Discover Midjourney V5's Hidden Powers with GPT-4!
Вставка
- Опубліковано 1 сер 2024
- Looking to get more out of ChatGPT? Get my free E-Book: myaiadvantage.com/newsletter
Are you tired of creating the same old boring images? Midjourney V5 is here to help! With improved functions and photorealistic qualities, V5 can replace models and bring your art to life. It may require more detailed prompts, but the results are worth it. Plus, with GPT-4, you can generate detailed visual descriptions of scenes. So why settle for ordinary when you can create stunning, unique art? Join our Discord to try it out and let your creativity run wild!
Links mentioned:
www.nikonusa.com/en/learn-and...
www.bhphotovideo.com/c/produc...
🎨 My Favorite Midjourney Course: courses.christianheidorn.com/...
CHAPTERS:
0:00 Midjourney V5 Is here
0:37 Why use it?
3:03 GPT-4 Prompt Generator
6:28 Using It
---------------------------------------------------------------------------------
Prompt (Version 26.3.2023):
Act as a stable diffusion photography prompt generator that accepts a visual description and outputs a detailed paragraph of 100 words that I can copy into my diffusion model. Always start the prompt with "/imagine prompt:". Include a variety of photography-related terminology including the description of the lens you use and most importantly a description of the lighting. Now give me a clear and concise natural language visual description of the following: [description]
Thumbnail Prompt:
Capture an image of a young girl exhibiting extraordinary abilities, bestowed upon her by a futuristic computer, situated in a vibrant and kaleidoscopic environment. Utilize a Zeiss Otus 55mm f/1.4 lens to achieve impeccable sharpness and outstanding color fidelity, ensuring the vivid hues in the scene are accurately represented. Position the girl in the center of the frame, surrounded by dynamic elements, such as floating holographic interfaces and neon light beams that showcase her superpowers. --ar 16:9 --v 5
Thanks to visualsynthesis from the comment section for the little "Always start the prompt with "/imagine prompt:"" trick
0:00 What Changed In MJ5?
3:03 GPT-4 Prompt Generator
#midjourney #v5 #midjourneyv5
---------------------------------------------------------------------------------
🧠 ChatGPT Prompt Engineering Course: aiadvantagecourse.com
🤯 E-Book with 750+ ChatGPT Use Cases: myaiadvantage.com/ebook
💬 Discord: / discord
🐦 Twitter: / theaiadvantage
📸 Instagram: / ai.advantage
---------------------------------------------------------------------------------
My Video Equipment:
📷 Camera: amzn.to/3AZVjyk
🔍 Lens: amzn.to/3nzBdYt
🎤 Mic: amzn.to/41px8Eh
🔊 Audio Mixer: amzn.to/42aQBtt
💡 Light: amzn.to/420Xcqd - Наука та технологія
For what it's worth, if you'll append "Return the completed prompt inside of a code box." It'll return the response inside a code box with the "Copy" button at the top right. Nothing groundbreaking there. It just makes the copy and paste that much easier.
That detailed generated prompt including the focal length is nuts Absolutely NUTS..
I'm an old school photographer that moved over to East Asia and my access to live models has been severely limited. I've been looking into using AI to help me generate some of the images in my mind. Your video was SO helpful in getting me started. Thank you!
Priming chat gpt with prompts before requesting is absolutely genius and the new mid journey works are absolutely stunning. I also used gpt4 for Blue Willow prompts and I got some really cool images
Thanks for taking the time to research and making it simple for us to understand.
This is amazing. The way you went all detailed with the camera lens and even goin all the way to B&H Photos to get the most expensive one, that’s genius!
Anamorphic Cinema Lenses would have been a great bonus tip too :)
Thanks, the lens info and using GPT to prompt Mid-j v5 has helped improve my images by leaps and bounds, thank you.
Loving the new setup! 🎉
this is fantastic and i respect the quality thanks for sharing.
Thanka for the excellent videos mate. Great resources.
this is ssooooo amazing!!! so helpful!! thank you
Love your content. Your news letter have helped me find new tools and just general updates in the Ai world. Keep it up man! 👏
Aw man, love to hear that :) Thanks for the support
What's the email to the newsletter?
