These tips are AWESOME!!! Thank you as always also i think the word is pronounced orchid- tronics. But thanks to you I have decided to rename my dog Orchi-don-tricks😆
Thanks, Nolan - That --sw info you got for me is awesome. I use it all the time. Thanks for including it here. I'm sure a lot of other users will find it helpful!
Another great batch of tips. BTW, you don't need to zoom/isolate a single image on the web interface to hide it A right-click on an image pops up a menu to like, hide or perform just about any other action upon it. Works in Create or Organize modes.
Mister Future Tech Pilot. Thanks for your video. Please tell me can I do GIF in opposite way? I mean changing in rectangular but no changing environment around rectangular?
Most useful tip of this video is using --q 0.5, 1 and 2. Just learned that --q 0.5 is enough for me, and --q 2 just looks over sharpened. Moreover, I can always use Topaz Photo AI to sharpen and add details if needed. That tip will save me many fast hours. Thanks Nolan, awesome tips.
you do your thing and you do it well so many nuggets straight from the bedrock there's a swag in this that takes time effort and coffee always enjoy to listen learn do because i like you
Great tips!!! Something I've been having a lot of fun with is using nonsense words in my prompts. Combining that with your new styles workflow should generate some interesting results
style _VERSION_, what?!! 🤯Any idea what they, well, stand for? I clicked like the second I read the title man. 😆That's awesome. These tips videos are full of gold, even for people who have been generating for a while. Thank you! 👍 edit: if the person who put the style reference together is reading this - You're a SUPERHERO! 👏
as a photographer i can chime in on quality generation and explain something youre missing. the 0.5 images you show are showcased with a F1.4 lens which blurs the outside of the image giving more of an artistic look which is wanted into portraits photos. the 2.0 quality images are shot at around F4.0 which is still desired but shows more clarity behind the face, meaning midjourney has to generate more around the face, giving it a more detailed look. both come out really good because its still focused on portrait photo but this example for quality would be different if you were getting photos of different things :D
Great tips! I have a small question: when creating design, I would like to make sure that the background color matches perfectly a certain HEX color (to copy paste the design on T-shirt mockup without seeing the difference in color), do you have any idea if it is possible to write a prompt and specify this to Midjourney?
Damn 🤘i named my company "Reimagined Works" few years ago when I started it. I felt word Reimagined described what I was here to do. Reimagine and solve problems from new perspectives, bring new ideas, discard the old approaches. Seems I tuned into bigger spirit as I've seen this word repeat many times in last few years since I've started my company
Excellent tips as always -- thanks for the shout out. On the subject of the difference between Discord and the Website, have you come across any website version of the [prefer_option_set] that we used to use for naming a character so it could be easily called up by name, or do we just have to live with the single picture CREF function?
Using quality levels makes a big difference when you are working with illustrations. Especially in niji. I prefer to stop at about 75% on niji sometimes. It gives him more realistic feel in my opinion. As far as photorealistic goes… I've never seen the reason to use quality levels.
Thanks! HIDE IMAGES: works on the Create page, if you right button click on each image you get a drop-down menu with the option to like, hide, and more.
You are my favorite on here. You have a personality but don’t waste our time blabbering. Perfect editing and super easy to follow and you add details that always help
@FutureTechPilot i watched quite a few more last night and I realized people must get on you for talking too fast. Welcome to the club. Lol Maybe that's why like you. :) Visual people tend to do that. :) keep up the great work. Graphics are always helpfully on point too!
Thanks again Nolan for these fantastic tips! Something completely different, perhaps related to the 'hiding' of images. My creation page is also filling up with generations of images. I know you can't really delete those images except on Discord with the red X. However, I don't see all my images there. It probably won't work on the Alpha site. Hey?,...(except for the "Hide" option) Another thing, (not very important by the way) but I have quite a long and complicated username on MJ. I already looked it up, but it's not easy to change, right? That's it Nolan, and thanks again for all the tips and information about MJ! Very much appreciated! Greetings from Tom, Amsterdam
@@FutureTechPilot Thanks for following up. And maybe I am asking a stupid question, But how do you generate that --sref dataset To get my personal codes you just --list p or what ever that was! LOL But how do you generate a style code like 4024715852? Am I asking this right? On the MJ site you can add an image and use it as a Style Reference, but there is no number there that I can see. How did the guy on that safari site discover all of those --sref 4024715852 codes?
