Learn to Re-Create EXTREMELY Difficult CAMERA ANGLES in Midjourney
Вставка
- Опубліковано 3 тра 2024
- 📙 Midjourney COURSE mastersofmidjourney.com Beginners to Advanced
🚀 FREE Midjourney Cheat Sheet tokenizedhq.com/freebies/mj-c...
🔗 FREE Supporting Material tokenizedhq.com/freebies/vide...
Folder: 2023-08-04 - Impossible Shots
🔗 FREE Promptalot Extension promptalot.com/extension
-
📰 AI Newsletter 👉 Your Inbox tokenizedhq.com/newsletters/ai
🐤 Follow me on Twitter / chrisheidorn
💼 Follow me on LinkedIn / christianheidorn
📸 Follow me on Instagram / christianheidorn
💬 Join the Tokenized AI Discord tokenizedhq.com/invite/discord
-
📺 Recommended Related Videos & Playlists 📺
Commands, Syntax & Parameters: • Commands, Syntax & Par...
Prompting Methods & Experiments: • Prompting Methods & Ex...
Character Design Series: • Character Design with ...
Storytelling: • Storytelling with Midj...
Watch Me Prompt: • Watch Me Prompt in Mid...
This Brand Does Not Exist: • This Brand Does Not Exist
Midjourney for Merch: • Make Money with Midjou...
-
Imagine you're working on a project that requires very specific camera angles.
So, for example, things like a graphic novel, comic book, a movie pitch, or perhaps even an entire storyboard for a film that's been green-lit.
One of the tricky parts about Midjourney is its element of randomness.
This randomness is part of what makes Midjourney so fascinating and incredibly appealing, however, in certain situations, it can be a problem.
That's why it's so important to have a solid foundational understanding of how Midjourney works.
You're going to have a tough time controlling Midjourney if you don't understand how to give it the "instructions" that it needs.
In today's video, I show you a reasonably reliable method for re-creating extremely specific camera angles in Midjourney while still maintaining the style that you need.
You know, the kind of stuff that some of the most iconic movies of all time are made of.
Now, be warned, this isn't going to be a simple process and it will require flexibility on your part.
But once you get the hang of it, a world of opportunity will open up to you.
-
⏰ Timestamps ⏰
00:00 Creating Impossible Shots
01:00 Aligning Our Settings
01:37 Example 1 - Man on a Ledge
11:11 Example 2 - The Iconic "Frodo" Look
#midjourney #midjourneyart #midjourneyv5.2 #storytelling #storyboarding #aiart - Наука та технологія
Awesome contribution...thank you.
Thank you for the extension. Thanks for all
Many thanks for your sharing
Amazing work!!! Thank you for sharing your creativity with such knowledge and patience!
Epic, informative and a great workflow as usual - Christian you're the best, thank you!
Absolutely brilliant video! 👏Great job taking it to a whole new level. 👍💯
That’s fantastic! 👏🏼
Awesome tips!
This is gold. Very soothing and calm voice and to the point. Thanks mate. God bless you 😊
It has been an amazing and extremely valuable video
Excellent video, as usual! :)
Thank you for uploading this. I've been trying really hard to do some fun angles for a while now, glad to see I'm not the only one finding this difficult in Midjourney. Hopefully one day Midjourney will make this more easier to do.
My pleasure:)
oh thanks so much for sharing 👌👍
Excellent. This one combined with faceswap and/or character consistency would/could/should be great. Maybe those great character consistency videos you´ve made can applied to this?
Great information
Incredible work on this tutorial! I've been searching everywhere for guidance on camera control, especially for flat front, side, and back shots. Midjourney always seem to add unwanted perspective, leaving me frustrated. But your clear and insightful approach has taught me some fantastic techniques. Thank you so much for sharing your expertise!Other tutorials
Thanks for this Video. Always learning something here. I made a similar picture from one shot in Die Hard where Hans Gruber falls down. It was challenging but fun to create.
brilliant!