Another Gem! I can't wait until I get paid so I can get all the subscriptions I need to do this stuff right.
The rate at which technology is evolving is simply astonishing
Well done!
can’t wait for in painting 😮
these are great!
I like how you explain everything slowly and calmly. Midjourney should begin making plans to raise their 25 free limit right away. As an alternative, AI technologies like Blue Willow will become more and more popular over time
A much needed and helpful video. Thanx
Lightning fast comment ⚡️
@@aiadvantage I have been testing midjourney v5 since yesterday and that video is a great addition.
@@JoDi2015 It's my fav starting point to craft longer prompts. Share your results in the discords if you feel like it!
Thats really cool I'm gonna build something similar into a new app i'm making would be thrilled to have you try! I got access for GPT-4 full api so there's some interesting things that can be done!
MJ v5 is just in Alpha right now. It's thrilling to imagine how the AI will be in the future.
Subscribed!
Thanks
Hey Igor, I just watched your video and I must say that it was really informative and well-made. I loved your videos. I was wondering if I could help you edit your videos and also make highly engaging shorts for you.
2:40 It's not that the image upscales really fast, but rather the 4 results are already upscaled. They said they are working on further upscaling images to even higher resolutions but it's not implemented yet.
these text to image ai systems will use gpt4 to filter unusual pictures like weird fingers. Because gpt4 has eyes and detects unusual details. That's gonna be how gpt4 will actually revolutionize ai art
Wow genius 🎉
@@joshikaustubh9445 thanks bro
Fingers are ALWAYS CURSED AF
@@TheCUBARY nightmare fuel ⛽️
Lol u can actually use gpt4 to enumerate all combinations of these ai’s and how they can b combined kekekekke
thx igor
Thanks for the video, however, I am having an issue when using the same "prompt" you have used, rather than ChatGPT outputting as it did yours, it is acting like it is a script or scene including narrator text and closing shots, etc, any advice? Thanks.
Thanks for sharing this useful information! Does anyone know how to use - - v5 and certain aspect ratio in the same prompt?👀
That moving shot to the ship in a bottle was a flex
:) my favorite shot in there for sure
So Midjourney themselves said that it basically ignores everything after roughly 60 words, this was before V5 however. It's probably worth finding out if it's still the same, but it's something you should take into consideration while prompting.
Is there a way to use the generated image and further prompt it to update just the hands or a specific part of the rendered image when there is an 'error'?
Use remix setting and --no to remove things
Thank you so much for your content, you do a great job of explaining your process in a concise way that is easy to understand. I'm having trouble using my own uploaded photo in a prompt not looking very close to the original person in my photo when the Mid Journey art image creation is finished. Does the prompt process you describe in this video help me with my issue?
So you are using the image link as a part of the prompt?
cool channel bro
Thanks Bb :)
Great info! I used the first prompt with my own description and was flagged for a possible content policy violation. Nothing in my description seems to get even close to Open AI's content prohibitions and ChatGPT won't (can't?) diagnose the problematic content. If you know of a good resource for content policy trigger terms, please share. Thanks for the excellent videos.
With the new /shorten command, it's looking like the devs are telling people to look at how little of their prompts is being processed. Have you tried /shorten on these paperbacks? I've used the Bing chat bot to generate prompts for MJ. They're shorter, and are (I hope) what MJ is expecting. It doesn't seem to me that MJ does well with normal speech patterns. This is a command line interface, and we need to understand the parameters and keywords, and how they are interpreted, to get good results.
should have described your ship in bottle a little bit: bottle lying down lengthways, 4x4 white horizontal sails plus one at front, light blue sea and red base. all on a light wooden table
Absolutely! As mentioned I didn't edit much in the prompts. In practice you will want to do exactly what you suggested and use the GPT-4 output as a starting point.
the link to the differnt Lens types has stopped work :( is there another one?
oh wow you are right. thanks! Here is a new one: www.nikonusa.com/en/learn-and-explore/a/tips-and-techniques/understanding-focal-length.html
Love the prompt. I have a question. I want to use the prompt but have the subject 10 to 20 feet away. How would you do that?
