is getting declined
Thanks for the video and great tips. Suggestion: use humans - much easier to relate to.
Most of my other videos use humans. I was told to do more other stuff. Whichever way I go, someone's always going to suggest the opposite 😅
@@TokenizedAI Yes but you can much better valuate the quality by doing humans
Quality of what exactly? This technique isn't about characters. It's more about environments and visual styles.
@@TokenizedAI Quality of Output
You can no longer see subtitles in any language with any smartphone
🤷🏻♂️
Midjourneys biggest failure is being locked to discord.
What do you mean? They have a Web UI that they're in the process of opening up to everyone.
Help me undestand your mindSet as well as other CZcams content Creators?
Why is Dystopia such a consistent Theme?
In over 60 to 80% of what is out there? in generative Art? This Dystopia theme runs amok.
Seriously? with ALL the NEGATIVE shit happening in world?
Why contribute to the Predictive Programming of NEGATIVE outcomes? As AI art is an incredibly powerful medium.
*Love your channel.*
But...
Please do a better job on what projections you are creating. As its very subtle, (psyOp). Please be more Conscious 🙏.
I showed you cute bunnies and badgers. I think that's a pretty broad range 🤷🏻♂️
Why not just let people create what they want to create? I'm just providing examples. You can obviously choose your own.
PS: I chose these at random btw. It has nothing to do with my taste.
It's called the zeitgeist. If you feel you're seeing too much of it, maybe you're looking in the wrong places. Maybe "Nanny-net" is more your thing. HF!
🙄🙄🙄
This is emotional art. People want to generate what excites them, both what they dream about and what they fear.
If a person has a calm, measured life, he can tickle his nerves with a horror movies, thrilling news from unstable countries, and this helps him to hold on to his peaceful life more tightly.
If a person lives in the midst of dangers, then, oddly enough, he may want to generate the 'final apocalypse', against which his troubles will no longer seem like something too scary.
All that is psychology.
@@o.k.2968 Well said. I'm glad you're "O.K" 🤓👍
it is greate