@@FutureTechPilot Ah I see. So the sites like Safari probably generated these him self? I thought there was a way to see the codes like you can for Personal. Thanks!!
I am in Midjourney Bot, in Discord, I tried the Wonder Woman -- sw 1000, it came back with an error message, am I in the wrong program ? I pay for my time monthly , would love to try some of these tips. Thanks
Re: Tip 15 - you don't have to go one by one. Head over to Organize and select multiple images at once. Use the ...More button and select Hide Images. Pet Peeve: They still show up on the create tab
I dunno if MJ is giving me a glitch. I'll try explain in bullet points ot it gets too long... Ok. So I produce a mean looking 50s detective using a character image. I get him looking real mean etc. Now I got a few to chose from, one looks super mean, turned up nose and very downward eyebrows (hes a real. meanie) Next I introduce the blonde bombshell...Because of the character reference she looks like him. So I go to editor, drag in a gorgeous looking 50s female face and start to erase etc. But they all still look odd? Bit strange. Just had bad result after bad result? Is it best to remove prompt in editor? Why do all these new women have that upturned nose and 'frown-like' brows?? Most frustrating 😢.
I have been using udio lately and neglecting Midjourney. I am getting overwhelmed by all the changes happening. At sixty I am struggling to keep up with a young mans game.😉
Nobody can keep up! Just do your best, I’m older than you and just keep trying. Keeping up with everything new is impossible at any age these days. If you’re regularly watching Nolan’s stuff you are ahead of most people! Good luck! 💪🏼
I think you need to know what you want at first, then go to MJ….. trying out „ every“ ( new) possibility without having a vision in your head is stressful and may leave you lost and overwhelmed ..😅
I'm new to udio, any tips. Having a nightmare. Generated an absolute banger track, think it's one of the best I've heard, 50s style, the hook is incredible, the emotions, the rises and lows... Heres the catch 100 generations later I'm getting nothing but rubbish 😢. Maybe it is a question of happy accidents. And waiting for that 1 in 1000 generations that literally come perfect from start to finish 😂.
Q: How hard would it be for MJ to replace all of these clunky --codes we have to remember, with simple sliders? MJ need to focus 100% on creating an elegant User Interface instead of a low-end early beta release of a geek-fest for tech-curious nerds.