You would pick an appropriate lens focal length. Maybe a “24mm” could do
@@aiadvantage thanks
It would be hilarious if you pressed enter after the last prompt and it showed an exact photo of you sitting at your desk
ahahhaha what a missed opportunity
If I purchase any Plan on midjourney, can I able to sell (commercial use) on sites like Shutterstock, Adobe Stock, Freepik, Getty Images...etc,.?
According to their terms of service, yes.
Thank you for this very helpful video. I have one question about ChatGPT Plus or Version 4: Are those versions free or do I have to pay for it? Thank you.
Yes you need to pay for those right now. ChatGPT Plus is a monthly subscription and you need it to access GPT-4 through the browser like I showed.
@@aiadvantage THX
When I can create long form video (1h plus) with text to video - ill be impressed.
Wait you use a bag image took a your phone and make it batter?
I used your prompt with Chat GPT and then added "a campfire in the evening next to a cabin by a lake with fireflies flying around". The GPT output with Midjourney was not nearly as good as just telling midjourney to imagine a campfire in the evening next to a cabin by a lake with fireflies flying around .
Thats entirely possible. This gives you more control withiut having to lay out every detail yourself.
How do I make it stop scrolling up to where I lose my image amd have to search for it? That is so frustrating
either use a less crowded room (my server) or communicate via dms
@@aiadvantage Excuse my lack of knowledge, but what is the option to do so? I've played around with it and the only options I see that I can use to create pictures is the "newbie" servers. What do I need to look for when looking for a less crowded room?
great content, but check your audio compression and mic work to improve the audio.
Absolutely! Appreciate the pointer this was a little mishap on my side (changing up the setup). Already fixed it.
Thanks for this informative video, since the GPT hype, I'm looking fromward for every video from you.
I want to pay for the GPT plus and check out its cool features, however, I'm encountering that card is declined every time even though I used multiple cards and all results in failure, any idea on how to overcome this issue???
You can use a virtual credit card
@@hfycentral which websites offer these cards??
Interesting, the first try with visa worked for me so not sure how to help you.
@@aiadvantage best guess is that I'm using gpt from non-supported region, therefore, card is not acceptable, even with vpn 😟
🤯🤯🤯🤯
Do you have to pay for it?
And how do you get both midjourney 4 and 5
Yeah MJ5 is only available to subscribers. It's a simple "/settings" command in a server that has the bot to switch
i just started using GPT but where is the button to get to GPT-4 ?
you need chatgpt plus
Sir, midjourney had a trial period I have lost it! now what I can do, I loved midjourney than any other ai art generator.
Pay the $10 😊
@@Logical_Human_ not quite. 200 generations a month I think. Roughly
Have you done any A/B testing with lenses? I don't see any difference by changing 85mm f/1.8 lens to 20mm f/5.6 lens. In real life rhe difference would be huge, but MJ seems to ignore camera settings. Changing Zeiss to Lomo makes a difference. Also: GPT creates long prompts and MJ5 seems to ignore half of the prompt. "volumetric lighting" in end of longish prompt makes no difference. If you add camera brand like Canon or Nikon then you will most likely get image of a camera or person holding a camera.
Lenses seemed to make a real difference in my testing. But it's correct that sometimes it just ignores certain parts including the lens. It's usually a multi-step process to arrive at something final but to me this detailed description in natural language is the ideal starting point when I don't have a precise idea.
In my experience, mj doenst weigh lens type that much. The camera brand does make a difference. If you choose a film camera, the film make also has a certain influence.
@@realricardoruiz you tested this with mj5? I found both camera and lens to make a real difference
@@aiadvantage Can you make a video of testing camera setups so you can replicate the reasults? You can get image look different by just adding random letters/words, but to replicate the result is harder.
Could someone please tell us how you get midjourney 5???!
