Mastering AI prompts with Stable Diffusion
![Vladimir Chopine [GeekatPlay]](http://yt3.ggpht.com/C8QIoKOqO3xNbQhrLw7Q48NVh8ynPa99GNZ2qYvEN4Csg_nwKRnLcMFYfEdljEseP84ZCfUixQ=s48-c-k-c0x00ffffff-no-rj)
Vložit
- čas přidán 22. 02. 2023
- How to create fine-tune prompts in Stable Diffusion with advanced functionality. In this video, I am explaining how they work and where to use them.
Very impressive AI driving image and video upscale topazlabs.com/ref/1514/ , try for free.
THANK YOU for your support!
Please subscribe and leave your comments, and don't forget to click on the notification bell: “Look daddy, the teacher says, every time a bell rings an angel gets his wings,” ZuZu Bailey from “It's A Wonderful Life.”
What do I use:
Image upscaling topazlabs.com/ref/1514/
Micro SD cards 256Gb V3, I use these in Drones, Insta 360, and GoPro's amzn.to/3HsTU7j
Must have: Micro SD storage/reader/Cables: amzn.to/3GZ33D8
Simplified Photoshop, no annual fees, just one-time purchase amzn.to/3DaUpAi
Canon R5 is not cheap, but it is definitely the best amzn.to/3DclTFX
Must have lens adapter EF to RF amzn.to/3Dk1qhp
Insta 360 X3, I am really impressed by this camera: amzn.to/3HqdMb4
My to-go camera GoPro 11 : amzn.to/3XMcKMc
DJI Osmo 6 amzn.to/3zsKlAG
Portable lights, very versatile usage : amzn.to/3wsUqMa
Another DJI great product. Drone Mini 3 Pro, my togo: amzn.to/3XKN9mG
One of my favorite modifiers from Fotodiox - amzn.to/2Rfr1Px
Another modifier, that helps with fill light - amzn.to/2ReC2jX
Corsair keyboard: amzn.to/3DlclHN
Adobe Photoshop CC - amzn.to/2TNrLwL
My #1 dress in Baroque/Rococo photoshoots. very good quality: amzn.to/401rGHQ
Baroque outfit for males: amzn.to/3kBdEg2
I recommend getting this shirt for the outfit above: amzn.to/3Wzcv5Z
My Vue book - amzn.to/2TGUkvQ
3D Art essentials - amzn.to/2RfqPjh
My Patreon webpage - / geekatplay
Tutorials and packs - gumroad.com/geekatplay
Tutorials website - www.geekatplay.com
Photography - www.chopinephotography.com
Subscribe to my channel for fast notifications on new tutorials - / @geekatplay
one tip is if you highlight a prompt then press ctrl and arrow up/down key you can rapidly change weights also your videos are very useful thank you
Thanks, will be helpful.
I'm relatively new to SD on a1111,been doing it for about a month now, this is so helpful!!! thank you!
As someone who just got into playing with SD (yes I know I'm late), this has been immensely helpful information. I have been getting extremely frustrated with trying to get desired results (clearly I have no idea what I'm doing lol). You just gained another subscriber. :)
sorry that i have to ask maybe i just missed the part but i don't understand whats the main difference between using [ ] ( ) and ' ' or ( ' ' ) [' ']
Thank you, that explanation with square brackets and weighting gives me a lot more freedom in structuring space and composition!
thank you
Finally simple and straight to the point explanation. Thanks.
Most in depth, straight forward guide. Was looking for exactly this. Thank you so much for your time and dedication.
thank you
Love your content and your accent. Keep up the great work helping us learn new tricks!
Thank you Vladimir, I still have to get Stable Diffusion running, so far only using Midjourney for my content.
Great insight. Thank you.
Super useful overview, thx!
Excellent! Thanks.
Great tutorial. Creating prompts is hard work, and this one is really helpful. I really hope you will make more videos like this, since nearly none exists.
Thanks a lot!