Glad to find your channel...love your videos
Wow . Thanks
Great job sir thanks for your efforts, you're great at explaining stuff. Do you think with the new "Vary Region" it's gonna make solving the issue of the fingers/black bars for example way easier nowadays!?
I would love to see a video of you trying to emblement that Vary into your great artistic works like this video
this is just perfect!! Thanks so much!
I think image-to-image can be great for angle shots.
Excellent ! I guess you could take the image with the black bars into the latest photoshop beta and use the expand crop tool retaining the 16:9 aspect ratio, and see if that helps.
Sure. You can edit all sorts of stuff. I guess I was just trying to avoid it because tons of people who use MJ have no idea how to use an image editor.
@@TokenizedAI well put:-)
Liked this video! Regarding the black bars on your picture... Try custom zoom with same aspect ratio and set zoom to 1. Midjourney will attempt to fill in those black bars. Doesnt always work, but pretty often it does.
Yep, seems like that would fix it.
Can these same moves be done in Leonardo Ai?
10:43 What if you wanted to add a specific character to the scene? Would you then use the face swapper to make that happen?
"this picture of a car under a bridge from a movie.."
...that movie would be The Matrix.. not just some movie. The Matrix.
What's your point?
I found that using custom zoom at small increments like 1.1 will remove black bars
If you have Photoshop with AI you can easily just generate the rest of the scenes that black bars are covering. Great tutorials!
Yes, I am aware. But don't underestimate how many people who use MJ do not have image editing skills or even an image editor for that matter.
@@TokenizedAI Fair point! I learned so much from you about Midjourney, thanks for that. Keep expoloring!
Another great tutorial - thank you. Earlier this week I just could not get someone standing underneath a cascading waterfall....I'll give that another shot. Just one question though - why Style Raw? Is that just for the graphic novel look or another reason?
It is “Style RAW” so that the model is not biased and works closer to your prompt; it requires more specific prompting, but for more specific results
@@ayaneagano6059 Thanks. Yep, I did a bit of experimenting after I posted this comment. It's not something I need to use often, so wasn't really familar with it :)
What @ayaneagano6059 said.
Photoshop 2023 (Beta) can easily help you to get rid of the two black bars with Generative Fill.
Just be sure that the selection you make around the black bar has its longer side of no more than 1024 pixels, in order to maintain the resolution. If the image is bigger, you can still use it by creating more chunks of 1024 on the larger side. It works perfectly.
Yeals, but don't underestimate how many people who use MJ don't have access to an image editor or don't know how to use them. If it ever seems like I'm sticking within MJ religiously (for no apparent reason), then that is the primary reason.
When I heard impossible in the beginning, I was going "yes, the feast is served" 😂🎉
Speaking of impossible... I want my characters consistent and combine it with impossible shots 😂 what do you think, is it possible?
Is it possible to build a fashion model scene for a e-commerce site but with using our apparel designs?
Depends...but not really. Not with MJ.
As cool as it is without possibility to use exact products it's usfelulnes is very limited. What use are non existing whisky bottles?
You're commenting under the wrong video.
@@TokenizedAI yeah no idea how that happened, sorry
That "Treasures Beyond Gold" thing looks GORGEOUS!!!!
Thanks for the help with this. What I find frustrating is that when swapping faces with InsightFace, the new image cannot be edited as a Midjourney image. Which appears to require adding the image into a new prompt which usually results in an image too far from the original to use. Any suggestions?
Well, it's an entirely different software after all.
Imagine a prompt syntax where you can specify what aspects of an image prompt is used and not used: style, texture, color palette, layout, character pose, scene etc. Even better, imagine an interface where all this mad iterative fiddling was more controlled. Mid Journey has great image quality, but a terribly imprecise and inefficient interface.
Im afraid MJ is the most user friendly tool we'll have for a while, but who knows, something out of the blue might suddenly launch
13:00 _and I think I'll go with the top-right one._ *No, no, no, no, no!* Please go with the bottom-left one with the smiling trippple chin and the extra palm mixed in with the fragments on the floor. 😁
🤣
Thank you and bravo for this video. It is very informative.