Please
I told and showed you. You need a payed plan and then type /settings
I think your mic is very sensitive for sounddirection. Volume changes a lot.
Midjourney should begin making plans to raise their 25 free limit right away. As an alternative, AI technologies like Blue Willow will become more and more popular over time. Blue Willow, in my opinion, will be the one to take down MidJourney
Does it take long to generate? This is actually so cool! What are your views on BlueWillow ? please share tips on this new tool next for us designers currently using it. That would truly be amazing 🤩
you need to readjust the sound compressor, the attack is too slow, sounds terrible... but the content is great, thanks
Done! Thanks man 🙏🏼
no wqhd on your videos?
What i hate is that its a discord application. Everyone is able to see what you generate. Discord truly is the limiting factor for this technology.
you can use it in dms with the bot or just invite it to your server
Use stealth mode
@@sarah-dz1ly thats not free.
Can i use the normal version of chatgpt ? 😢
absolutely! just found gpt-4 to be slightly better for this (richer description)
Do you use a limiter on your Mic? Remove it, when you're in a silent room. ;-)
Looks like the preset got messed up on this one 🥲 will do
If you add "always start the prompt with "/imagine prompt: "" to your instructions it'll save you from having to type /imagine all the time
What a great little trick I did not consider. Thanks! Updated the prompt and gave you credit
Your audio settings seem a little off. Your mic gate seems to continuously open and close. Your voice is getting louder and quieter through the whole video. Might take a look into it. Good video though :)
Yeah I noticed that and when you notice it you can't unnotice it and it's so annoying 🙈
I used this technique of using gpt to make longer more detailed prompts for V5 and I know they say the longer prompts are better but for what I was using it for it turned out not nearly as good as my short 10 word prompt
I think you make a good point! It’s another tool in the toolbox but I find it to be a fantastic starting point for precise outputs.
Sites like Facebook will soon be able to “see” what photos are harmful so content moderators aren’t required anymore.
There’s thousands of people employed in that horrible job and huge numbers end up having PTSD or other mental issues.
I did it for 1 week and left. I couldn’t handle it. Some of the shit I saw still haunts me 😞
How much did you earn? And if I wanted the job how do I get it?
👋
I liked the Video but the Audio was weird, big highs and lows where your voice would dip out like you moved your mouth away from the mic, which made it hard for me to focus, good content though, I subscribed.
I tried this and just got a bunch of pictures of cameras lol...
Excellent content! You could EQ your mic to get better audio. High frequencies are too quiet, the deep ones way too loud. Your mic reacts very sensitive to small changes in distance to your mouth. Just listen to your voice without looking at the video. Apart from that thumbs up!
I totally agree! Messed up a bit as I was changing my setup throughout the last week. Fixed from now on.
@@aiadvantage Great! It'll be a pleasure watching your videos with even better quality. Keep up the superb work.
just a tip your voice fluctuates in volume. try to manually adjust it so it is the same volume and doesnt go up and down all the time
You can automate this using a compressor. TDR Kotelnikov is a great free audio compressor plugin that has tutorials on YT.
Might be a compressor issue, just a guess.
Seems to be an issue with your audio in this video. Maybe its the mic, keeps going up and down slightly.
Yes its a compressor on the soundboard that went rogue. Already fixed it but appreciate the pointer
jailbreaking chatgpt to response like chatgpt4
How they will improve the image generating?
Its extremly complicated. Many pictures with so much problems, 7-8 fingers on hands etc..
Blurry parts and so on. How they fix it?
Keep in mind that this is not the final release. MJ5 is still in alpha so I would expect it to get better over time.
Its not about creating something great but creating a certain something you WANT to create. GPT cant help with that.
something is wrong with volume in your video, some gating, or compression, check it
Already fixed for future uploads. Been playing with the setup and this escaped me.
Not sure if it's just your video but the audio of your voice is pulsing, might be a compressor issue if you're using one, I've heard the term pumping in relation to compressor settings.. not an engineer, but it was audible enough to make me turn off the video..
We made major changes to the setup and this slipped my attention. Already fixed for the next one!
