Midjourney Version 4 is here of Alpha Testing and it is absolutely insane. In this Video I will show you want it can do and what is still problematic. The styles, anatomical correct bodies and designs it can create are mind blowing. Midjourney V4 is also a completely new model. In Midjourney V4 you will get 4 previews of stunning quality. And Midjourney V4 will bring more details, in correct perspective and beautiful lighting. The rendering with Midjourney V4 feels very fast too. I love it!
Follow Robomar: https://www.instagram.com/robomar.ai.art/
Support my Channel:
https://www.youtube.com/channel/UCCKx8mAHiFus-XYQLy_WnaA/join
Subscribe to my Newsletter for FREE: My Newsletter: https://oliviotutorials.podia.com/newsletter
How to get started with Midjourney: https://youtu.be/HQLnB6qLvUg
Midjourney Settings explained: https://youtu.be/jYfULc99XI8
Best Midjourney Resources: https://youtu.be/I-EWJ8MFNks
Make better Midjourney Prompts: https://youtu.be/q1taNR3V6Fc
My Facebook PHOTOGRAPHY group: https://www.facebook.com/groups/oliviotutorials.superfan/
My Affinity Photo Creative Packs: https://gumroad.com/sarikasat
My Patreon Page: https://www.patreon.com/sarikas
All my Social Media Accounts: https://linktr.ee/oliviotutorials
Support my Channel: https://www.youtube.com/channel/UCCKx8mAHiFus-XYQLy_WnaA/join
Follow Robomar: https://www.instagram.com/robomar.ai.art/
Big Ben is standing in the middle of the road. 🙂
Those images are truly amazing!
I'm one of those "local install SD boys", but seeing what the capabilities of the new version of MJ are amazes me!
Version 4 is indeed insane, I friggin love it!
dreambooth + midjourney v4 could break the internet
version 4 i feel is a bit cartoonish rather than realistic
Wow!!!
Aspect ratio can’t be used with v4 yet 😭
i found niji , how to get it .
I haven't used MJ for quite some time. Love the images it creates but it seems you have little control on the outcome and feels more like the roll of a dice. Correct me if I'm wrong but can you actually reproduce a consistent look once you've created a set you like? Now if they could only have it run on a regular webpage rather than Discord I would consider using it.
You can get 2048 x 2048 by upscaling and then beta upscaling. It does slightly change the resulting image, but it's a good workaround for now.
V4 is insane, it's so good.
Hi Olivio, excellent video. May I ask what the prompt was used for the young realistic girl at the beginning? Im curious what you wrote for a prompt for an image like that?
the AI is already in the REM phase of their sleep
SD has some catching up to do
I saw a couple of mention mention it but maybe adding the prompt would be great, I am not able to get as good results but I know is cause my prompt game is weak haha
Thanks again, great video as always.
Do you make your prompts public? Some of it – like for example the colored face in the beginning – would be interesting for me to play with in StableDiffusion.
Olivio – Great work.. as always
🛠🔥👌
It still gets hands wrong 🙁
Killer images. Now when this tech is combined with real 3d space based compositions created by AI…
I second having the prompts shown. I've been loving V4 since finding out about it today! Rapidly going through my remaining 14.50 fast hours 😀
Hey dude, check your Instagram handle on your YouTube profile page, I'm not sure it links to the correct account?
its motherfucking incredible !!!!!!!!
Arggggg, I had it for two months and cancelled my subscription because I generated everything I wanted to and couldn't think of anything else to generate. Looks like I'm subscribing again lolol
Details!
Wow awesome, mediocre people can create "art" with words and real artists are losing their job…. so exciting….
At first I was amazed, but I quickly grew tired of these AI images. No matter how sophisticated, they are all surface, no structure. Garbled details that look correctly lit, but are nonsensical and wrong. They are learning, however…
And then there is the matter of copyright. Nobody was asked if their artwork may be put in a learning bundle for a machine that is training to replace them. The huge corporations that are pushing this sinister shitshow are admitting that they would never do the same thing with music. Big labels have expensive attorneys, while the artists are a clueless bunch without coordination, that are even happy for the opportunity to help train their own replacement for free…
hai,, found your youtube, thank for invite me in nijijourney
I stuck with MJ Test and have noticed a considerable increase in quality of facial details.
Took mah jerb!
It's amazing but real shame about the square-only thing. That makes it useless for me.
My first tests with version 4 looked actually much worse than the results I got with version 3. Okay, I didn't make very elaborate prompts, I'm more using simple prompts, but I didn't do complex stuff with version 3 either… at least most of the faces I got are worse than before.
what did u promt for your image into painting one 18:30 ?
I'm only a few weeks into following what these AI generator are doing… and I am stunned at the speed of progress. The fact that this is V4. My head can't wrap around the advancements. By v20, we will not be able to distinguish photoreal images from real images. IMG2IMG will be the real area I'll be watching. Slightly morphing real images into fantasy is so fascinating.
it would be interesting to compare the same prompts with v3 and v4 to see the differences
Unbelievable how fast it is going. Peoples' commentary can't even keep up with the pace of how fast this is growing and improving.
This is getting to the point where I don’t have to edit these images anymore lol
Can you tell me why I keep getting a message that says that version 4 is not compatible with
-v 4 –stylize.Oh wow some exquisite images coming out of V4 blimey the detail is superb! The cyborg girl at the beginning is particularly of interest >> Really looking forward to testing this out with prompts I have used before, to get a really good comparison… perfect thank you 👌💯👀😎🌟🖼
Glad I got a subscription that it is looking more and more polished.
It's so detailed that I'm afraid to use it. Don't want a random artist to go copystrike me.
