This sounds really interesting, not to mention helpful! I wasn't aware that the text weights weren't included if someone else copies your prompt. Off to try it out :)
Рік тому
Fantastic workflow to structure a composition we want
Interesting. I’ve done similar things with taking midjourney generations and chopping up different things from different generations in photoshop for a starting image.
15:00 Is what makes your final image unique is the simple text prompt you added at the very end @TokenizedAl? I’m wondering if what you mean by “unique” is being able to have the final prompt not reference the previous image urls and not the same as the previous prompts.
It's the fact that you can add weights to each individual image reference and your chosen weights aren't known to the public. Unlike the rest of your prompt.
Chris, your tutorials are amazing. You explain topics in very easy and especially logical way. It makes it a lot easier to understand even difficult issues. Could you tell why do you sometimes use weight without any number only with double colon? In this video you did it in refining blended image part. Thank you!
Do you have any advice on turning photos into a specific painting style *without* changing the contents of the original photo? I have a lot of my own photos that I just want to stylize (i.e. turn into a 1950s style airbrush painting or realistic acrylic painting)
Thanks for another great tutorial. My understanding is that MJ will prevent you from using a seed or job number that wasn't generated by your account. MJ's website doesn't include all the info you'd need to reproduce the exact result... for good reason.
@@TokenizedAI But you can't expect to replicate an exact copy of an image generated by someone else's account by using the information on the MJ website, right?
So weight sum up to 100%. Im bad at math so using 0.7, 1.3, 5 for me difficult. But i can easily understand 20, 30, 50. Im gonna try it today. As always- good vid. Straight to my "MJ for future reference" playlist 😂
Hey Christian! Excellent video as usual! Can you please try to make MJ produce something for *Wimmelbilder* style? I've never seen any prompt able to do so... I thought about asking you! keep up the great work! 👍🏻 🙂
MJ is actually pretty good at "wimmelbilder". Just put it in your prompt. You may also need to try out v4 for this because I haven't tested it with v5 yet.
Sorry, I'm new to this midjourney. I heard midjourney is one of those machine learning algorithms. I thought making someone's picture with it was tough, even with the same seed and parameters, because my MJ is fed totally different keywords than the other person.
@@TokenizedAI Rain dripped, pooling acid-lime by the garage. Iridescent puddles mirrored dark brown and indigo skies, blending day into night. Sludgepunk haze lingered. Bio-luminescent organisms danced in the water, lighting the road. Nearby, waste-powered lights hummed, casting low-angle shadows. Transfixing marine scenes glowed, a manapunk masterpiece with tech art details --ar 16:9 --upbeta --seed 83851817
I didn't mean that you should actually give me your prompt. I was just saying that with that prompt, I could reproduce it perfectly. The only limitation here is that I don't know which version you used. Btw, --upbeta is an old deprecated parameter. Unless you're using an older MJ version, it doesn't actually do anything. v5+ doesn't even have a separate upscaler yet.
Unless you have a Pro subscription and are using Stealth Mode, every single one of your prompts is public on the Midjourney web app (the browser platform). Whether you generate your prompts in the official MJ Discord or your own, doesn't matter.
??? who cares if someone takes your picture?? it's IA generated based on several styles copied on real artists,it didn't worried you to take transformed pictures from artists, so...
What’s kind of especially bizarre about this video, is that there are 100s of videos SOLVING for the problem that Midjourney CAN’T recreate the identical image. Go ahead and try. Put in all of your parameters and then do it again. The images will be wildly different. This video is solving for a problem that not only doesn’t exist, but so much so doesn’t exist that people have created tutorials for integrating Midjourney into stable diffusion in order to compensate for the massive issue that it’s impossible to recreate an image you created in Midjourney using only Midjourney.