Thank you for your support!
wow. this was insanely helpful! i just started with this whole AI thing and this will help a lot, especially with the finger issue! thanks dude
thank you
very useful thank you
This is an excellent video, THANK YOU!! As for the fingers, it won't render perfectly every time, but it helps significantly.
Excellent video! Thank you very much!
Glad you liked it!
great tutorial I didn't even know it was possible to do that
Excellent help thank you!! Works fine in PlaygroundAI too.
Glad it helped
One of the best and intresting video i have seen to better understand prompt.
thank you!
juicy information
Big thumbs up and a grateful new subscriber after watching your excellent video. I have been getting adequate but inconsistent results and now I see it’s because I am misusing the prompt system! There are so many sites with poor or outright wrong ‘information’. Thank you for setting me straight. Now, if you’ll excuse me, I’ll check out your other videos. Thank you, Vladimir!
thank you for your support!
One thing I've noticed -- and you'll see it, too, if you haven't already -- is that if you specify a color _for anything_ in the image, the AI will apply it to other things in the image you didn't tell it to.
Look at the ball. He didn't tell the AI to make it a _red ball._ But it took the instruction 'red' in 'red coat' and applied 'red' to the ball, as well.
yea, you need use negative prompt, example: prompt: 'red coat', ball. Negative: 'red color ball'
@@Geekatplay you can also use the cutoff extension to denote order of colors to each object
@@Rasukix What is the name of the extension?
Finally! Someone did a good explanation!
thank you
greate! :D
I make the images with web interfaces like Playground AI, Stable Diffusion site and Leonardo AI. Is weighting there working the same way as shown in your fabulous video?
Thanks.
what a trip bro
yes it is
do you have a tutorial that separates each work use in the prompt in categories.
like style: surrealism. mood: gloomy. etc?
How do you get aspect ration buttons in UI?
Good video
thank you!
I have a "ControlNet" folder with "model" folder inside with pixar, monaliza, etc in pt. files how use it ?please
Thank you, great info. I’m on an iPad and need to know how to prompt the image ratio so its portrait or landscape, not 1:1. Can you help? I’m using stability ai. ❤️🇦🇺
The aspect ratio refers to the width in relation to the height (w:h). Landscape is 16:9 and Portrait is 9:16. I hope that this helps!
Learned some stuff thk ! One thing i didnt figure is how target prompt , like a scene with 2 model , and i want a spécifique one to wear a hat and glasse .
Most of time it just appear randomly on one or another
thank you
10:23 that was you making h3nt a l I caught you🤣🤣(seems like I have a lot to learn about stable defussion to make complex images after seeing that verry long prompt you had on your clipboard)
yea :)
More clearly its impossible. Great video.
thank you for your support!
This is what I need but the ui is really bright too me I cannot see well the words in the prompt my eyes hurt, and it's too zoom out if only you more focused on (zoom in) the prompt tabel and make it dark mode it's going to be great
thank you for suggestion, i will switch to dark mode
where can I get the anything 4.5 model? I can't seem to find it atm
huggingface.co/andite/anything-v4.0/blob/main/anything-v4.5.ckpt
brother had negative prompts for nsfw generation pasted into the positive prompts at 10:23 xD
intresting
why is when i put "red" to an item, it gives red to everything else as well
be sure to isolated in ()
Ive use this prompt...
pencil drawing of a man
I get a pencil drawing of a man and actual pencils in the image. When all I want is a drawing of a man done in pencil ...
how do I get rid of the images of pencils without getting rid of the "PENCIL drawing"?
I feel like everyone is just bloating their prompts with generic nonsense that harm output. This is particularly noticeable with negative prompts. You need to be selective because you can cripple the generation with even seemingly unrelated phrases.
As for having 50+ strings of adjectives and dozens of different artists and styles, you're kinda kidding yourself that this is achieving anything
(((Thanks:2)))
Why do all the pictures look like cartoons and not like photos?
I now need to learn how to type square brackets lmao
hehe
damn thick accent makes it hard to understand
damn thick people make it hard to enjoy youtube
Use the captions then. Smh