Do you think we will ever get more logical text on minjourney? It seems it can't replicate words very well
If you train your own models then hands no longer become an issue
Olivio, how do I use V4? I tried to use it, but it doesn't allow styles and wouldn't let me use it.
artist in industry and business became obsolete. now we will have art technicians. no more studying perspective anatomy etc but manipulating different programs for products. this displasment going to be painful. the q is will the small private ppl will be priced out with access to the final ai programs only to big conglomerates like Disney or blizzard that can put the cash or 2 years from now is a world where anyone can do animation and games without knowing art.
My God ! …. I never new Zombies can run. Super helpful video Olivio
Still struggles with hands. Do you know of any way to stop writing appearing in the shots?
i luv how it tries to add Bee's…..from Zom..bies '')
Anyone have documentation on how to reference your own uploaded image?
oh my gosh… oh my gosh
hay que pagar 35 pavos para usarlo
Great video!
Olivio, I'm learning to make these arts and I have a very big doubt:
Is it better to work on Midjourney v4 or on Stable Diffusion??
BTW, you don't have to go into the bot DM to use /setting . It will work in any channel that you can generate images in. –your friendly neighborhood MJ guide
the MidJourney with Remaster option on V4 is amazing. Just a simple prompt like "The Milky-Way galaxy is a glass bottle –q 5 –s 625" produces mind blowing stuff
Insane… Now I really do think visual artists will have some serious competition from AI.
The faces have definitely improved. Although everything kind of still has that airbrushed look.
Does v4 stop fucking up limbs and adding hairlips? I swear it is better at making nightmare creatures
Midjourney is paid right no way of running it locally.
Amazing. Midjourney has somehow learned what we think is beautiful, I guess? And appears to default to just pleasing your eyes, adding details and elements most people will simply enjoy. Really different from other models. It's as if they've taught it with opinions – whereas other models seem to have been taught mostly with facts? 🤔
V4 is amazing. Does anyone know what terms / phrases to use in a prompt when making portrait images so that the entire head or body of the person is included in the frame of the image (no head or hair sliced off at the top? I've tried 'zoomed out', 'in shot' , 'in frame', 'cover art', 'wide angle' 'full head' etc but nothing seems to work reliable. Any suggestions? Thanks!
May I ask at the current time which AI would allow to replicate the same “character/subject” over and over in different poses/scenes/locations/scenarios? In short, if there is a software that may be used to character (I.e. a robot, avatar, goku, whatever..) and then use it to create sketches of the same character performing different actions. Very nice video thanks for sharing it
Wtf this is amazing. Like a computer game upgrade of 20 years.
These images are mind-blowing, all of them, if you told me they were created by a human artist, I'd totally believe you
Man becomes unnecessary.
Indeed insane
I see how all of OnlyFans “creators” are loosing their income 😂
You talk of this very well. It would be nice to see the actual prompts used though.
Will it free for use some day?
I wonder when they gonna fix hands
I keep getting the error: "–version 4 is not compatible with –stylize"
But there doesn't seem to be a way to disable stylize in the settings?
Is there any way to run a stable diffusion, even very shrinked version of it, on an old videocard with only 2gb of video memory?
I just wish I never believed that the human being had any value,
I should have dedicated myself to being an office worker from the beginning,
dreaming is forbidden in the future
why bother learning art anymore?…so I can be fuel for the machine?
We are just at the beginning of all this and it's going to be an amazing ride. We will look back on this in a few years and see how basic this is but for now it's the best we have.
That's pretty awesome. I would ask less of the AI, because I want to start with my own artwork and photos, and have it generate things like costumes and makeup and lighting. Once, I priced armor for my model. and the breast plate started at $400.00.
Try this:
Create an image using –v 4
Upscale it to the max you can have now.
Ask a new version, and change –v 4 to –v 3
Now you will have the an image with the same command but the mind of the AI. v3. So the image can be very different. And it has the strange zig-zag lines that V3 has.
Ask an upscale.
Ask a Remaster.
Now the lines are much better, and the image is more clear. And you can upscale to the max.
You will not get back the image that V4 made. You will get something very interesting (that sometimes better then the V4, sometimes worse)
V4 is even better than Dalle
how did you do the last animated part?
I loved the results I was getting out of my phone until I checked them on my desktop monitor, I knew they were smaller than previous versions but these were very fuzzy. Anyone else ?
I am baffled
😮🎩🎩👍👍👍👍💕💕💕💕
I am so close to giving up my dream of being a digital painter. I feel like I am observing being robbed of my life in real time.
Let's hope for a leak or hack, I'm not paying a dime to generate my waifus. Not to artists commissions, not to AI companies.
This video is sooooo good, I would love to see what prompts you were using to get some of these images!
How do you access v4 ?
OK!
I just tried this prompt: juxtaposition (object 1) and (object 2) and it worked😀
Glad I didn't spend wasted time learning PS! 🤔
It sucks and always will because it can't create anything. It just takes from art that already exists… Isn't high resolution or anything. Anyone that thinks its good is a total noob.
You can see some great examples of the top trending images on the community showcase page. I put together a slideshow of some of the top creators. This diversity of what people come up with is amazing. And to think this creativity would be lost by those without traditional art training. The creation of Logos is also very interesting. Keep up the good work. Love your videos
yep – eyes are much improved
Great videos. I like the first rabbit best. 🥰
I find that Midjourney V4 gives a lot of text on the pictures. I have tried –no text, –no words, and –no script but nothing gets rid of it. Does anyone know a good way of stopping it? It never happened on V3.
This is amazing but as an artist this is pretty terrifying
i want to join your server
Insane!
Midjourney is the best. Now it's possible to create fine art which will blow your socks off!