There's nothing bizarre about it. Again, I'm not sure where you seem to have taken the wrong turn in the logic, but it really sounds like you've never used the --seed parameter before. The problem that you claim does not exist is a problem that I keep hearing from seasoned professionals from the creative industry (in workshops) that emphasize that they need a way to create 100% unique images. No offense, but I'll take their word over a random YT comment. 🤷🏻♂️
I'm curious. Are you using an old version of Midjourney? Like v3? Cause with v5 and beyond, a 100% identical prompt (incl. the same seed) will always return the same images. I mean, they even put it in the official docs: docs.midjourney.com/docs/seeds See the section "Jobs run two times with --seed 123". This is exactly the behaviour that everyone here, except for you it seems, can observe.
I never clicked a video so fast, thank you for uploading. your content is always gold.
Wow, the methods for ensuring Midjourney images are completely unique are amazing! This guide offers great strategies for standing out.
!!!! Chris... they did the thing I JUST asked you about. Incredible!
I must admit I was initially a bit skeptical, but I see how the final step brought it all together. Interesting how similar but different they can be!
Chris -- you always seem to answer questions I've been wondering about. Once again, nice job.
14:02 This, in particular, was something I'm quite sure many of us didn't think of! Fantastic. 👍 Thank you very much.
This sounds really interesting, not to mention helpful! I wasn't aware that the text weights weren't included if someone else copies your prompt. Off to try it out :)
Fantastic workflow to structure a composition we want
Your explanation is so good! ❤
Great vid. Do you have a video detailing that final step a bit more? A lot of my images have that 3D comic look that I'd like to refine.
Not on my YT channel. I explain Scaffolding Remix in detail in my course though.
Always on point and ahead of the game!!!🎉
What is your stylize setting? Lately I’ve been very disappointed with the stylization parameter.
I just use the default
Thank you! That’s very good content ❤
Thank you for sharing this information. I have been working on a image for a month…
Thank you so much for this video. I definitely needed it for my workflow!!
WOW! This was so helpful! Thank you!
good thumbnail!
Interesting. I’ve done similar things with taking midjourney generations and chopping up different things from different generations in photoshop for a starting image.
Thank's, thats great 😀
15:00 Is what makes your final image unique is the simple text prompt you added at the very end @TokenizedAl? I’m wondering if what you mean by “unique” is being able to have the final prompt not reference the previous image urls and not the same as the previous prompts.
It's the fact that you can add weights to each individual image reference and your chosen weights aren't known to the public. Unlike the rest of your prompt.
@@TokenizedAI I get it now. Thank you 😃. Nice photos you took at the end 👍 15:45 .
Love it...eye opening technique
Chris, your tutorials are amazing. You explain topics in very easy and especially logical way. It makes it a lot easier to understand even difficult issues. Could you tell why do you sometimes use weight without any number only with double colon? In this video you did it in refining blended image part. Thank you!
If you don't set a specific weight, it just defaults to 1.
@@TokenizedAI thank you a lot!
Do you have any advice on turning photos into a specific painting style *without* changing the contents of the original photo? I have a lot of my own photos that I just want to stylize (i.e. turn into a 1950s style airbrush painting or realistic acrylic painting)
Afraid not, because that's not really the purpose of MJ.
@@TokenizedAI Ah ok, had a feeling. Thanks for letting me know!
Tu curso se encuentra disponible en otros idiomas?, Implementando doblaje con IA? ó Subtitulado, tal vez?
Unfortunately, it's only in English right now. But I may have to think about adding subtitles 🤔
Thankyou Christian
Thanks for another great tutorial.
My understanding is that MJ will prevent you from using a seed or job number that wasn't generated by your account. MJ's website doesn't include all the info you'd need to reproduce the exact result... for good reason.
Job number yes. But not the seed. The seed is just a random number.
@@TokenizedAI But you can't expect to replicate an exact copy of an image generated by someone else's account by using the information on the MJ website, right?
Depends on how they prompt. If they always add all parameters (incl. seed) and never use any defaults, then sure you can.
@@TokenizedAI Viva La Stealth.
🥷
Thank you!!
