How to use Stable Diffusion. Automatic1111 Tutorial
ฝัง
- เผยแพร่เมื่อ 18 พ.ค. 2024
- Part 1: Install Stable Diffusion • How to Install Stable ...
In this Stable Diffusion tutorial we'll go through the basics of generative AI art and how to generate your first images. Chat with me in our community discord: / discord
Thank me on Patreon so I can keep making these videos / sebastiankamph
Prompt styles here:
/ sebs-hilis-79649068
Check out sponsor RunDiffusion here: app.rundiffusion.com/?ref=seb...
Use promo code: sebastiankamph15 for 15% off Creator's Club. Stable Diffusion in the cloud.
NMKD Siax 4x upscaler model: icedrive.net/s/43GNBihZyi
Place inside stable-diffusion-webui\models\ESRGAN\
Support me on Patreon and make me happy! / sebastiankamph
Chat with me in our community discord: / discord
My Weekly AI Art Challenges • Let's AI Paint - Weekl...
My Stable diffusion workflow to Perfect Images • Revealing my Workflow ...
ControlNet tutorial and install guide • NEW ControlNet for Sta...
Famous Scenes Remade by ControlNet AI • Famous Scenes Remade b...
LIVE Pose in Stable Diffusion • LIVE Pose in Stable Di...
Control Lights in Stable Diffusion • Control Light in AI Im...
Ultimate Stable diffusion guide • Stable diffusion tutor...
Inpainting Tutorial - Stable Diffusion • Inpainting Tutorial - ...
The Rise of AI Art: A Creative Revolution • The Rise of AI Art - A...
7 Secrets to writing with ChatGPT (Don't tell your boss!) • 7 Secrets in ChatGPT (...
Ultimate Animation guide in Stable diffusion • Stable diffusion anima...
Dreambooth tutorial for Stable diffusion • Dreambooth tutorial fo...
5 tricks you're not using in Stable diffusion • Top 5 Stable diffusion...
Avoid these 7 mistakes in Stable diffusion • Don't make these 7 mis...
How to ChatGPT. ChatGPT explained in 1 minute • How to ChatGPT? Chat G...
This is Adobe Firefly. AI For Professionals • This Is Adobe Firefly....
Adobe Firefly Tutorial • Adobe Firefly Tutorial...
ChatGPT Playlist • ChatGPT
CHAPTERS
0:00 Welcome to Stable Diffusion
2:21 Text2img
12:10 Sponsor & Dadjoke
13:46 Hiresfix
15:47 More Text2img
16:49 ControlNet Lite
18:05 Img2img
22:45 Inpaint Lite
24:27 Extras/Upscaler
26:14 Png Info
![Supernova - aespa エスパ 에스파 [Music Bank] | KBS WORLD TV 240517](http://i.ytimg.com/vi/bepqfR-pxY0/mqdefault.jpg)
![🔴 [TRỰC TIẾP] U20 VIỆT NAM - U20 THÁI LAN | Cúp VTV9 - Bình Điền 2024 | JET STUDIO](http://i.ytimg.com/vi/HxN2G6YpB-U/mqdefault.jpg)
Download Prompt styles: www.patreon.com/posts/sebs-hilis-79649068
👋
how do i download these styles? i don't see any downloadable link on the page. thanks!
@@redshift3696 ctrl f styles.csv
I don't have the buttons underneath Generate. Only have a delete button and the blue one. I reinstalled many times and still not there. Did they move the button?
It's only for people with subscription.
I installed it locally, and been messing with it on my own for about 4 days. But now it's time for me to watch these tutorials and truly learn!
i'm subbing to you because you're explaining properly and good and you're touching on a lot of small details that seem insignificant but combined together paint a much clearer picture on what's going on and i appreciate that,
Thank you very much, both for the sub and the kind words 😊
Hear hear that’s why i subbed
Hi, newbie here. Your channel and tutorials so far have been a huge help and easy to follow. I subbed to your Patreon. Keep up the good work, Thank you sooooo much! I am so excited to keep learning AI.
Just want to say your content is easy to follow and very helpful in learning this! Appreciate you 🙌
Awesome! Thank you, Jeff! 😊🌟
updated! this is fantastic a friend of mine just started to learn more about Stable Diffusion, ill share this video to them
Amazing, I hope it will be helpful! 😊🌟
This is the best video so far to go from absolutely noob to intermediate.
Just need to play around the settings he mentions and that's it.