저는 바쁜 분들을 위해 23개의 MidJourney 팁을 요약하고, 각 팁에 대한 추가 설명을 제공했습니다. Reimagined: 프롬프트에 'reimagined'를 추가해 물체나 동물을 새로운 형태로 재구성하세요. 예를 들어, '고양이를 돌고래로 재구성'하면 AI가 고양이의 형상을 돌고래와 결합해 재해석합니다. 이 방법은 창의적인 결과를 도출하는 데 적합합니다. Pixel Art: 픽셀 아트를 일정한 스타일로 유지하려면 먼저 원하는 픽셀 아트 이미지를 생성하고 'style reference'를 사용해 그 이미지를 참조하여 새로운 장면을 만들어보세요. 'style weight'를 1000으로 설정해 원본 스타일을 최대한 유지하세요. 그리고 버전별 스타일 차이도 시험해보세요. 웹사이트 줌 조정: 웹사이트에서 옵션이 보이지 않으면 페이지 확대/축소 비율이 문제일 수 있습니다. 줌을 조정해 숨겨진 옵션(예: 화면 비율 변경)을 다시 표시할 수 있습니다. Discord와 웹사이트 연결: 웹사이트에서 생성한 이미지를 Discord에 표시하려면 'job ID'를 복사한 후, Discord에서 '/show' 명령어로 입력해 이미지를 불러올 수 있습니다. 이는 두 플랫폼 간 콘텐츠를 쉽게 공유하는 방법입니다. 비디오 그리드: Luma Dream Machine 같은 비디오 생성 도구에 MidJourney 그리드를 입력해 비디오로 변환할 수 있습니다. 그리드를 전체 이미지로 저장한 후 비디오 생성 도구에 업로드하여 비디오 효과를 확인하세요. 품질 조정: D-Q2로 이미지 생성 시간을 늘려 고품질 이미지를 얻을 수 있습니다. 반대로, D-Q5로 빠르고 저렴하게 이미지를 생성할 수 있습니다. 품질 차이를 확인하면서 자신의 목적에 맞는 옵션을 선택하세요. Fast Hours 만료: 2024년 10월 1일부터 구입한 Fast Hours는 60일 후에 만료됩니다. 이는 더 나은 서버 자원 관리를 위해 변경된 사항으로, Fast Hours를 효율적으로 사용하세요. 카드 디자인: 카드 디자인을 변경하려면 이미지 편집기를 사용해 카드의 테두리를 유지하고 중앙만 변경하세요. 이를 통해 다양한 스타일의 카드를 쉽게 만들 수 있습니다. Arrangement: 프롬프트에 'arranged'를 추가하여 이미지의 요소를 기하학적 형태로 배열할 수 있습니다. 예를 들어, '삼각형 안에 배치된 고양이들'처럼 주제를 특정한 형상으로 배치해 독특한 구성을 만드세요. 개인화 코드 추적: Discord에서 '/list P' 명령어를 사용해 생성된 모든 개인화 코드를 확인하세요. 여러 코드가 있을 경우, 이를 정리하여 나중에 사용할 때 혼동을 줄일 수 있습니다. 편집기 워크플로우: 이미지를 편집할 때 여러 문제를 한 번에 수정하려 하지 말고 하나씩 수정하는 것이 좋습니다. 한 번에 모든 문제를 해결하려고 하면 원하는 결과를 얻기 어려울 수 있습니다. GIF 만들기: Chaos 값을 높여 이미지를 다양하게 생성한 후, 이를 GIF로 변환하세요. 여러 이미지 간 변화를 통해 역동적인 시각적 효과를 얻을 수 있습니다. 캔버스 확장: 3:1 또는 1:3 비율로 이미지를 확장하면 독특한 구도를 만들 수 있습니다. 넓거나 좁은 화면 비율을 통해 작품에 새로운 느낌을 더할 수 있습니다. 개인화 코드 추가: 탐색 페이지에서 마음에 드는 이미지를 찾아 'D-P' 코드를 추가해 자신의 개인화된 스타일로 재생성하세요. 이는 다른 사용자 스타일을 기반으로 자신만의 독창적인 작품을 만들 수 있는 좋은 방법입니다. 생성 페이지 정리: 이미지 상단 옵션에서 'hide' 버튼을 사용해 생성 페이지에서 원하지 않는 이미지를 숨길 수 있습니다. 생성 페이지를 깔끔하게 정리해 필요한 이미지만 집중할 수 있습니다. Scorch 코드: 스타일 참조 코드 '36101 37247'을 사용해 독특한 텍스처와 조명 효과를 얻으세요. 이 코드는 고유한 텍스처와 그라디언트 효과를 제공하여 독특한 이미지를 만들 수 있습니다. 형태의 X: 큐브, 구, 피라미드 같은 3D 형상을 프롬프트에 추가해 해당 형상 내에서 창의적인 아트를 생성할 수 있습니다. 다양한 형상과 주제를 결합해 흥미로운 결과물을 만들어보세요. 가중치 조정: 스타일과 개인화 코드를 사용할 때, 'D-S'와 'SSW' 파라미터를 통해 각 요소의 영향을 조절할 수 있습니다. 예를 들어, 개인화 코드를 더 강하게 적용하거나 스타일을 덜 적용하는 식으로 조정할 수 있습니다. 향수적인 2000년대 스타일: 'Unreal Engine render'와 낮은 스타일 값을 사용하면 2000년대 초반 컴퓨터 그래픽 스타일을 재현할 수 있습니다. 이 스타일은 고전적인 3D 게임 또는 애니메이션 느낌을 주는 데 적합합니다. 새로운 스타일 발견: 'style reference random'을 사용해 새로운 스타일을 찾은 후, GPT에게 해당 스타일의 이름을 요청해보세요. 이렇게 하면 새로운 영감을 얻을 수 있습니다. 캐릭터 스타일 변환: 특정 스타일의 캐릭터를 생성한 후, 다른 스타일로 변환하려면 'style reference'를 사용하세요. 캐릭터 가중치를 조절해 원하는 스타일로 변환할 수 있습니다. Style Safari: 스타일 참조 코드를 검색할 수 있는 웹사이트를 사용해 다양한 스타일을 탐색하세요. 이를 통해 새로운 스타일 코드를 쉽게 찾고 적용할 수 있습니다. 인물의 다른 시대 표현: 특정 인물의 다른 시대적 표현을 생성하려면 인물 참조와 스타일 파라미터를 사용하세요. 과거 사진 스타일을 사용해 인물의 역사적 표현을 재현할 수 있습니다. 이 팁을 활용해 MidJourney에서 더 창의적인 이미지를 생성할 수 있습니다!
