Create Amazing Art in Stable Diffusion the Easy Way
Vložit
- čas přidán 2. 06. 2024
- The site where I got all the Embeddings, LoRA, and Model downloads - civitai.com/
Ever wonder why your Stable Diffusion Art isn't as great as you expected? Well check out this video and I'll show you a way to fix it as easy as possible and also get a particular style not offered by the vanilla flavor of Stable D.
This video will show you how to install Textual Inversion Embeddings, LoRA models, Custom CheckPoint Models, negative embeddings, VAE's and more. All without spending hours training your own model. You'll also see the same prompt with and without these custom files.
VAE code - case sensitive and commands are separated by commas:
Code to Add VAE to UI sd_vae
Code to add VAE and Clip Skip to UI: ,sd_vae, CLIP_stop_at_last_layers
VAE Model Option 1 - huggingface.co/stabilityai/sd... (WARNING: This one no longer works for me!)
VAE Model Option 2 - huggingface.co/stabilityai/sd...
Negative Embeddings collection - huggingface.co/nolanaatama/em...
The site where I got all the Embeddings, LoRA, and Model downloads - civitai.com/
Another site to get embeddings. Not as user friendly as Civitai but it has embeddings - huggingface.co/spaces/sd-conc...
The model I used, Deliberate (best all around model) - civitai.com/models/4823/delib...
The Negative Prompt for the Model Deliberate - civitai.com/models/30224/nega...
The Textual Inversion / Embedding Emma Watson Model I used - civitai.com/models/4944/emma-...
The Lora File of Gigachad I used - civitai.com/models/18177/giga...
You can Find the Lora File of Jinx from LoL on CivitAI but it’s too spicy to link to youtube NSFW warning if you search for it.
Here is the Stable diffusion Bible (not my site) - stable-diffusion-art.com/embe...
Other videos you might like:
Have you heard of Control net? Get the exact Pose or shape you want - • Master your AI Art Pos...
Fix Hands in Stable Diffusion with Control net - • Let's Fix Those Hands!...
Need to install Stable Diffusion Automatic 1111? • Updated Stable Diffusi...
Want to Animate an image using stable diffusion the easy way? • SadTalker - A Free & E...
What was updated in MidJourney V5? • Mid Journey V5: Your U...
Have you seen MidJourney's new features? Permuations, repeat commad, NiJi V5 • 3 New Mid Journey Feat...
Want to Animate an image the easiest way possible (not free or cheap) - • Let's Animate a 2D Ima...
Here are 3 Midjourney Alternatives - • Three Amazing MidJourn...
Chapters
----------------
00:00 Intro and what you will learn!
01:15 Embedding & LoRA - What are they and where did they come from?
02:35 Embeddings / Textual Inversion grabbing your embeddings
04:45 Embeddings / Textual Inversion Second Example
05:11 Install Embeddings into Stable Diffusion/embeddings Folder
05:24 LoRA download and paste into model Stable Diffusion/models/LoRA Folder
06:29 Custom Models - CIVITai Deliberate Model V2
07:45 Put Custom Model in Stable Diffusion/Models/Stable-Diffusion Folder
07:55 Negative Embedding
08:24 Negative Prompts
08:59 Custom Model vs. V2.1
09:20 Changing your Models
09:34 How to Access Embeddings & LoRA in Stable Diffusion (Extra networks)
10:13 Testing the Emma Watson Embedding
10:23 Quick Info about adding weights or bias'
11:03 Embedding vs. Non Embedding
11:19 Lowering Weights
11:35 Atomic Punk Embedding
11:52 Fixing a Ugly Picture w/ a Neg Embedding
12:30 LoRA Model Example JinX from League of Legends
13:00 Quality of LIfe - 2 ways of Adding Preview Photos to LoRA, Embeddings, Models, and Hypernetworks
14:41 Dropping gigachad in SadTalker just for fun
15:02 Downloading VAE's & What Folder it goes in (models/VAE)
15:43 What a VAE can do and how to enable it (also CLIP Skip)
17:30 Latent upscale with VAE AMAZING and best way to upscale
18:28 Combing a LoRA & Embedding example
19:12 SadTalker output of GigaChad (see my video on SadTalker)
#StableDiffusion #aiart #stablediffusion2023 #stablediffusiontutorial #embedding #civitai - Věda a technologie
amazing tutorial series, please don't stop, you are my hero.. 😭
So much depth to stable diffusion so def more to come.