Hugs from distance.
Subscribed, commenting, and liking
I was able to get everything up and running with your last video on how to install SD. Thanks, man. I like the way you speak.
Thank you, appreciate it! And glad you got it all sorted 😊
Thank you so, so much. I really do appreciate you sharing your knowledge in such a calm and precise manner.
Glad it was helpful! 🌟
Great tutorial, easy to follow and love the examples you included
Glad you enjoyed it!
This video explains it so well... I was looking for this kind of explanation. Thank you.
Happy to help! Recommend it to a friend 😊🌟
Thanks for your new video. I always learn something new watching your videos. This time i learned about the ESRGAN NMKD Siax model.
You're very welcome and I'm happy you can expand your knowledge about generative AI art :)
the best video tutorial on stable diffusion I've seen so far. I will check the other ones. One thing I am looking for is how to convert architecture sketches into real images and how to organize setting in stable (rundiffusion) I just started to use. Thumbs up for all the things already provided and for the knowledge shared.
Thank you! Glad you enjoyed it. You'll want to use ControlNet for your usecase.
Thank you so much for making this video. It was exactly what I needed.
Happy to help! I understand Stable diffusion can be confusing at first, and it keeps updating all the time, changing how things look. Good luck!
How come this video doesn't have more than 1k likes in four days since it was published, c'mon guys. This video is GOLD for newcomers.
Thank you, that's very kind of you 😊
Give it time
I’ve never seen such a detailed explanation of each knobs and dials on the Automatic1111. Beautifully rendered! ❤
because new users like me are viewing n reviewing the first one from time to time n not yet come to this far hhhh
Thank you so much Sebastian, great tutorial!
This is very clear and concise. I could follow along easily. Thank you!
You're very welcome!
Wow, thank you Sebastian, that is a very nice video for newbies, I already started everything and following the steps!
Glad it was helpful!
I love your tutorials, they are so helpful and professionally made. I feel like I'm learning as much from you as I would from a paid course! I had a question though, at 3:32, I don't have the 'apply selected styles to the current prompt' button in my Stable Diffusion. Also at 17:13, I don't have a depth model option.
Same here. I am thinking it must be extensions we didn't install yet.
Thank you for this, it was extremely helpful and well explained. I was at my computer with a paper and pen taking notes as if I was back in college, haha!
Wonderful! Glad you enjoyed it 😊
Yo...You the man!!!..i'''m a noob (@ 67 years of age)...just jumping in to find out that the installation process should be called " Unstable conffusion""...but now that I''ve finally installed it I look forward to trying it out. I am a graphic artist and a computer nut..I started using DOS back when DOS was cool..but seriously I just wanted to thank You for working so hard on this info and sharing it with us all...Great presentation and concise information...Thanks again!
Thank you kindly! Appreciate the comment. I'm a tad younger but I also remember DOS from back in the days. I was probably 10 or 11 when I wrote little .bat files that just redirected to another with a text output 😅
This guide is so good 😢 thank you
thank you so much! please keep on covering it. your explanations are 🚀🚀
Thank you! Will do!
I took a break from Stable Diffusion for a few months and after picking it up a few days ago and fumbling with it I decide I should review the basics. I go looking for the monst recent tutorial I can find (since in the AI world anything older then a month tends to feel put of date.) And find this just a few hours old, funny how that works out sometimes.
I got you, don't worry! 😁
Your tutorials are so good!
Thank you! Tell a friend 🌟
Already did ^^@@sebastiankamph
Excellent pacing and full of value. Thank you
Glad you liked it! 😊
Thank you, this was a very informative and well made tutorial. Appreciate it! ♥
Happy to help!
You're a great teacher this is the perfect series for SD beginners
Thank you kindly :)
Youre the best teacher bro, thanks a lot homie!
Happy to help!
Boom, fantastic as always!
Thank you for the best tutorials!
Glad you like them! Tell a friend 😊🌟
Thank you so much for this tutorial!
You are so welcome! 😊
From what I learned, the higher the resolution, the more sampling steps you need
picture me, chilling in my chair with my gown on can of cidre with my pet rat max nibbling on some food next to me. Just switched from the calmest and most chill tutorial to the second vid. Clip volume goes up by 300% scaring the shit out of me. not that im complaining, that was super funny to me
You are an excellent teacher!
Learned a lot thx
Thank you very much! Very interesting and useful to watch your videos
You're very welcome! 😊🌟
wow, great info here. Thank you!