Too bad character reference can't use real photos. It would be awesome to use "Lineage" that way. (Unless we can create one image that looks almost exactly like our subject there doesn't seem to be a workaround)
Every tried prompting Midjourney to prompt historically accurate Medieval or Roman armor? Its a nightmare no matter how detailed your prompts are, or the pictures you use.
Fun fact : i got access to flux 1.1 and its rubbish compared to MJ. maybe its better at prompt, but it just produces the most messed up images. Been working with 50s aesthetic and flux keeps putting very modern looking women in 50s costumes. Misses the haircuts, make up, the coloring, those little details that give the shot an authentic look... Ughhhhh.
@@audiogus2651 is it more sophisticated prompting? Cant get retro looks. To be fair I wanted an old irish bar scene, all men in waistcoats having a great party around an old piano. It gave me something incredible. MJ didnt do ok, failed to capture the room and size.
@@armondtanz hard to describe the influence/impact of using Lora, but by combining Lora (particularly ones you have trained well) in your prompt and essentially blending them by assigning different weights you can pretty much achieve any style goals you want, provided you invest the time to get good at it and have some degree of visual sensibilities and vocabulary too.
@@audiogus2651 I think MJ is the best solution for ppl like me who hate going under the hood. I'm more artist than tech guy. I tried lora training using open art AI. It was abysmal. Looked like something from 18 months ago ... lol. I'm fascinated by comfyui but man that nodes setup looks insane.
Great job, Nolan. You really are the MJ test pilot with the common touch. They need to put you on the payroll.
Cheers!!
Thanks again Nolan for an insightful tip video. Always learn something new. You are one of the best sources for info and insights.
And thank you for the support!
Style Safari is an excellent resource - thank you!
Cheers!
These tips are AWESOME!!! Thank you as always also i think the word is pronounced orchid- tronics. But thanks to you I have decided to rename my dog Orchi-don-tricks😆
😂😂 yah maybe you're right
Thanks, Nolan - That --sw info you got for me is awesome. I use it all the time. Thanks for including it here. I'm sure a lot of other users will find it helpful!
Cheers!!
Another great batch of tips. BTW, you don't need to zoom/isolate a single image on the web interface to hide it A right-click on an image pops up a menu to like, hide or perform just about any other action upon it. Works in Create or Organize modes.
😂 there's always something I didn't know!
Mister Future Tech Pilot. Thanks for your video. Please tell me can I do GIF in opposite way? I mean changing in rectangular but no changing environment around rectangular?
Yah for sure!!
@@FutureTechPilotI tried do like that but it didn't work out( which prompt flag should I use? Or which highlight mode should i choose?
@@A1exJane I think you need to try rewriting your prompt completely - but it can be frustrating for sure
Nolan, great tips as always! Time to practice this new ones! Thanks, have a good week!!
Cheers!
Most useful tip of this video is using --q 0.5, 1 and 2.
Just learned that --q 0.5 is enough for me, and --q 2 just looks over sharpened.
Moreover, I can always use Topaz Photo AI to sharpen and add details if needed.
That tip will save me many fast hours.
Thanks Nolan, awesome tips.
haha yes .5 is great for rapid experimentation!
the arranged + reimagined thing is brilliant! thank you! very valuable for my work. this is team work :)
Enjoy!!
love it and this is really useful thanks for having you, Nolan , wish i can be person like you that always share the good stuff
Cheers pal!!
All respect and appreciation from Egypt 🤛
And thank you for taking the time to comment!
you do your thing and you do it well so many nuggets straight from the bedrock there's a swag in this that takes time effort and coffee always enjoy to listen learn do because i like you
hahah I really appreciate the kind words!!
Amazing as always!
As for 15: you can right-click and then chose "hide", instead of clicking the image, the menu, etc
haha there is always something I didn't know!
You gave some valuable gems here, thanks!
I'm glad to hear that!
Thanks again Nolan. Gonna check out the Style Safari site now, so cool!
Enjoy!!
Thank you very much for sharing these tips. They really helped me!
Hey I'm glad to hear that! Thanks for the comment :)
Wow, dude you saved me by showing me how to open images generated on the website in Discord. Thanks :)
haha there is always something new to learn!
