More Consistent Ai Comic Characters with MidJourney's New Remix Setting
Vložit
- čas přidán 29. 08. 2024
- Read the Camp Peculiar ai Webcomic on Instagram
/ camppeculiar
Get my free Instagram Comic Template and more Ai art comic resources at:
camppeculiar.com
Not wanting to be the last Ai Art channel covering the release of the remix setting in MidJourney we take a look at how this exciting and workflow changing tool can help you create consistent characters facial expressions and story details for you Ai comic characters.
Also in this video I release the Camp Peculiar facial expression Character sheet which can help you organize and categorize your facial expressions for an ai comic character.
bro I suggest you start usign the seed number in order to have a more consistent chatacter
this was 2months ago, i am sure he advanced since then, hehe
Sidenote: You do not have to upscale. Remixing from the grid works just fine. I only advised upscaling for 3 reasons. 1. I wanted to be able to easily find my base image and prevent style drift. 2. I wasn't sure about the time/cost savings when making the video of having to take every remixed variant up the chain of upscale, remaster, upscale vs just starting the process at a good upscaled base image. 3. I wasn't sure if the extra detail/data added by a regular or beta upscale helped with the style consistency. That was my thinking. Thanks for asking such a great question.
Thank you again!👍
What about the body/poses? I find it so difficult to change the characters poses.. And also the view from close-up to medium or medium long.. How to do that?
@@demsyciu AI is currently very bad with body parts/poses. But for broader views, try typing “wide angle perspective/shot”
Here my story, looking forward to trying out my journey, great video czcams.com/video/zmMO0TVAHD4/video.html
You can also give it the prompt for multiple panels: 6 panels, 9 panels, 16 panels. You can get it to do the same character in multiple poses.
Can you elaborate on that? What are these panels?
@@climate-civilizations Try a prompt like, “/imagine 9 panels of a cartoon character with different emotional expressions”
@@AJCrabill Thanks, I will
The efficiency of these videos really makes them a notch above the other Midjourney videos that I've seen so far. Good work :)
"More" consistent is the key qualifier here. Actual consistency on demand will be a game-changer (as will proper hands...) but for now, it's about rerolling for what you want an ungodly amount of times.
I LOVE this channel! The rate in which you are cranking out these super cool, super educational videos is amazing. If I had an art teacher like you when I was younger they'd be calling me Spielberg or Michaelangelo. Simply awesome.
Hey Prospector. Thanks for the kind words, I really appreciate it. Still plenty of time for you to rival Speilberg, Kirby or Ditko :)
@@camppeculiar thanks for the vote of confidence but unfortunately, I'll need to wait for the Academy to create the, "Best Stick Figure Flip Book" category to stand a chance at a dishonorable mention😭
Well, now I really want to hear what the acceptance speech for "Best Stick Figure Flip Book" sounds like :)
A channel for comics and animation! My life is complete and I haven;t created anything yet. I can't wait to watch all your videos and see what's on your web page!
Best character mj video I’ve found 👏✌️
Hey Jean! Thanks for the comment.
Using the seed number will help with consistency
Thank you , I’m doing a very intrincate coloring book, but I have a story on mind about 2 Mayan kids, this was very useful!!!
I tried getting into this but don't like it seems over rated to me , they look different
The do look different and the more you look the more different you notice they are. I think at this stage in the development of diffusion and comics we are going for "good enough to not completely break the suspension of disbelief". Does that make the techniques and technology overrated, maybe, probably, still if you are looking to tell a story and think your readers will let go of the inconsistencies its a good option. Thanks for the comment.
Awesome video! Thank you. A suggestion for future videos: Can you make a video where you show how you put it all together to make your comic book? Here's another one: What about body movements? What if we want to capture different motions like running,
, etc?
Hey Marilyn. Did your other fantastic question get answered? The one on upscaling. I don't see it anymore :( Replying to this question below.
I would love to make a video on my whole process. Can I ask you what your end goal would be? Print, Comixology, Webtoon/Tapas, Social Media? Are you looking for the process on a standard us comic printed through print ninja, mixam, or ka-blam?
Dynamic poses are hard, especially with facial and outfit consistency. My workflow at the moment relies on photoshop, but I did start prepping a video on the subject last night :) more soon.
@@camppeculiar I thought it was a bad question so I deleted it :) But since you read it, please let me know the answer if possible
@@camppeculiar I was thinking about making a cute children's book/ebook. I am not much into comic books but I'd love to make a regular book. 🙏
Love it! This is what I try to find for so long
You can also use the /blend command and combine the two images for more consistency.
hi, I like it. Thank you for sharing
Hi there Jenny. Thanks for the comment.