I have a question please: can we import film images on Midjourney without fear (I mean concerning copyright)... You seem to be doing it, is it a common thing or is it better to use the stealth mode?
Fear of what exactly? I'm doing this for educational purposes.
@@TokenizedAI Thanks for your reply. It's not a personal attack on what you do in this video. I'm new to Midjourney and I'm trying to do well. I'd like to know if it's common for users to use images from movies (or whatever). I don't know what the laws are and I'm wondering if these images can be copyrighted? After that, I'm probably asking myself too many questions...
No worries. I didn't read it as an attack. I guess I just don't see the real risk here. The only reason anyone knows what image I used here is because I told everyone.
Plus, I'm not sure anyone knows how to interpret this from a legal standpoint. That's kind of the point.
@@TokenizedAI Thanks for your answer, it helps me. Like your video. I've already bought 2 Midjourney training courses that I'm trying to apply. I think yours will be next.
The course says more updates on the 11th...
So, once am signed up I wud have liferime updates, as MJ evolves? (Amazing value, if yes)
Or is this training i need to download?
Essentially yes, within the scope of what the course is meant to cover.
Hi I’m very new to midjourney and I’m trying to create a hybrid of a pair of handcuffs and an engagement ring that merges the symbolism of an engagement ring and a pair of handcuffs. I want to highlight the contrast between the symbol of commitment and constraint. However, I’m noticing that midjourney doesn’t handle things like fingers or groups of lines or strings like spaghetti or ropes very well. I can’t even get it to create a simple police issue handcuffs, even if I use an image reference. Can you please help me with a prompt that might achieve that result? I want it in a comic book type line drawing with no shading sort of like a coloring book art. Thank you in advance.
has this tutorial been superseded by sref and sw? thanks.
No really. It serves a very different purpose.
amazing video, thank you! Trying to create a motion blur photo of a face but I just cannot get anything working even when I use describe and upload a motion blur face Pic. Midjourney always "fixes" the face :/
Why not add "motion blur" to the prompt?
Since you're now tinkering w/ Stable Diffusion - you can save time on rerolling here, by just taking your Pretty decent results but w/ whacky hands, and inpainting the hand in SD... or even better, inpainting the hand, but using a depth map image controlnet.
You can also "outpaint" using controlnet - to get rid of the black bars.
Yes, I realize that. Don't need SD for that. Asobe's generative fill does this quicker for me.
The reason I insist on doing it all in MJ is because the majority of my viewers want to do it all in MJ.
I don't understand why you didn't use the seeds reference if you wanted to accentuate the style ?
Many roads lead to Rome...
And seeds do not control style.
is there any way to midjourney draw a certain type of object, equipment or instrument in a scene, following some image reference for that certain element, but without spoiling the rest of the image? I'll give you an example, we generate an illustration of a rebel in a cyberpunk style, and we want one of his arms to be replaced by a chainsaw and the other by a mechanical claw. If we try this directly from the prompt or using reference images of a chainsaw and a mechanical claw, it tends to mess up the image, mixing the graphic elements in places that have nothing to do with it.
What you want/need is inpainting. That's coming soon.
20:28 Instead of using screenshots from movies, would it be possible to create these compositions using 3D models instead?
I’m uncertain about the legalities of using screenshots from movies or stock photos that I don’t own the copyright to as references (unless they’re public domain). For difficult shots like this, I’d like to use images that I’ve created with 3D models, or photos that I’ve taken myself.
Sure, why not. You can use any image you like.
You talk in this video about another video about the button "very strong", could you just paste the link again here please ? cheers
Probably this one? ua-cam.com/video/lCFzMnBDqEc/v-deo.html
hmm very good resource although i have come across a issue (Invalid parameter
Unrecognized parameter(s): --style, raw, --ar, 16:9
/imagine Graphic novel illustration, aerial view of man laying on floor looking at ceiling, lovecraft, dark, --v 5.2 --style raw --ar 16:9 --q 2 --s 750) for some reason it keeps trying to put comma's in between
Who is "it" in this context?