Great tips!!! Something I've been having a lot of fun with is using nonsense words in my prompts. Combining that with your new styles workflow should generate some interesting results
hahah yah that sounds fun!
Thank you so much, you have helped me no end with this great video!
I'm happy to hear that!
That Style Safari site is insane!
Yah, so cool!
style _VERSION_, what?!! 🤯Any idea what they, well, stand for?
I clicked like the second I read the title man. 😆That's awesome. These tips videos are full of gold, even for people who have been generating for a while. Thank you! 👍
edit: if the person who put the style reference together is reading this - You're a SUPERHERO! 👏
haha hey I'm glad I could help out!!
Thanks so much for the website to discord instructions!
I'm happy I could help!
as a photographer i can chime in on quality generation and explain something youre missing. the 0.5 images you show are showcased with a F1.4 lens which blurs the outside of the image giving more of an artistic look which is wanted into portraits photos. the 2.0 quality images are shot at around F4.0 which is still desired but shows more clarity behind the face, meaning midjourney has to generate more around the face, giving it a more detailed look. both come out really good because its still focused on portrait photo but this example for quality would be different if you were getting photos of different things :D
Fascinating!
I'd forgotten half of these features. Thanks for sharing these tips and suggestions.
haha happens to the best of us!
I couldn't figure out how to get the images from the website to Discord for the ever-loving life of me! THANK YOU!
Cheers!
Thank you, We like you too!
haha cheers!
4:49
There was a period of time when MJ disabled the quality parameter...... I didn't know they've now enabled it again
Yes it's hard to keep track lol
Great tips! I have a small question: when creating design, I would like to make sure that the background color matches perfectly a certain HEX color (to copy paste the design on T-shirt mockup without seeing the difference in color), do you have any idea if it is possible to write a prompt and specify this to Midjourney?
No that's not possible at the moment!
Fantastic tips, thank you very much 👍
Happy to help!
Damn 🤘i named my company "Reimagined Works" few years ago when I started it. I felt word Reimagined described what I was here to do. Reimagine and solve problems from new perspectives, bring new ideas, discard the old approaches. Seems I tuned into bigger spirit as I've seen this word repeat many times in last few years since I've started my company
Life is funny that way!
Excellent tips as always -- thanks for the shout out. On the subject of the difference between Discord and the Website, have you come across any website version of the [prefer_option_set] that we used to use for naming a character so it could be easily called up by name, or do we just have to live with the single picture CREF function?
no lol :( I don't think 'prefer options' is a priority for them unfortunately
So many good tips! Thank you so much!
Cheers!
Nolan, you are awesome!
no u
Using quality levels makes a big difference when you are working with illustrations. Especially in niji. I prefer to stop at about 75% on niji sometimes. It gives him more realistic feel in my opinion. As far as photorealistic goes… I've never seen the reason to use quality levels.
Yah? I'll have to try that out!
Amazing bunch of tips! Thanks!
Cheers!
This rocks - thank you for this video.
Cheers pal!
Thanks! HIDE IMAGES: works on the Create page, if you right button click on each image you get a drop-down menu with the option to like, hide, and more.
There's always something I miss!
Well done!
🤝
14:21
These cubes are cool!
The coolest!
Thank you 🎉
Cheers!
thanks for these. Does upscaling cost hours
Yah upscaling does for sure
@@FutureTechPilot I guess I should have known. Might be worth pointing out that there are free Upscalers available which might help bulk users
You are my favorite on here. You have a personality but don’t waste our time blabbering. Perfect editing and super easy to follow and you add details that always help
Well thanks a lot for the kind words! And I'm glad my style works for you :)
@FutureTechPilot i watched quite a few more last night and I realized people must get on you for talking too fast. Welcome to the club. Lol Maybe that's why like you. :) Visual people tend to do that. :) keep up the great work. Graphics are always helpfully on point too!
Excellent video, thank you sir!
Happy to help!
🔥 Thanks !!
Cheers!
Thanks again Nolan for these fantastic tips!
Something completely different, perhaps related to the 'hiding' of images. My creation page is also filling up with generations of images. I know you can't really delete those images except on Discord with the red X. However, I don't see all my images there. It probably won't work on the Alpha site. Hey?,...(except for the "Hide" option)
Another thing, (not very important by the way) but I have quite a long and complicated username on MJ. I already looked it up, but it's not easy to change, right?