ALSO, NEGATIVE PROMPTS can have NEGATIVE effects on the output, it all DEPENDS on which model you are using and HOW MUCH negative you are using. Sometimes NO negatives gives BETTER result, sometimes a SIMPLE negative word fixes all, and sometimes LOTS of negative is NEEDED if a model requires it, , but not always......
I am creating anime characters and the upscale takes a long time.
"Looks like a real photo" while it looks like a strange 3D render. Much hype here?
What ablut that cat with a hat!
I predict hand artist becoming a highly paid job in the not too distant future... Seems like these models need to be trained on good source data 😅
this is an important comment.... from what i underestand this AI just bases its output on other peoples work. When people get pissed off because AI takes there vision (and income) and stop creating what happens next?
@@hairynoggin it's all fall flat on its face. Because ai art is parasitic in its nature
Why photographes need midjourney?
I keep getting the AI putting GIANT images of cameras right in the middle of the images. it's really annoying lmao
How come UA-cam videos now a days are, 98% talk and 2% content?
Your audio keeps oscillating. It's making my head feel like it's spinning.
I think you exaggerated with the excessive speed of your voice. The video becomes virtually impossible to follow. I understand that public attention is getting scarcer and there are more and more UA-camrs in this space trying to hoard this valuable resource, but there are limits to everything. Removing absolutely all the pauses in the speech and perhaps also artificially increasing the speed, for me it's too much, and I pass.
seriously it seems that people who are fans of midjourney have no eyes ...no matter what you do nothing goes in these pictures nothing ..at first we have the impression to have something and then ...we look and any child can realize that it's just not a normal thing, the proportions have no sense, there is no details, the anatomy is completely fucked, the colors come out of a dirty ass the general sense is distorted in short nothing goes . .apart from the people who hate the others and who hate the image and who just need fast and tasteless content I don't know who can say wow except someone who has never seen an image in his life
This is what's happened when talentless people, with no knowledge, understanding and experience get the keys to ferrari. Non of them see mistakes, because of those things.
On word, AMAZING 😄Can I use BlueWillow for this process? Please share for us beginners who just started using this new tool. Your insight would be amazing
Sorry, but it has been tested MANY times, if you use a ChatGPT generated prompt, PLEASE EDIT IT AND REDUCE the excess.... no NEED to use a prompt like the ones you are showing. YES, we can use MORE words than before, and YES, even without editing chatbot you will get NICE results, SAME as NOT using a long paragraph...... you just need to do BETTER DETAILED promtps, not "LONGER" prompts.,, and i apologize but its just too much text for achieving something THE SAME as fewer texts.....
Man’s been smoking why too much, that does not look like a real photo bro
The sound of your video is so bad my dude, the compressor is insane and it sounds like my ears are shutting down everytime you start talking. Just a tip
Appreciate that tip! It was just on this video though.
Stop Promoting GPT-4 this AI is so Stupid, it fails with simple tasks, idk why you are Promoting still such a Bad Product
Can you be more specific? I personally find it extremely helpful on a daily basis but your job might differ.
Oof. AI art, the epitome of talentless hackery
I fully understand this is just my current gut reation... I could be wrong. But as an artist, I feel this is tech signals the end of human art. Sure people will still do physical artwork because they love it, but no one will care and no one will pay for it on a commercial basis. This makes me extremely sad. As I mentioned, maybe theres something I dont get, or maybe theres an angle Im not understanding. But it really gives me a terrible sinking feeling for the future.
Man... I think that you have a very real and realistic point right there. Not much we can do about it though. It's simply put a new era.
On the bright stuff this stuff is super fun to play with and you can literally 100x your creative output. It's just a different discipline.
@@aiadvantage o for sure - can’t put this toothpaste back in the tube, and hopefully it’s early enough that people like me can pivot in some way
You see, it's all parasitic in nature. When artist stop creating new styles, techniques and designs, it will start cannibalising itself, creating never ending loop of grey unimaginative goop of the same looking content, as non of these people here, have any talent or ability to create something from the scratch. At the end, you will have the last laugh