7:52 I've done some cursory testing (definitely not exhaustive) of SHORTEN and in it's details breakdown its claims about weight percentages of certain words .......but I've found noticeable flaws in those percentages. For example, it claimed "0.00%" for a certain group of words but the generated images with the same prompt with only those words removed had a noticeable impact on the generated image. So those words couldn't possibly be 0.00% as indicated. Granted, I wasn't using the same seed, but multiple rolls of the same prompt only missing those words all had more or less the same outcome.....a noticeable difference. Anyway, I'll have to do more thorough testing to flesh this out.
@MakingPhoto Yes, using the same seed is actually the best way to go and I'll admit that I could be mistaken in my conclusion. But from my initial sub-optimal testing, there does appear to be inconsistency in some of the 0.00% or near 0.00% claims. And I'm not referring to words like "a" or "the" but rather phrases such as "shallow depth of field focal aperture blur" where if I recall it correctly, MJ weighed each of those words in my longish prompts (and the fact that they were longish could make a difference in comparison to short prompts) as 0.00 or maybe 0.01....... but in my testing, there often was an effect of the background and foreground being blurred but without that phrase the blurring didn't seemingly occur at all or it was noticeable less.
Kind of irrelevant, but I know we can weight words, but how about 'concepts? E.g... I want a cat wearing red shoes. Would I weight red AND shoes equally? Or just shoes because it's the end word of the concept? You get me? Maybe MJ in the future will allow you to contain a phrase somehow ((Red Shoes::2)) Like that maybe?
@@MakingPhoto Because if you write red shoes::2 it's just weighting shoes, right? So being able to tell MJ that the preceding words are part of the weight would be cool don't you think? Sorry, I'm rambling :D
@@Shanghaimartin It should weight the phrase between the :: so should weight "red shoes". But to clarify, I don't think /shorten works with weighted prompts yet
PROMPT: "High Fashion Model --v 5.2" This is my go-to prompt when I want something in this style. You'll get wildly different details, but a similar vibe.
Excellent video! I've been playing around with 5.2. They've really raised the bar. And they're being smart about not naming it 'v6' yet. --shorten has a way to go I believe. Thank you. 👍
So its finally adding some of the stuff that we've had in stable diffusion for months now with limited options, why people are still paying for this baffles me.
@@MakingPhoto Tech god??? Not being rude but How hard do you think it is to install stable diffusion? Theres tutorial all over the place now that shows you how easy it is so you dont even have to research it anymore. SO much Easier....?! using loras and textual inversions with SD sounds alot easier to me to get exactly what you want then all the crap you have to go through with mj ignoring your prompts and generating randomness, so im not to sure about that tbh sry. I can link you a tutorial if youd like, all you really need is a decent graphics card. btw i appreciate your content keep it up
@@snuffsterx Don't be rude. Not everyone thinks this tech stuff is easy. When I started a year ago, I couldn't install SD on my Mac. If I couldn't do it, then there were a lot of folks who would struggle. I decided AI was about the art rather than the struggle with tech. I try my prompts in SD all the time and it's so much harder to get nice images in SD than MJ. You have to work so much harder. Why???
@@MakingPhoto Wow...oook. I'm sorry if I sounded rude I really didn't mean that. Did you only call me rude because i said "not to be rude" Good job! The mac was prob the issue. I was only informing you it's gotten easier to install I literally posted a link that installs in 1 click¿¿¿¿ alot has changed in the last year you'd be surprised what you can actually do with SD including animating/posing and much more but that's fine. I apologize for trying to help I guess. I realize now people would rather empty their pockets instead of literally reading 3 sentences. Sorry but wow 👌
@@MakingPhoto what ever SD you've used previously prob uses old trained models. BTW midjourey models are available within SD so I'm not to sure you actually know what your talking about now. Maybe you not as tech literate as you think.. now that sounds rude 😆
It won't be an exact rendering of you. AI GENERATES - which means it changes. You can up the image weight --iw 2 which helps. It also depends on what the image looks like. If it's just your face, the AI will have a hard time extrapolating and giving you a full body.
Thanks for curating this update & highlighting freeflow.
No problem! Having fun pushing around these new tools.
You rock, Janet!
😃😃
Freeflo is amazing thank you
Yes! So many lovely images!
excellent video , thanks for your share!!!
Thanks! Have fun playing with v5.2
"Perpendicular to the square, I rise like Flair. Welcome to the dragon's lair."
The perp is Dracula, at day a square, wolves are were, turn around, you won't well fair.
Wu-Tang.. Wu-Tang
Thank you so much for your fast responses to your followers thank you 😊
😃😜
7:52
I've done some cursory testing (definitely not exhaustive) of SHORTEN and in it's details breakdown its claims about weight percentages of certain words .......but I've found noticeable flaws in those percentages.
For example, it claimed "0.00%" for a certain group of words but the generated images with the same prompt with only those words removed had a noticeable impact on the generated image. So those words couldn't possibly be 0.00% as indicated.
Granted, I wasn't using the same seed, but multiple rolls of the same prompt only missing those words all had more or less the same outcome.....a noticeable difference.
Anyway, I'll have to do more thorough testing to flesh this out.
I'll do some deeper testing too. I like o use the same seed with the testing.
@MakingPhoto
Yes, using the same seed is actually the best way to go and I'll admit that I could be mistaken in my conclusion. But from my initial sub-optimal testing, there does appear to be inconsistency in some of the 0.00% or near 0.00% claims.