That's it Nolan, and thanks again for all the tips and information about MJ!
Very much appreciated!
Greetings from Tom,
Amsterdam
Cheers pal!!
I am interested purchase your courese but it's not mention it's life time or not
Yes it is. I'm waiting for Version 7 to release and then I am going to remake and update the course
How do you discover / list / find these codes? I know how to find the Personal Code, but not these.
What do ya mean?
@@FutureTechPilot Thanks for following up. And maybe I am asking a stupid question, But how do you generate that --sref dataset To get my personal codes you just --list p or what ever that was! LOL But how do you generate a style code like 4024715852? Am I asking this right? On the MJ site you can add an image and use it as a Style Reference, but there is no number there that I can see. How did the guy on that safari site discover all of those --sref 4024715852 codes?
@@FutureTechPilot How do you discover the codes? You can't do a --list p like for you personal coldes.
@@lineardrift Use '--sref random' at the end of your prompt!
@@FutureTechPilot Ah I see. So the sites like Safari probably generated these him self? I thought there was a way to see the codes like you can for Personal. Thanks!!
Thank you !
Cheers!
I am in Midjourney Bot, in Discord, I tried the Wonder Woman -- sw 1000, it came back with an error message, am I in the wrong program ? I pay for my time monthly , would love to try some of these tips. Thanks
If you use --sw , you need to use a style reference! Try just '--s'
That shirt 👏
💯
Thank you for pronouncing gif correctly
😂
And that's why we like you
😂
Re: Tip 15 - you don't have to go one by one. Head over to Organize and select multiple images at once. Use the ...More button and select Hide Images. Pet Peeve: They still show up on the create tab
haha yah there is always a trade off eh?
The Midjourney Master
🤙
I dunno if MJ is giving me a glitch.
I'll try explain in bullet points ot it gets too long...
Ok. So I produce a mean looking 50s detective using a character image.
I get him looking real mean etc. Now I got a few to chose from, one looks super mean, turned up nose and very downward eyebrows (hes a real.
meanie)
Next I introduce the blonde bombshell...Because of the character reference she looks like him.
So I go to editor, drag in a gorgeous looking 50s female face and start to erase etc.
But they all still look odd?
Bit strange. Just had bad result after bad result?
Is it best to remove prompt in editor?
Why do all these new women have that upturned nose and 'frown-like' brows??
Most frustrating 😢.
If you're trying to insert a new character, it definitely matters what your prompt says!
❤❤❤❤❤❤❤❤
can you please give tips to generate anime characters?
Orchidontrics 😂😂
right??
Hi 👋
🤝
I have been using udio lately and neglecting Midjourney. I am getting overwhelmed by all the changes happening. At sixty I am struggling to keep up with a young mans game.😉
😂 at 33 I can barely keep up either
@@FutureTechPilot hahha😂
Nobody can keep up! Just do your best, I’m older than you and just keep trying. Keeping up with everything new is impossible at any age these days. If you’re regularly watching Nolan’s stuff you are ahead of most people! Good luck! 💪🏼
I think you need to know what you want at first, then go to MJ….. trying out „ every“ ( new) possibility without having a vision in your head is stressful and may leave you lost and overwhelmed ..😅
I'm new to udio, any tips. Having a nightmare. Generated an absolute banger track, think it's one of the best I've heard, 50s style, the hook is incredible, the emotions, the rises and lows...
Heres the catch 100 generations later I'm getting nothing but rubbish 😢.
Maybe it is a question of happy accidents. And waiting for that 1 in 1000 generations that literally come perfect from start to finish 😂.
are you telling me there are reasons to come back to mj?..
There's always something to do!
Q: How hard would it be for MJ to replace all of these clunky --codes we have to remember, with simple sliders?
MJ need to focus 100% on creating an elegant User Interface instead of a low-end early beta release of a geek-fest for tech-curious nerds.
That's above my pay grade 😂
저는 바쁜 분들을 위해 23개의 MidJourney 팁을 요약하고, 각 팁에 대한 추가 설명을 제공했습니다.
Reimagined:
프롬프트에 'reimagined'를 추가해 물체나 동물을 새로운 형태로 재구성하세요. 예를 들어, '고양이를 돌고래로 재구성'하면 AI가 고양이의 형상을 돌고래와 결합해 재해석합니다. 이 방법은 창의적인 결과를 도출하는 데 적합합니다.
