A1111 - Best Advice for amazing Stable Diffusion Images
HTML-код
- Опубликовано: 10 сен 2023
- Here are some of my best Tricks for A1111. How to get amazing images with ease. Highly detailed. Anatomically correct. High Resolution. Working with older GPUs. Using the best Models and Loras.
#### Links from the Video ####
Addetailer Video : • AI News: Draw 360 Scen...
XL More ART Lora Video: • This Lora makes REAL A...
Controlnet Videos:
• DW Pose for A1111 - Be...
• NEW ControlNET SDXL Lo...
• SDXL ControlNET - Easy...
Roop Video: • ROOP Deep Fake for A11...
Downloads:
Upscale Models: huggingface.co/uwg/upscaler/t...
Realistic Vision: civitai.com/models/4201/reali...
Rev Animated: civitai.com/models/7371/rev-a...
Deliberate: civitai.com/models/4823/delib...
EpicRealism: civitai.com/models/25694/epic...
XL More Art Lora: civitai.com/models/124347?mod...
Add Detail Lora: civitai.com/models/58390/deta...
#### Join and Support me ####
Buy me a Coffee: www.buymeacoffee.com/oliviotu...
Join my Facebook Group: / theairevolution
Joint my Discord Group: / discord
AI Newsletter: oliviotutorials.podia.com/new...
Support me on Patreon: / sarikas 
Хобби
![Gunna - whatsapp (wassam) [Official Video]](http://i.ytimg.com/vi/bY1kLt4iz94/mqdefault.jpg)
Probably the best 1st-day introduction to SD that anyone has made so far. I knew most of this stuff, but it's taken me several months (and many of your other videos) to gradually get all this knowledge.
Yes!
Absolutely agree!
I can confirm, this is stellar as I am just starting out
I just started and this video is essentially hieroglyphics. Idk what A1111 is or what this man is talking about. Does anyone know a video that shows how to use stable diffusion as a beginner? This is for well versed people, as I don't know anything about anything he is speaking of
Thank you for this fantastic overview of A1111! I've been dipping my toes in it every now and then, but kept finding myself getting overwhelmed with the UI. This was really helpful.
Here's a suggestion for a future video, or something you could maybe keep doing from time to time whenever there's been enough updates to justify a new upload : A start to finish video where you start with a rough concept and take us through your workflow to achieve the final result. You don't need to go though the details of each step, you can just point us to your other videos like you've done here. This would be very helpful in terms of having a bird's eye view of what the main steps in the process are.
Thank you for all the work you've done in making and sharing these tutorials. 🙏🏽
Great advice. I've been using A1111 for about 10 months now and I still learned a few tricks.
Follow the instructions in the video up above lol @OmarDahmen-fh9eq
Thanks for all your amazing tutorials and different showcases. I have a lot of fun making pictures of my favortie character or landscapes. Keep up the awesome work :)
Thank you so much for the straight forward tutorials, helps so much!
00:04 Choose the right model to get amazing results with AI
02:24 There is a lot of information about different models and settings that can be used.
04:38 There are two important types of Laura models: XL and detailed tweaker.
07:11 Adjust settings for high quality rendering
09:38 There are different important folders for specific types of models.
11:55 Adjusting weights and settings can improve the image generation results.
14:16 The Image to Image feature is powerful for making subtle changes in an image.
16:35 Use lower denoise strength for subtle image changes and higher denoise strength for intense changes
18:49 Control net extension allows for tracking and adjusting body parts in images
20:56 Render an AI image that closely matches the body and head shape.
23:04 Using batch size can improve performance on slower computers and older GPUs
Thank you!
Great video!!! Can't wait for another amazing lesson!
You have provided almost every update that SD have gone through since its beginning in one single video, keep up le good stuff King 👌
thanks this is the most complete guide i've seen, i finally understand what are loras and other settings
Amazing as always. I learned a bunch and have been doing this for 3 months!
I've spent the last month trying to learn via a bunch of tutorials, and with aaaaall of that combined, this video STILL provided me with more information than I have spent hours of researching for. You are the MAN
Thank you! I love your comment!
Thank you for excellent tutorial!
Thank you very much! LoRA in ADetailer and roop extension are excellent hints!
agreed, the Lora´s in ADetailer are very powerfull. And yes, roop amazingly good. Just fon´t use the offical one, the uncensored is better
You’re amazing dude! Thanks so much
wow, this is packed with so much great information!
Wow. I really enjoyed watching this one despite the fact that I have watched your in-depth videos where you have already covered stuff in much detail. Thank you.
VERY good tutorial, Olivio :) Very detailed and understandable. Thank You.