00:13 🔄 MidJourney's new Remix setting revolutionizes creating webcomic or graphic novel characters with AI.
00:27 💡 Previously, creating varied facial expressions for AI-generated characters was challenging, requiring manual effort and prompt hacking.
01:37 🚀 The new Remix feature in MidJourney allows easy generation of AI comic characters with customizable facial expressions.
03:13 🖼️ The demo uses a prompt for a "female comic book elf Ranger" with "no expression, no emotion" for baseline creation.
04:39 📏 Upscale the chosen image, then use a character motion sheet in Photoshop for organizing and editing.
06:01 🎨 Utilize the Remix feature to create variations of the character's expression, using descriptive words like "big smile" and "happy."
07:53 🎭 Adjust style words to guide the AI towards desired artistic characteristics in the generated image.
09:05 ⚠️ To maintain consistent style, start each new emotion variation from the base image rather than iterating from previous iterations.
09:33 🤖 MidJourney's Remix feature enables the creation of consistent and expressive comic characters, opening up new creative possibilities for visual storytellers.
It’s interesting, but some of the facial expressions don’t look like they belong to the original character. I think this is a good feature that Midjourney has added, and I really hope they will continue to improve by offering features that can consistently repeat the same character in different scenes. Hopefully, it will also become easier to create profile images of a character, as well as any angle that is needed, since comics will require us to work with a variety of angles. Additionally, we don’t just need portraits, but we also need to be able to pose a character in full body shots.
Artists are now prompts people. Now that's honing your skills.
looks like it still needs a lot of work. But it looks great to build a prototype and get something that represents what you are going for.So in that sense I think its still pretty remarkable!
This is amazing and an awesome demonstration of a practical AI workflow to create things 🙏
Thank you for sharing. Just a suggestion on your character sheet, you'd be better off if you use Indesign for that type of layout. You'll have lighter document sizes, multiple pages, searchable text, easy to modify sizes of picture boxes, fewer layers needed, borders and a lot more. Photoshop although very capable, was not really designed to create page layouts. Respectfully.
That thumbnail is horrifying, these images kinda lead into the uncanny valley.
Amazing
awesome stuff. really really awesome , thanks so much.
Hey Montana! Thanks for the kind words and comment. I really do appreciate it.
now it needs to do the same for different posses, swinging a sword, sending out flames from their hands etc
Hands……lol
@@101touchapps yes from all three of their 4 to 8 fingered hands.
Nice, please keep it going. Could you please also show it all with Stable Diffusion?
Hi SW. More Stable Diffusion videos...you got it.
@@camppeculiar nice
Good info. Thanks for the video. I think you coudl layouver the expression on the face and just erase in the expression you are wanting ontop of the base. Cheers
The algorithm gods are showing me this cause on corridor crew they're about to have some crazy AI generated anime video, in some future weeks
Thanks for the vid. I have come up with a similar style of working - slightly less organised than yours. It’s a bit sad MidJourney is not quite there yet.. needs more work to help with comic book production workflow. If they created a separate prompt function for determining the style of drawing - that would be greatly helpful. Also, allowing easier commands which, let’s say, allow reusing an image you feed in with only the minimal changes required.. that would be very helpful also.
Thanks for the content and have subbed. Fantastic work. Your workflow/structure to get the comic out is fantastic and I got to learn from you. Thank you sir!
To your point, variations of variations does take away some character consistency. As I have been experimenting, I have gotten better consistency using a root image and writing different prompts with the root image +seed. I am currently using variations to introduce smaller delta changes.
Hey Vibin, thanks for the comment and kind words. Welcome!
Brilliant again!
Ya know...didn't want to be the last video about Remix to come out, CZcams moves quick :( Thanks for the comment, appreciate it.
Hey thanks for the superb content, trying my best to learn the skill of the a.i. to create stories, love story telling, ill check more on your channel.
Great content, keep it up!
Thanks Miho!
Still clearly needs a bit of work. But that consistency is impressive for an ai
Excellent
Thank you so much 😀
I am in remix mode but it doesnt prompt me to input more instructions, it goes right to creating variations. What am i doing wrong?
same !!! somebody?
I already know! You have to go into /settings and then click remix mode! You're welcome!
@@dariawilsz6586 thats what i did that time, but for some reason it didnt stick. Had to do it again.
Thank you!
I love this
Wouldn't it be way easier to do all the framing in Adobe XD, export a high quality PNG and then make the edits in photo shop?
Superb, thank you so much. Love your style.
Thank you! Very helpful. Just subscribed too 🙌 🎄🎉 Merry Christmas to you and to all.