Glad it was helpful! Welcome aboard 😊🌟
muchas gracias, funciona perfecto
I usually come for the jokes and u never fail them 😂
Glad to hear you're enjoying them! If you guys keep coming back for them, I'll keep providing 😅
@@sebastiankamph i'm sold
Subscribed. Thank you for your videos!
Thanks for subbing! Welcome aboard the AI hypetrain 😊🌟
The scar and veins by your right eye make you look like a movie villain, which is pretty cool.
do you have a tutorial video on how to download and use stable diffusion on a Mac?
Great !!! C'est génial, merci beaucoup
Thanks so much, this was super useful!
You're very welcome!
sooo usefull tutorial series :)
Glad you think so!
Thank you Sebastian, I have watched a couple of these videos on how to get started in A1111 and this has been the best. I have only recently began my journey into generative ai, I began using Fooocus which was nice and simple to begin with. I have dabbled an little in Comfy and now looking to play with this a little. Which out of Comfy and A1111 would you recommend as a beginner? Looking forward to binge watching your other content 🙂
Glad to hear you enjoying the content! I recommend to start with a1111 and then also learn comfy a little on the side. They complement each other very well.
Thank you for these great videos!
Thank you for commenting!
awesome tutorial!
Thank you!
I did my install and was able to make some images with some models, but it keeps failing to generate and I get this response: "NansException: A tensor with all NaNs was produced in Unet. This could be either because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the "Upcast cross attention layer to float32" option in Settings > Stable Diffusion or using the --no-half commandline argument to fix this. Use --disable-nan-check commandline argument to disable this check."
I can restart and sometimes things work fine, sometimes they don't. I can change models and it sometimes fixes the problem, sometimes doesn't. Image to Image always gets this response. I've found this problem mentioned on some forums but couldn't find an applicable fix. Any ideas?
Thank you for all the help, but for some reason I cannot use the Styles options. Whenever I click the arrow or anywhere around it nothing pops up and the generated images don't look good. Do you have any clue as to why this is happening?
Thank you very much for the tutorial!
I wonder what should i fix if some downloaded models simply give a grey box result instead of an actual image?
Great video, and cute doggos!
Thank you my friend! 😊
any tips on fixing eyes? and overall face xD the latest version for example don't have "restore face" and look slightly different that version in the video
😍thank you for this video
You're very welcome, I hope you'll enjoy it! 😊🌟
Thanks for the video. When is it possible to run on Mac's ?
Thank you.
Thanks so much for this - coming late to the stable diffusion game but your series of videos are a great way to learn (so am supporting you on patreon!). By the way, where would I be able to find the deliberate v2 model you've shown there? Do you still recommend it over SDXL etc? Cheers
Thank you very much! I think a lot has happened since deliberate2, but it's still good if you want an overall digital painted look. See epic realism for 1.5 photorealism or JuggernautXL in SDXL. SDXL is good, but 1.5 still is a little better for photorealism. You can get them all on civitai
thankyou so much for explaining
Glad it was helpful!
@@sebastiankamph can you make a video on sad talker eye blink and head movement
Thanks!
Thank you for the support, greatly appreciated! 🌟❤️
Link for the NMKD Siax 4x upscaler is not working anymore, where can I download it?
Is there any setting or controls to increase the size of the controlnet inpainting window for more precise masking? only thing I've found is zoom the display size of text and images in my browser
thank you for the video Brother
You bet!
Thanks so much
You're welcome!
Great! I've been using Prompt Chan AI, which is great but not as good as what Stable Diffusion looks to be. BTW the Stable Diffusion free web version is junk compared to others like MidJourney or Soulgen. The best web AI I found is Prompt Chan AI, but I wanna start using Stable Diffusion.
Thank You. I just subbed annd will follow your channel.
Welcome aboard!
You are so helpful I have a question on my Stable Diffusion checkpoint, some models I downloaded and installed I can't use. I keep getting errors. Also, I can't change them without shutting it down and reloading it. If I do get it the change, it gives me an error.
Here is one of the errors:
NansException: A tensor with all NaNs was produced in Unet. This could be either because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the "Upcast cross attention layer to float32" option in Settings > Stable Diffusion or using the --no-half commandline argument to fix this. Use --disable-nan-check commandline argument to disable this check.
Time taken: 0.5 sec.