@@TokenizedAI ahh midjourney seems to want to put commons in between style and raw ar and 16:9
@@TokenizedAI and if that to show me where the mistake is then not sure. Because after i removed those words it created the art just fine
After I tried following this video, the extension bar suddenly disappeared from my Discord. Any idea why? Promptalot is still on according to the button in my message bar.
Have you tried reloading the page? That usually sorts out the issue.
@@TokenizedAI I did, but after restarting today, it's working again.
The "long street"s in New York City are the "avenue"s.
And it wouldn't make the slightest bit of a difference.
@@TokenizedAI Technically incorrect. There's always a chance that an adjective like "long" can drift to other parts of the prompt, like "man" or "window" causing problems. Any time you can condense, you'll have more reliable results.
You are welcome to prove me wrong by simply replacing the words and producing the exact type of image I meant to create. I'm not beyond admitting when I'm wrong. But for what I was demonstrating in this video, it is entirely inconsequential. Even if it did produce a better angle, it simply wouldn't provide you with the exact shot.
@@TokenizedAI Just trying to save you some time, buddy. It's the "open" part of "OpenAI".
Privacy is important when creating artwork for a book you want to own. Anyone can copyright the artwork and claim it's theirs. It's your prompts that got the results. Isn't there a better platform for this?
If you think that seeing a single prompt will allow anyone to copy a multi-step process like this, then you are gravely mistaken.
If you're a professional or whatever you are working on is very valuable, then just get the Pro subscription and use Stealth mode. Problem solved.
And if not, then you're underestimating how little people care about your prompts specifically.
So many people obsess over the privacy of their prompts as if they discovered gold.
PS: As it stands, no copyright office will accept your prompts anyway, unless your creative process adds degrees of transformation to the artwork.
@TokenizedAI I was focusing on the artwork being protected and the user, due to other people can take the artwork and copyright it and claim it's theirs. Someone can use the same prompt as you and get a whole different piece of artwork. I get that! It's the privacy mode part. Which you say get a paid account. But is your already done artwork on its platform always private once your subscription is over. These are the questions people are searching for answers on.
Again, you can't copyright the artwork right now. Not only do the MJ terms of service only give you a commercial license. All relevant copyright offices on this planet also won't allow you to copyright an AI-generated image unless you've made sufficiently transformative changes to it. So, if YOU can't copyright it, nobody can. Check out what "transformative" actually means in this context.
I strongly recommend you have a closer look at MJ's license terms and also check out the latest news about the legal copyright debate. I understand your concerns but they are mostly irrelevant because what you're worried about isn't even legally possible.
👋
How is it these AI image generators can get human expressions correct more often than not, but STILL choke on the number of fingers on the hand? What's special about fingers?
Always wondered the same. Might have to do with biomechanics as hand and finger is extremely subjective to positioning of body and objects being interacted with in the environment. (This is my best guess atm)
The AI models simply aren't able to count reliably. Sounds dumb but that's essentially why. There are too many images in the training data where you don't necessarily see all 5 fingers of a hand.
Hire an illustrator
he doesn't care
Not everybody can afford to hire an illustrator. Not everybody is an artist.
лучше самому рисовать научиться)
🥱
Make a Firefox Extension - who cares about Chrome !!
You do realize that Firefox users represent less than 8% of the market, right? Plus, it doesn't even have full support for Manifest v3.
@@TokenizedAI so is that a no for the request? All I’m doing is asking
No. I'm saying it's simply not a priority right now.
It's far easier for you to use a privacy-focused Chromium-based alternative (e.g. Brave Browser; assuming that's your reason for using FF) than it is for me to port and maintain a separate extension. Mind you, I do this in my spare time and the extension is free.
If that's too much hassle for you, then that's alright. But then using the extension isn't among your priorities either.
See my point?
Burn AI generative. Poor humanity of tomorow incompetent and no skill typers...
Ok 🤔
Top Midjourney channel for nerds. It’s that simple. A few other good ones out there… @thaeyne and @inbetweenstops. The rest are pedestrian.