Yo this is cool. I'm trying to work on a cartoon! I animated a video game pretty much from scratch 3 years ago totally alone. I strangely enjoyed a lot of the key framing and realized if I was more of an artist I would 100% have made an animated cartoon. Well looks like that time has come!
Hi. Amazing video! Thanks.
The link to the template on your website does not seem to work. Can you please let me know where I can download it?
To be fussy, is the remix icon a “4” side of a dice? Or is it four tiny knobs that you could twist independently to change the settings? If you squint real close it looks like there’s dial setting numbers around the top of each knob.
Wow. Mind blown. No I have to rethink everything I have assumed to be true about user interfaces and ui icons. :)
I saw a similar video that adds --seed to it, does mixer mode automatically use the same seed ?
Thats such a good question, I always assumed remix was, at the very least, using the seed if not doing some seed and image prompting behind the scenes
10/10 video!
Hey thanks Still Ready, much appreciated.
This is informative, but we are facing difficulty in instructing Midjourney to make a profile of the character. How are you solving this issue?
See my video on how to get side profile ai comic characters. It's hard. Thanks for the comment, good luck!
Would it work better if you used the seed prompt?
I think so, remixing plus --seed or --sameseed, plus throwing the image url in there for kicks, dropping the --chaos maby lower --s a tad, may improve your consistency but there is still going to be what I can drift for the next for months or maybe a year or so. Thanks for the comment!
Thanks, this was indeed very helpful!
Learning to draw is more fun
Midjourney really mangles the number of fingers on a hand...and toes. The more characters (people) in the scene the more likely there lots of distortion and weirdness going on. Any ideas on how to cure that?
Wow, so much hard work.
This is Genius
Everybody tells me to turn on Remix mode but that just makes it a lot harder to keep consistent images for me. Turning it off gives me way better and almost identical variations of the previous picture.
Wait I'm confused. I'm not really a cross my fingers kind of person, but every now and then I do. Either way, do we cross our fingers before or hitting enter? (p.s. subscribed just now, like the dry humor)
You should really be crossing your fingers more frequently, it'll change your Starbucks order accuracy percentage by a few points at least ;) Thanks for the sub.
Try "white background" to make it easier to remove the background.
using --seed might make them more consistent too
you may want to try white background instead
Thanks for this. I subbed!
Great stuff, very helpful!
Amazing. Thanks much
Hey Marthino. Thanks for the comment.
I don't know why your lady stills look like the Default version. When I start to make variations from my char, then it starts to add different utilities. The default has no crown or bandana, but when I create a smile or angry face expression, then the variations get too some equipment. That is kinda annoying.
Incredible what some people do to avoid actually learning to draw.
It’s not a dice. It’s a metal box with four knobs.
Also, sorry, that sounds so grumpy! Thank you for this awesome video. Love your website too!
Why don't they just create an Ai comic creator. I am so motivated to finish my graphic novel and hopefully in a couple of years someone created a comic Ai creator we don't have to tweak too much.
Strange. I'm using paid pro version of midjourney. The Make Variations doesn't do anything on mine. I have Remix Mode turned on. Doesn't do anything. No new images get generated. Tried several images and changed. No luck. ?
Make sure v4 is enabled in settings or add v4 to the prompt
Thank you for sharing... i would like to share with you this prompt " cartoon character model sheets for YOUR CHARACTER, THE STYLE YOU WANT , --s 5000 --ar 32:9 " ... its working like a charm :)
Hey Fero! Thanks for the prompt. I was making a video about a similar type of prompt, and found the 32:9 super helpful. Can I credit you in that video or mention this comment?
@@camppeculiar yes of course, thanks… i shared for “cartoon character model sheets” in a side, and “ar 32:9” on the other side.
🙏😁🙏
Yay! Thanks. I got one to add to the mix, that I hope works for ya. Video soon Thanks again.
Are you going to do one for side profile?
Can you not use --seed to keep it to the original?
You mentioned creating side profiles views... I'm assuming it's ridiculously hard.. but any advice there? Thanks so much!
This is not dice, it is mix console icon :)
Thank you 😄
Have you had issues with this? How? You just use the image as a prompt.
I am new to midjourney and this video was helpful but i still can't seem to do what i want to do. I am making a story and when i create a character in midjourney and i have it do something (play on a swing), if i try to re create that exact same character doing something else (riding a bike) i can't figure out how to make the character look the same. it seems when you change what the character does,, even if you describe it the same way, the character changes, sometimes quite markedly. i tried the remix feature but keep getting the same problem. is there a way to address this? this is a critical capability, is midjourney the best program for me to use? really appreciate any input, thanks
This is nice, but the fact that you still have to do everything on their discord page is a turn off for me.