Heyo, glad RUclips recommended your videos to me about a month or back and has helped me greatly over a few days. Can say that I'm using a NVidia GT1030 2GB GPU and the img2img tiling upscale helps relieve the pressure dramatically. I haven't tested out the max upscale resolution yet but I can get a constant 1024x1536 upscale down, a long time no doubt but possible indeed.
Great great introduction!!!🥳
Thank you for your hard work!
Is there a possibility to publish a tutorial how we can train a Lora for a character (myself as example) for SDXL in near future?
Great Video. Thanks so much for all the info you provided!
Great video. thanks for the tips
Thanks Olivio. Can you share how to get perfect hand or finger (what setting to use?). Usually the hand or finger is not perfect on stable diffusion.
This is Great advice.
So nice ! so interesting! fantastic job!
Another amazing video, this will be really good for beginners. Regarding the Roop extension, i think it is no longer available? I did a fresh install a few days ago and i had to use an alternative.
Great advices! Thanks for detailed video. a few months ago I can use "Infinite Zoom" but now I can not find tag on UI. is it the problem of my local computer issue or they stopped infinite zoom no more?
Sir. You have saved me again!! 😂 I got so depressed after the SD UI updated and changed. I spent so much time learning how to navigate the old controls. Thank you for this video. I feel like I’m back in the game now. Lol.
great video as always. Thank you .
Keep on meaning to leave more comments...thanks so much for all your awesome content!
영상 감사합니다~!
THANKS THANKS MAN
First ! love your work man , keep it up =) =)
Wondering why you never put any hi res steps? also can you review the dynamic shareholding extension?
Nice video, useful advices 👍
Hi Olivio can you pls explain how to move automatic 1111 to a new partition, I think many users may be slowly facing the problem of lack of partition space due to the size of the models
Thanks you, highly informative and absolutely fascinating. I could spend the rest of my life on these tools!
Awesome, thank you
does the upscale script also help make face and details better as Hires fix does?
I'm clapping.
If I knew that the video was going to be this good, I'd have put on a suit.
Any remote chance you could cover all this for Invoke pretty please specially for up scaling
hi and thanks for all this useful information! i recently updated to the newest A1111 and installed SD XL. since the update loading the program takes a long time, this is what it says in the cmd each time:
Installing sd-webui-controlnet requirement: changing opencv-python version from 4.7.0.72 to 4.8.0
but rendering is the real problem. it can take up to 5 minutes to render 1 photo in 1024. do you have an idea what the problem could be? i have a GForce RTX 3060 card
wieder mal ein sehr informatives video 👍hier noch 2 Extensions mit denen ich sehr gerne arbeite: 1. Posex zusammen mit Depth Library (perfekte Hände IMMER!) und 2. Inpaint Anything (erstellt ein Bild aus unserem ersteltem Bild, das von lauter Masken besteht, man markiert welche man ändern will und ab die Post! hat auch verschiedene Models die verschieden stark wirken). Würde mich freuen wenn Du sie testen könntest und Deine Meinung dazu äußerst.
P.S. würde gern alles auf englisch schreiben (für alle), aber mein englisch "is not the yellow from the egg" 🥶
Your explaination is very good, I would like to know what AI you used for the final talking cartoon
At first I thought wow his beard has grown really quickly 😅 then I saw it was your Mic 😅
I have the exact same thought. 😁
How do you check whether you're using SDXL 1.0 or SD 1.5 and how can you change it?
How does the seed behave in batch size vs batch count? Is it always gonna make a new seed in both cases for any image?
Just to mention there is an error at 1:03 in the positive prompt. I'm a photographer and the name of the Fujifilm X-trans camera is misspelled. It's Fujifilm X-T3, not Fujifilm XT3. And yes, it's important because it's the correct name or reference, but also because the name X-Trans is a combination of X, which stands for the X mount system that Fujifilm uses for its lenses, and T means Trans, which is short for transverse, meaning crossing or intersecting. And 3 because it's the third model of this series. So, Fujifilm X-T3 ... otherwise, the prompt wouldn't work at all.
Hi Olivio, thanks for your videos! I was wondering though...you were saying that Controlnet extensions need to be for SDXL or for SD 1.5 to work with the appropriate model. How do you know what SD version a Controlnet extension is for? For example, is OpenPose for SDXL or for SD 1.5?
If you load/change model for controlnet, you can easily distinguish them by file name. You have either "1.5" or "sdxl" in filename somewhere.
Why doesn't Civitai allow us to create "Detail Tweaker LoRA" on the site like the some of the others that has the create button instead of the download button?