Hey from Morocco 🇲🇦
❤️
So weight sum up to 100%. Im bad at math so using 0.7, 1.3, 5 for me difficult. But i can easily understand 20, 30, 50. Im gonna try it today. As always- good vid. Straight to my "MJ for future reference" playlist 😂
super !!🙏
always incredible
I've been doing it for a while like this: /Imagine prompt: [ ]::2 [ ]::3 rest of prompt
Hey Christian! Excellent video as usual! Can you please try to make MJ produce something for *Wimmelbilder* style? I've never seen any prompt able to do so... I thought about asking you! keep up the great work! 👍🏻 🙂
MJ is actually pretty good at "wimmelbilder". Just put it in your prompt. You may also need to try out v4 for this because I haven't tested it with v5 yet.
Thank you.
Awesome.
Sorry, I'm new to this midjourney. I heard midjourney is one of those machine learning algorithms. I thought making someone's picture with it was tough, even with the same seed and parameters, because my MJ is fed totally different keywords than the other person.
All your prompts are publicly visible. If you give me the exact same prompt with a seed, I can reproduce it pixel for pixel.
@@TokenizedAI Rain dripped, pooling acid-lime by the garage. Iridescent puddles mirrored dark brown and indigo skies, blending day into night. Sludgepunk haze lingered. Bio-luminescent organisms danced in the water, lighting the road. Nearby, waste-powered lights hummed, casting low-angle shadows. Transfixing marine scenes glowed, a manapunk masterpiece with tech art details --ar 16:9 --upbeta --seed 83851817
I didn't mean that you should actually give me your prompt. I was just saying that with that prompt, I could reproduce it perfectly. The only limitation here is that I don't know which version you used. Btw, --upbeta is an old deprecated parameter. Unless you're using an older MJ version, it doesn't actually do anything. v5+ doesn't even have a separate upscaler yet.
@@TokenizedAI What do you mean by saying that all prompts are visible to everyone? Even in the paid version in a private room?
Unless you have a Pro subscription and are using Stealth Mode, every single one of your prompts is public on the Midjourney web app (the browser platform).
Whether you generate your prompts in the official MJ Discord or your own, doesn't matter.
My images are already unique lol
👍🏻
Gold
You used style raw. It seems you wanted it in a cinematic photorealistic style… like your vacation pics were. Why did you use style raw?
Raw gives me more control and less of MJ's default stylization, which isn't necessarily always better.
I can see that you have been captivated by dune. did you find shai hulud in dubai? or maybe enjoyed some fremen ladies?
I wouldn't say captivated. But it makes for a good theme
Use the envelope icon then you will have all 4 image addresses..........don't need to waste upscale time.
I have more GPU hours than I'd ever need. Plus, the images are already upscaled. Makes little to no difference in terms of fast hours usage.
??? who cares if someone takes your picture?? it's IA generated based on several styles copied on real artists,it didn't worried you to take transformed pictures from artists, so...
You don't get it. I'm not worried. Others are.
❤🔥
What’s kind of especially bizarre about this video, is that there are 100s of videos SOLVING for the problem that Midjourney CAN’T recreate the identical image.
Go ahead and try. Put in all of your parameters and then do it again. The images will be wildly different. This video is solving for a problem that not only doesn’t exist, but so much so doesn’t exist that people have created tutorials for integrating Midjourney into stable diffusion in order to compensate for the massive issue that it’s impossible to recreate an image you created in Midjourney using only Midjourney.
There's nothing bizarre about it.
Again, I'm not sure where you seem to have taken the wrong turn in the logic, but it really sounds like you've never used the --seed parameter before.
The problem that you claim does not exist is a problem that I keep hearing from seasoned professionals from the creative industry (in workshops) that emphasize that they need a way to create 100% unique images.
No offense, but I'll take their word over a random YT comment. 🤷🏻♂️
@@TokenizedAI Then make a video of you doing it. Make a video of you creating two identical images in Midjourney.
I'm curious. Are you using an old version of Midjourney? Like v3? Cause with v5 and beyond, a 100% identical prompt (incl. the same seed) will always return the same images.
I mean, they even put it in the official docs: docs.midjourney.com/docs/seeds
See the section "Jobs run two times with --seed 123". This is exactly the behaviour that everyone here, except for you it seems, can observe.