I couldn't find your website 🙁🙁🙁
Great video buddy, new sub from me.
Just started working on my own.
you gotta look up for the word "consistent" in the dictionary
is there a method to upgrade existing art using AI art? I got a few pieces I had contracted out that are lack luster.
Hi James. I think you'd be looking at using them in an image prompt with some heavy weighting to favor the image over what ever text you choose to accompany your image. So, yes it's possible in almost every Ai art engine and will just require you to balance the image weight prompt with whatever details you try and add or correct. May work, may not. Its probably worth a shot.
It's cool but the outfits are way too wildly inconsistent for people not to notice.
I feel like the AI is definitely capable of reproducing the character at this point. The problem is that the programming/engine just isn't good enough. The AI is beyond us, the human programming controlling it is what's failing.
Everything is beyond you mediocre person.
Looks like greta
Lady Cabbage lol
thanks a lot
You're welcome!
Muito bom!
Cool! How much do you pay for midjourney?
I pay for the base monthly option and for the private option, so $35 maybe? I switch between /relax and /fast. a dozen times a day :)
There is very little consistency there, perhaps learn how to draw and then you could.... Oh wait....
use --seed .
Remaster go bye-bye.
Is there a way to hack your subscriber count up to a million?
Find the one that looks like your character 😂
Hi everyone, I'm a programming teacher at univ. I would like to work with someone to make a story about a dystopian programming university and the adventures of a class of students that somehow mange to pass exams in an extremely challenging environment. Would someone want to hear more about this idea?
As the students at CodeTech struggled to pass their exams in the challenging programming curriculum, they also faced a number of real-world struggles.
Alex, the leader of the study group, was constantly struggling financially. He came from a low-income family, and paying for tuition and living expenses was a constant struggle. Many nights, he went to bed hungry because he couldn't afford to buy food. But he refused to let his financial struggles hold him back. He worked hard and took on multiple part-time jobs to make ends meet.
Mia, another member of the study group, struggled with mental health issues. She had always struggled with anxiety and depression, and the high-pressure environment of CodeTech only made it worse. She often found herself feeling overwhelmed and alone, but she refused to let her mental health issues define her. With the support of her classmates, she sought help and learned coping mechanisms that allowed her to stay on top of her studies.
Despite these struggles, the students were determined to succeed. They formed a close-knit study group and worked tirelessly, pouring over textbooks and practicing code for hours on end. As the weeks went by, they began to see improvement. They were able to complete assignments with ease, and their test scores began to rise.
The professors took notice, and soon, the students were being hailed as the top performers in the school. But their success did not go unnoticed by the powers that be. As they neared the end of the semester, the students were confronted by a group of shadowy figures who claimed to be representatives of the university's administration.
The figures told the students that they were too good, and that their success was threatening the status quo. They offered the students a deal: give up their hard-earned knowledge and return to being mediocre, or face the consequences.
The students were torn, but Alex knew what had to be done. He stood up to the figures, telling them that they would never give up what they had worked so hard for. And with the support of his classmates, he held firm.
In the end, the students stood their ground, and the shadowy figures backed down. The class graduated with top honors, and the students went on to become some of the most successful programmers in the industry.
For Alex, the struggles he faced only made his success that much sweeter. He knew that he had earned it, and he was proud of what he had accomplished. And for Mia, the victory was even more meaningful. She had overcome her mental health struggles and proved to herself and the world that she was capable of anything.
The dystopian programming university of CodeTech was never the same again. The students had proved that with hard work, determination, and a little bit of support, anything was possible, even in the most challenging of environments.
@@obomasinladen Thanks for the ideas, however it comes a little late, I've already done the first chapter on my own and I have a plan for the rest. Give it a look here czcams.com/video/ElChNWY8JWQ/video.html
lol easier is draw it..... rather then spend time to get lucky on midjourney :D
In a few more years we will be able to make comics with AI perfectly and probaly vdo too.
Well looks like I have to start looking for housbend, or a way to do sex work safely, because outside of art, and visual story telling I don't really have any other options to surviv. I will still draw, juts my life goal will be shaterd.
It’s somehow reassuring for the traditionalists to see that AI comic characters are so lousy and in poor taste like bad human creators. 😂😂😂😂😂😂😂
;-)
:)
Stop stealing via ai
How is it stealing? He is using a tool to create images?
Weak
Can we make nursery rhymes videos by using mid journey ?
is any suggested platform where i get 3d cartoon animation for kids nursery rhymes >>