Is it because it is SD1.5???
I ask because I don't understand why "Juggernaut XL" which is SDXL yet doesn't allow you to create on the site??
I'm still a newbie at this and would like to know.
Great video as always Olivio!
Bester Mann!
Can you search the list of all your installed extension ?
Saw an interesting post on reddit that says seed numbers are very important (almost as important as prompts) with your resulting images
Can you do a tutorial on Dream Studio Animation node in ComfyUI. please
can you do a video with anime related stuff? the most videos on youtube are all about photorealistic....
Also why using Automatic-1111 when ComfyUi gives you more options??
For the roop extension, do you run the entire prompt again, while enabling the roop extension?
Or do you run that in img2img with the face masked out?
Doing these minor changes to pictures are still not clear to me unfortunately…
sir , how to use pose in new SD ?, ive try tutorial but it doesnt work , can you teach us please ?
Hi, ver`y huge and interesting videos, im new in stable diffusione and a friend of mine told me that the picke files (like all upscaling files that u mentioned) are not secure, could contain some virus, or something problematic for our pcs... is that true? there is some risk downloading and using those files?
I'm sorry about my english, i just hope that u can understand what i mean 😅
Why use sd upscaler instead of the ultimate sd upscaler?
any update to fixing hands once and for all?
Olivio, love your channel. I'm frustrated with A1111, faces that should be blurred in the background are being annoyingly sharpened. How can I turn this off? I've tried every setting in adetailer, even turning it off, no use. Thanks.
What I would LOVE to see is tips on how to avoid getting the same overtrained faces everytime you run a prompt. A lot of models are so overtrained that every woman, man or child looks alike. It just doesn't show any(or hardly any) variations.
Epic realism is a perfect example. Amazing results but every face looks the same even if you try to influence it.
I was trying to install the ADetailer, but I get the error: importerror cannot import name 'yolo' from 'ultralytics'
I was trying to solve this problem but no idea of how to do it :(
Please give me a advice. i am building my own website provide royalty free image (personals or commercial purpose). So this "Civitai models" image generations can use as a royalty free images?
its cazy how quick advancemnts go. i havent done anything in a half year or so and now theres so many innovations.
do i have to add the preview images for the model cards myself?
Only if they're missing and even then you don't have to add them. You can leave them blank and they still work just as well.
How did you load upscaler in Hi Res fixes ? Im not sure where to place downloaed files
Nvm had to continue watching
Hey, how do I update A1111 to the newest version and retain all of my current stuff? Is there an easy way?
You can click right next to the folder path for example C:\stable-diffusion-webui. It will now select the path. You remove everything and type "cmd". Cmd pops up and now you type "git pull". It will update everything
there are so many pt files in the link. Are they supposed to go to the same folder ?
where did "face restore" go in the new update?
Not sure but ADetailer is way, way better (it's in extensions) than Face Restore.
@@Elwaves2925 Thanks. My problem is, that the ADetailer is not saved in the image data. So I might creat a good picture, but I will not know how I did it after a while. ;-)
inpainting never works for me.. Do you leave the prompt the same?
You must use a model that has the word "-inpainting" in it.
You only need to keep the parts of the prompt that describe the spot you are inpainting the masked area only
my A111 exceeded nearly 480GB
Wawww clip skip on the main options bar!?!? ❤❤❤❤
👍👍👍
can someone pls tell me how to check my SD version.!!
You should make a video showing which cases ComfyUI is better than A1111. Ok, maybe you won't do it.
I don't see highres fix in my a1111
How about face swapping?
I stopped and was disappointed you did not regenerate the pose change :(
Is Comfy still better?
Define 'better'. It's never been better for me. Comfy is less resource intensive, quicker for some people (I disagree but it depends on what you do) and it's main feature is that it's highly customisable. The downsides of that are it can get messy, you're constantly zooming in/out and panning around (unless you have a large monitor). With A1111 it's mostly there in front of you, there's less scrolling, no zooming, no mess but it is resource intensive which is a big factor for many. A1111 also as a number of QOL features that Comfy doesn't (last time I checked). Comfy definitely has it's advantages, especially if you have lower VRAM but personally I still prefer A1111 and the results are just as good. The best way to know is to try both. 🙂
One thing often missing on civitai is the clip size and the vae sadly
I would really appreciate if anyone here can share a fitting vae for SD 1.5
What is A1111? I don't know what you're doing, I guess I need a more beginner friendly video that treats me like I don't know how to use stable diffusion, this is clearly for people deeply versed in it cause you have me lost from the first 10 seconds
Even in neural networking classes, French women have been displaced by African women😏
thx