And I'm not referring to words like "a" or "the" but rather phrases such as "shallow depth of field focal aperture blur" where if I recall it correctly, MJ weighed each of those words in my longish prompts (and the fact that they were longish could make a difference in comparison to short prompts) as 0.00 or maybe 0.01....... but in my testing, there often was an effect of the background and foreground being blurred but without that phrase the blurring didn't seemingly occur at all or it was noticeable less.
Still can’t create two separate subjects in one image.. I’m still waiting.
Are you using weights? I agree this is an issue.
Kind of irrelevant, but I know we can weight words, but how about 'concepts?
E.g... I want a cat wearing red shoes.
Would I weight red AND shoes equally? Or just shoes because it's the end word of the concept?
You get me?
Maybe MJ in the future will allow you to contain a phrase somehow ((Red Shoes::2))
Like that maybe?
I'm not quite following. Why can't you weight "red shoes"
@@MakingPhoto Because if you write red shoes::2 it's just weighting shoes, right? So being able to tell MJ that the preceding words are part of the weight would be cool don't you think? Sorry, I'm rambling :D
@@Shanghaimartin It should weight the phrase between the :: so should weight "red shoes". But to clarify, I don't think /shorten works with weighted prompts yet
Can you share the link to FreeFlo? I don’t see in the description or on a google search but def wanted to checkout
It should be in the description. Freeflo.ai
when i hit square on a single pic i get 4 new ones....
Yep. And they're all square 1x1 aspect ratio.
hi. what's promt red hat woman 01:35 thanks a lot :) she like my wife
PROMPT: "High Fashion Model --v 5.2" This is my go-to prompt when I want something in this style. You'll get wildly different details, but a similar vibe.
The most recent version of Midjourney is amazing! It looks like version 5.2 has a ton of amazing new features and tools.
Excellent video! I've been playing around with 5.2. They've really raised the bar. And they're being smart about not naming it 'v6' yet. --shorten has a way to go I believe. Thank you. 👍
I agree. I'm going to put /shorten through some more tests this week.
I've been excited for this video! You've got great content on this channel, it was an easy sub.
So its finally adding some of the stuff that we've had in stable diffusion for months now with limited options, why people are still paying for this baffles me.
We're not all tech-gods. Plus it's SO much easier to prompt a nice image in MJ.
@@MakingPhoto Tech god??? Not being rude but How hard do you think it is to install stable diffusion? Theres tutorial all over the place now that shows you how easy it is so you dont even have to research it anymore. SO much Easier....?! using loras and textual inversions with SD sounds alot easier to me to get exactly what you want then all the crap you have to go through with mj ignoring your prompts and generating randomness, so im not to sure about that tbh sry. I can link you a tutorial if youd like, all you really need is a decent graphics card. btw i appreciate your content keep it up
@@snuffsterx Don't be rude. Not everyone thinks this tech stuff is easy. When I started a year ago, I couldn't install SD on my Mac. If I couldn't do it, then there were a lot of folks who would struggle. I decided AI was about the art rather than the struggle with tech. I try my prompts in SD all the time and it's so much harder to get nice images in SD than MJ. You have to work so much harder. Why???
@@MakingPhoto Wow...oook. I'm sorry if I sounded rude I really didn't mean that. Did you only call me rude because i said "not to be rude" Good job! The mac was prob the issue. I was only informing you it's gotten easier to install I literally posted a link that installs in 1 click¿¿¿¿ alot has changed in the last year you'd be surprised what you can actually do with SD including animating/posing and much more but that's fine. I apologize for trying to help I guess. I realize now people would rather empty their pockets instead of literally reading 3 sentences. Sorry but wow 👌
@@MakingPhoto what ever SD you've used previously prob uses old trained models. BTW midjourey models are available within SD so I'm not to sure you actually know what your talking about now. Maybe you not as tech literate as you think.. now that sounds rude 😆
loved this video!
😃😍
did they sort out the messy feet and fingers?
I wasn't testing specifically. But I did notice some nice hands in a few of my zoom outs
@@MakingPhoto oh ok thanks.
Thanks!
Thank you!
wow....
😃😃
At last we can make room for framing or gallery wrapping our images! But, wow, I always learn something new from you. Thanks.
Thanks! Good point about the framing issue! I've always had to think about this with my photography.
If I used a picture of myself and told midjourney v5.2 to have me play Michael Jordan on the moon. Will it create that?
It won't be an exact rendering of you. AI GENERATES - which means it changes. You can up the image weight --iw 2 which helps. It also depends on what the image looks like. If it's just your face, the AI will have a hard time extrapolating and giving you a full body.
@@MakingPhoto ty kindly
Lots of cool new features. Thank you so much for breaking them down. 👏🏾👍🏾
No problem! Midjourney released quite a few.
can your personalized your own picture in midjourney ?
I'm not sure what you mean. But this video might help "Create Your Dream AI Images!: You Need these Tools!" ua-cam.com/video/AAPMIm4sTyc/v-deo.html
Hi, thank you. I really like the transition between different images (6:47 - 7:03) . How did you do that? Mid Journey also can do that ?
For image transitions I use Runway ML frame interpolation tool
@@MakingPhoto thank you, very cool effect! Have a nice day!
best midjourney channerl Learned a lot Thx!
Thanks, Lucas!
Your photo is mind blowing❤🎉❤
Thank you so much 😀
How cool ist it😮 Can't believe😍
I'm having a lot of fun pushing these new tools around!
8:08
😆 🤣 😂 🐈
😹🐱
🎉🙌🏻🎉💯
😃😃
hi 🥲 digital artist 👋
Hello 😊