Pixel Art:
픽셀 아트를 일정한 스타일로 유지하려면 먼저 원하는 픽셀 아트 이미지를 생성하고 'style reference'를 사용해 그 이미지를 참조하여 새로운 장면을 만들어보세요. 'style weight'를 1000으로 설정해 원본 스타일을 최대한 유지하세요. 그리고 버전별 스타일 차이도 시험해보세요.
웹사이트 줌 조정:
웹사이트에서 옵션이 보이지 않으면 페이지 확대/축소 비율이 문제일 수 있습니다. 줌을 조정해 숨겨진 옵션(예: 화면 비율 변경)을 다시 표시할 수 있습니다.
Discord와 웹사이트 연결:
웹사이트에서 생성한 이미지를 Discord에 표시하려면 'job ID'를 복사한 후, Discord에서 '/show' 명령어로 입력해 이미지를 불러올 수 있습니다. 이는 두 플랫폼 간 콘텐츠를 쉽게 공유하는 방법입니다.
비디오 그리드:
Luma Dream Machine 같은 비디오 생성 도구에 MidJourney 그리드를 입력해 비디오로 변환할 수 있습니다. 그리드를 전체 이미지로 저장한 후 비디오 생성 도구에 업로드하여 비디오 효과를 확인하세요.
품질 조정:
D-Q2로 이미지 생성 시간을 늘려 고품질 이미지를 얻을 수 있습니다. 반대로, D-Q5로 빠르고 저렴하게 이미지를 생성할 수 있습니다. 품질 차이를 확인하면서 자신의 목적에 맞는 옵션을 선택하세요.
Fast Hours 만료:
2024년 10월 1일부터 구입한 Fast Hours는 60일 후에 만료됩니다. 이는 더 나은 서버 자원 관리를 위해 변경된 사항으로, Fast Hours를 효율적으로 사용하세요.
카드 디자인:
카드 디자인을 변경하려면 이미지 편집기를 사용해 카드의 테두리를 유지하고 중앙만 변경하세요. 이를 통해 다양한 스타일의 카드를 쉽게 만들 수 있습니다.
Arrangement:
프롬프트에 'arranged'를 추가하여 이미지의 요소를 기하학적 형태로 배열할 수 있습니다. 예를 들어, '삼각형 안에 배치된 고양이들'처럼 주제를 특정한 형상으로 배치해 독특한 구성을 만드세요.
개인화 코드 추적:
Discord에서 '/list P' 명령어를 사용해 생성된 모든 개인화 코드를 확인하세요. 여러 코드가 있을 경우, 이를 정리하여 나중에 사용할 때 혼동을 줄일 수 있습니다.
편집기 워크플로우:
이미지를 편집할 때 여러 문제를 한 번에 수정하려 하지 말고 하나씩 수정하는 것이 좋습니다. 한 번에 모든 문제를 해결하려고 하면 원하는 결과를 얻기 어려울 수 있습니다.
GIF 만들기:
Chaos 값을 높여 이미지를 다양하게 생성한 후, 이를 GIF로 변환하세요. 여러 이미지 간 변화를 통해 역동적인 시각적 효과를 얻을 수 있습니다.
캔버스 확장:
3:1 또는 1:3 비율로 이미지를 확장하면 독특한 구도를 만들 수 있습니다. 넓거나 좁은 화면 비율을 통해 작품에 새로운 느낌을 더할 수 있습니다.
개인화 코드 추가:
탐색 페이지에서 마음에 드는 이미지를 찾아 'D-P' 코드를 추가해 자신의 개인화된 스타일로 재생성하세요. 이는 다른 사용자 스타일을 기반으로 자신만의 독창적인 작품을 만들 수 있는 좋은 방법입니다.
생성 페이지 정리:
이미지 상단 옵션에서 'hide' 버튼을 사용해 생성 페이지에서 원하지 않는 이미지를 숨길 수 있습니다. 생성 페이지를 깔끔하게 정리해 필요한 이미지만 집중할 수 있습니다.
Scorch 코드:
스타일 참조 코드 '36101 37247'을 사용해 독특한 텍스처와 조명 효과를 얻으세요. 이 코드는 고유한 텍스처와 그라디언트 효과를 제공하여 독특한 이미지를 만들 수 있습니다.