Aww I alway thought it would be nice if the Lora files had pictures to remeber them by, thanks!
Good tutorial for people starting to optimize the art. i think you meant the negative prompt not as usefull with regular checkpoint, which i agree to. Im bit suprised you dont have realistic vision, its my fav. Got best results with it. Whats your fav promt? Im going to try negative promts more from creators. Is positive as much valuable? I never used textual inversion, thought it was same as chekcpoints. How to even activate textual inversion? The last Jinx pic you made was amazing btw.
Hey thank you so much for your detailed tutorials! I saw in your description there's a link to a collection of Negative Embeddings. Should I download all of those, or only the EasyNegative safetensors?
It’s easy enough to download all negative embeddings since they are under a megabyte. You can also type in a negative prompt only in stable diffusion and click the disk icon at the top right to save that negative prompt. The civitai site is filled with examples for negative prompts you can use. After you click save you can access that negative prompt via the styles drop down near generate. I like the negative embeddings though because people actually rate them so you know if it’s good or not. I mainly try deliberates negative embedding first.
@@ArtificallyIntelligentAi Gotcha! Thank you so much!
Great video, Is there a quicklist command for embeddings?
Hey there! I just use the pink hypernetworks button under the generate button it has a search function. If you quick list now and click inside, it will show you all the commands available. At least I think it's all of them there are quite a bit.
Just a quick tip for the lazy ones "Stable-Diffusion-Webui-Civitai-Helper" is an extension that deals with downloading all the previews.
do i have toi put the images in the folders too?
Here is the site I got my custom models from civitai.com/ and at 18:20 I said I was going to do an infinite canvas (not to be confused with infinite zoom) which is a feature on Leonardo AI that allows you to outpaint for free without installing stable diffusion. Check it out here: czcams.com/video/QlTjGhRGJtc/video.html I also fixed my links in the video description.
How come CivitAI limits the textual inversions it shows me in all the tabs including the style tab? I can't see most of what you were seeing. For example the Atomic Punk. I can see it only if specifically search for it.
I know some of them are hidden if your not logged in, because of mature only filters. It also hides some by default when u sign up to keep it pg13
@@ArtificallyIntelligentAi I'm a silly billi. I was logged in but had published date set to Month instead of all time. Now I see it all.
@@peacetoall1858 awesome. Civit ai is good stuff. Good u can see All of it now.
bro my stable diffusion is running now but it is very slow it take all most 1 hour to generate 1 image i apply --xformers any solution do you have
get a better pc, u need 12gb vram rtx at least.
check out Invoke AI & Vlad Diffusion instead of Automatic 1111, they require less VRAM. There is also a trick you can do with your paging files that helps but I've only seen other people do that. Olivio Sarikas video on invoke and vlad diffusion. You can also lower your text2img creation to steps to 10 and render at 256x256 then do a latent upscale by throwing the low quality image you like into img2img and then generating with no prompt at 0.20 or less denoising and higher steps (at least 30), use one of the DPM Upscalers, and increase your resolution to 512x512, Then throw it into extras tab and do a 4x upscale, even then, might get an out of VRAM error. At some point, it just ends up that there is not enough VRAM, like Gutzim said.
thx
Anytime!
Downloaded the models ... stuck 'em in the folder .... soooooo .. how the hell do you get them to do anything? They don't show up in the drop down list for 'checkpoints'. How the hell does anyone get these things to work?!?!
So frustrating!
So sorry they should work if they are in the models / stable diffusion folder. You just have to do a full restart of stable diffusion and press the refresh button at the top left near the model selector or click on the pink button under the generate button to pull up your models and refresh it from there
hey hey hey pls reply, can all these be done in mac m1 ?
Hi 👋 I don’t have a mac but this is a really good site that has a install tutorial for stable diffusion stable-diffusion-art.com/install-mac/ . This site has it all , but I would start here and the custom models should work the same way
goofy