Hello Sebastian, after the initial guide I'm loving it, very descriptive and no fill out moments. One thing, you say the Styles are free, but I have to pay pantheon to download them, or am I missing something? With other styles i got from the web I don't see the icon "Apply selected styles to current prompt", is that because of the file?
Thanks and great videos, gonna watch them all.
Hi Alejandro and thank you for the kind words! I recently started experimenting with putting the styles for Patrons only as a means to be able to focus on making these videos. Initially that wasn't the case, but without it I might not be able to keep up. I will see how it develops and might change back in time. The button you ask about has been removed in the latest a1111 release. I did hear they considered taking it back though. Welcome aboard! 😊🌟
thank you!@@sebastiankamph
SD AI noob here... I suppose we can choose the new SD XL model and that will work fine with Automatic1111 as you've shown here to install it?
Great tutorial as always thanks so much bro 🙏🏾🙏🏾
I'm thinking of buying a new pc to for the AI staff (Image, Audio and Video) can you please suggest a Good PC that works smoothly for training and other AI related things.
If it's okay it would be great if you make a Video
Hey and thank you! Regarding Stable diffusion generation it really just ends up with the GPU. The rest are nice to haves for a computer of course, but SD mainly works with your GPU. I'd get anything with 12+ Gb vram if possible, depending on your budget. The more the better.
@@sebastiankamph Thanks for the suggestion🙏🏾🙏🏾 you're the best
Hey Sebastian could you tell me on an approx how big is your stable diffusion on your pc.
Can this be used to make a real portrait face to a MidJourney face?
i thought dall-e was cool until I saw the super insane stable diffusion images...
Thanks for you video let me know how to create an ai art
Hey do you know if rundiffusion is uncensored?
In the video you said the styles are free to download is this not the case any longer?
Lovely!
good tutorial and nice voice too
Thank you, very kind!
@@sebastiankamph you’re welcome and you deserve it!
This one is ultra nice and helped my beginning steps alot!
Somehow i made the experience, that i need to use Anime stuffs to get the charakters in the postures i want them to be in and then making a depth map to make the "Realistic" one to behave like that.... dunno but i have problems with realistic humans, to for example someone jump up like martial art kicking forward, with akimba barettas in theyre hands - but with anime that works well!! - and then, like i said, with depht its easy to reproduce it to a real image....
why is it like that ? arent there realistic checkpoints which you can prompt better in ways like that ?
i feel like its some kind of random with realistic one... also with negative prompts.
Thank you!
You're very welcome! 🌟
Thanks for the great beginners video. I can’t seem to get x formers to work though. I get an error no module ‘xformers’ proceeding without it.
Any chance you know about this?
I don't think xformers is needed anymore with new versions of automatic1111 with torch and all that. In my webui-user bat file I have this for my COMMANDLINE_ARGS=--opt-sdp-no-mem-attention so you can try removing xformers from that argument if it's in there for you. But also maybe double check Sebastian's how to install stable diffusion video from a month ago.
So, a question. I have created images to a certain model on Tensorart site. Now how to copy my work and continue on my pc? How to get that same model - face, body, hair etc?
Really usefull tutorial. I'm beginner and understand fine. Thank you for making amazing content for us.
Happy to help! One week away, you been doing great art since then?
@@sebastiankamph I had some problems with lunching SD, because I have rx 6600xt, and you know that amd gpu is so bad on making art with SD. Yes, I fixed some problems. But in future I need Nvidia gpu with more vram. About my art, I know not to much about making good design, but I leaning something new everyday ☺️. Now I'm on simple art style. Not photorealistic. Trying myself on POD.
Hi, love your tutorial, Can you please Share the link Preprocessor Model? its showing none on my end.while in folder of extension, it shows few Please share some link? I search in videos but cant find for a1111
Do you have a tutorial about how to use stable inpaint upload?
Is there a reason why many AI images seem to have the same rim lighting?
Is stable diffusion and automatic1111 with the use of models free to use in your company or how does it work with copyright and such?
Thank you Sebastian! But my generation always fail with this note: NansException: A tensor with all NaNs was produced in Unet. This could be either because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the "Upcast cross attention layer to float32" option in Settings > Stable Diffusion or using the --no-half commandline argument to fix this. Use --disable-nan-check commandline argument to disable this check.
Time taken: 24.1 sec. ----would you please explain why and how should I solve it? Thank you!