형태의 X:
큐브, 구, 피라미드 같은 3D 형상을 프롬프트에 추가해 해당 형상 내에서 창의적인 아트를 생성할 수 있습니다. 다양한 형상과 주제를 결합해 흥미로운 결과물을 만들어보세요.
가중치 조정:
스타일과 개인화 코드를 사용할 때, 'D-S'와 'SSW' 파라미터를 통해 각 요소의 영향을 조절할 수 있습니다. 예를 들어, 개인화 코드를 더 강하게 적용하거나 스타일을 덜 적용하는 식으로 조정할 수 있습니다.
향수적인 2000년대 스타일:
'Unreal Engine render'와 낮은 스타일 값을 사용하면 2000년대 초반 컴퓨터 그래픽 스타일을 재현할 수 있습니다. 이 스타일은 고전적인 3D 게임 또는 애니메이션 느낌을 주는 데 적합합니다.
새로운 스타일 발견:
'style reference random'을 사용해 새로운 스타일을 찾은 후, GPT에게 해당 스타일의 이름을 요청해보세요. 이렇게 하면 새로운 영감을 얻을 수 있습니다.
캐릭터 스타일 변환:
특정 스타일의 캐릭터를 생성한 후, 다른 스타일로 변환하려면 'style reference'를 사용하세요. 캐릭터 가중치를 조절해 원하는 스타일로 변환할 수 있습니다.
Style Safari:
스타일 참조 코드를 검색할 수 있는 웹사이트를 사용해 다양한 스타일을 탐색하세요. 이를 통해 새로운 스타일 코드를 쉽게 찾고 적용할 수 있습니다.
인물의 다른 시대 표현:
특정 인물의 다른 시대적 표현을 생성하려면 인물 참조와 스타일 파라미터를 사용하세요. 과거 사진 스타일을 사용해 인물의 역사적 표현을 재현할 수 있습니다.
이 팁을 활용해 MidJourney에서 더 창의적인 이미지를 생성할 수 있습니다!
I wish I understood what this said
@@FutureTechPilot How about do you ask that ChatGPT help you haha
Too bad character reference can't use real photos. It would be awesome to use "Lineage" that way. (Unless we can create one image that looks almost exactly like our subject there doesn't seem to be a workaround)
Keep that tip in mind for the future because I'm sure that feature will improve!!
I'm a bit confused, it was Orchid-tronics, did you deliberately mis-pronounce it as a joke, is it because I'm British...?
no I'm just a dummie and actually thought it said Orchi-dontrics 😂
The 0.5 photos look better to me
haha to each their own!
Every tried prompting Midjourney to prompt historically accurate Medieval or Roman armor? Its a nightmare no matter how detailed your prompts are, or the pictures you use.
Yah unfortunately that isn't really what Midjourney is designed to do right now
smogcore
I like
BALD bEN aFFLECK !!!!
In the last 21 Years, you're the third person to say that. Thank you
Great content! But you speak so fast :D calm down ;)
😂 I'll try to remember that
@@FutureTechPilot no your pacing is good, don't go changing..
Fun fact : i got access to flux 1.1 and its rubbish compared to MJ. maybe its better at prompt, but it just produces the most messed up images. Been working with 50s aesthetic and flux keeps putting very modern looking women in 50s costumes. Misses the haircuts, make up, the coloring, those little details that give the shot an authentic look... Ughhhhh.
MJ is definitely the easiest to use
Yah with Flux you pretty much have to be into training your own Lora to get the most out of it.
@@audiogus2651 is it more sophisticated prompting?
Cant get retro looks.
To be fair I wanted an old irish bar scene, all men in waistcoats having a great party around an old piano. It gave me something incredible.
MJ didnt do ok, failed to capture the room and size.
@@armondtanz hard to describe the influence/impact of using Lora, but by combining Lora (particularly ones you have trained well) in your prompt and essentially blending them by assigning different weights you can pretty much achieve any style goals you want, provided you invest the time to get good at it and have some degree of visual sensibilities and vocabulary too.
@@audiogus2651 I think MJ is the best solution for ppl like me who hate going under the hood. I'm more artist than tech guy.
I tried lora training using open art AI. It was abysmal. Looked like something from 18 months ago ... lol.
I'm fascinated by comfyui but man that nodes setup looks insane.
Great tips but that last woman just has a very weird face
😂
Midjourney is old news! Bring Flux or ideogram videos.
😐
The besttttt!!!