Do you have links for your comparison tables (Sampling method and steps, and cfg scale)? It can be very useful for reference!
btw, great video.. thanks!!
Thank you! Sorry, no. Just generated them for the video. You'll have to go back to the video to see them :)
Hi Sebastian, I have a problem generating image. since I'm new to ai, thought I'd try it. I've installed all there is to get it running' followed all steps in your tutorial. I got 32 gb pc memory, 8 gb graphics memory. made alterations as error messages suggest. but not much luck. Could you please make a tutorial about this issue and how to fix problems like this. Forums don't help, just people fiddling with the same issues.
So is Automatic1111 different from Stable Diffusion, or is it another name for the same thing? I had a lot of trouble getting Stable Diffusion to install but finally got it. Now I'm finding this and wondering if I installed a lesser version or something.
Stable diffusion is the base AI model. Automatic1111 is one of the interfaces. a1111/Vlad/SD.Next, ComfyUI and Invoke are probably the most popular and best ones. The main AI generators you have are Stable Diffusion, Midjourney and Dall-E.
@@sebastiankamph thank you
First, thanks for the videos. Your time spent is appreciated.
My questions are hardware requirements.
My ONLY understanding is 4-6 Gig VRam min, and higher video processor is obviously better,
but what generates 512x512x 4 images in 1 min?
What generates 4k x 4 in 1 min?
How, if it does, CPU and RAM come into the equation?
I've not seen any SD Local Install vids addressing hardware, maybe I'm looking in the wrong places.
Suggestions?
edit: "standard" image size is 512x512, not 540x540🤦♀
Thank you! It's all about the vram and speed of your gpu. You can almost run it on any old potato as long as you have a good gpu. My RTX 3080 10gb generates a 512x512 in a couple of seconds. I tried 8k with tiles, that took me about 30 minutes.
@@sebastiankamph Thanks for the detailed reply!🙂
My current build is incredible old and I'm going to rebuild from ground up, save the SSD's, case, and PSU.
Thinking 3070 with mid range MB, RAM, CPU should do for now.
Thanks again.
GTX 1070 can generate 1 picture in around 20 secs (512 x 512 ; 25 sampling steps, Euler A, CFG scale 7)
Without "styles" (negative & positive) applied its around 1-2 secs faster
Anyone with AMD card here, please tell us how fast are those GPUs!
Thank you so much for this. What I still don't understand is the difference between Dreamstudio and running SD locally, like your doing. Do they produce the same results?
Locally uses your own computer, your own gpu. Dreamstudio uses a cloud computer somewhere.
And running it locally is completely free, no matter how many images you generate?
@@peterpain9715 Yes. Except your electricity bill from using that computer and gpu.
@@sebastiankamph respectively automatic1111. But I think this is also free except for electricity? Just can’t believe they offer this for free!
the art of humorously delivery of AI generated jokes is just..
Hey Sebastian! Last week I followed a tutorial to train the Stable Diffusion 1.5 base model with my face using Google Colab (as I have a poor 1050ti 4Gb vRAM). Now I would like to know how to use my trained base model with the other cool models like Realistic Vision so I can generate photorealistic images of myself wearing a suit. Could you give me some guidance? I like your simple way of explaining things. Grateful!
Hey can you please make a video on how to use stable diffusion using google collab, please?
please a tutorial for creating our own styles 3:23
@sebastiankamph Tell please, if this tool can change the style of one already existing image to the style of the second one by example? for example, taking a picture in the style of a painting.
this implies that you need to load two images and remake one like the second.
Yes, with ControlNet extension (I have multiple tutorials).
@@sebastiankamph can u give please a link? there are a lot of videos, I don't see exactly which one.
@@AlexandraDeBerg There are multiple ways, I'd recommend trying reference only: th-cam.com/video/tBwmbTwMxfQ/w-d-xo.html
@@sebastiankamph Forgive me, my dear sir, but I checked it closely and it is not about I asked for. And I checked about 10 more videos and couldn't find it anywhere. I mean when you have two already existing images and you need to change the style of number 1 to the style of number 2, leaving the essence of number 1. you don't upload 2 images anywhere, you only work with one. Is such operation possible in stable diffusion?
Hello, Cool video. Which are your PC Specs? Thanks
Sharing prompts and modules is awesome, getting the avg user up to speed on hardware requirements is necessary.
I use a RTX 3080. Rest of the specs are irrelevant.
Are the styles still free? cant seem to find the download on the patreon