Stable Diffusion doesn't have to be complicated
ฝัง
- เผยแพร่เมื่อ 22 พ.ค. 2024
- This video briefly covers every component of Automatic1111 UI for Stable Diffusion, from txt2img, img2img, and inpainting. Timestamps below:
0:00 Introduction
0:28 Checkpoints
1:47 Picking a model - CivitAI
2:38 Downloading and installing models
3:07 Text to Image - Prompting!
5:39 Sampling method / Sampling Steps
7:17 CFG Scale
8:30 Seeds and Batches
10:30 Misc settings - Restore faces, Hires.fix, x/y/z plots
12:03 Styles
13:08 Textual inversions, Hypernetworks, LORAs
14:26 Image to Image - Denoising Strength
18:26 Sketch
20:08 Inpainting
23:03 Inpaint Sketch, Inpaint Upload, Batch processing
24:10 Extras
24:28 PNG Info
25:29 Checkpoint Merging
25:45 Training Embeddings
25:58 Settings
26:03 Extensions
26:37 Outro
Very helpful for me as a total beginner! Thank you very much!
Hey mate, I just wanted to congratulate you and thank you for those videos. You videos about Stable Diffusion are BY FAR the best! I've been playing around with SD for a bit more than a week now and I've probably watched 100 videos about the topic. Unfortunately, almost all of them are complete garbage. Most "tutorials" I've seen so far a re just a collection of "click here", "download this" and "" enter that" - but nothing is actually explained. Your videos are so much better!
I was hoping to find more and had a look at your channel - just to see that there are only 7 videos and all of them are already a few months old with no updates at all. I really hope you haven't given up already and will be back soon with a new video (LoRA and LoRA training, please). Your videos are really excellent and I hope there will be some more uploads soon!
I've rarely heard someone explaining things that well. Thanks!
You are wonderful, calm and beautiful, explaining just what is needed and in a very very wonderful manner, I am currently, as you describe lost and confused with the many parameters in the program, but your simplification of them is very wonderful. Thank you very much for that.
I allready use it 2 month... but this was what i would have loved to see i n the beginning... great one.. abo clicked
This was EXCEPTIONALLY helpful as a new user. I watched your tip videos as well and they were perfect. I've only been using Stable Diffusion for a week and this upped my game exponentially. Thank you!
Just wanted to take the time to drop a comment and say this is exactly the type of video I needed to understand fully what it is that I'm doing. Thanks so much for doing what you do!
What an excellent video. Thank you.
9:57 batch size is often overlooked but for many cases I think it's preferable over batch count.
The UI will generate "batch size" images in parallel (whereas with batch count they're generated one after another), and that will speed things up (not proportionally though, batch size 4 should be faster than batch count 4, but not 4 times faster).
Wow this is great, you taught me a lot in this short video. Keep the great content coming!
I love the simple breakdown with examples format, this is a great introduction for beginners.
very good and helpful tutorial to start with S D
Thanks, this is super useful 🤗
Thank you man, this helped.
super good breakdown, keep em coming!
Great explanation!
Thanks man, awesome video 2 days I'm struggling trying to understand everything by myself by doing a lot of tests. This is really helpful, wish I found this video sooner but anyway everything is clearer now.
Really well done. Thanks
One of the most helpful videos for SD. Thank you so much for this.
thanks for this great and detailed video!
Great and clear
Excellent!
Very helpful, thanks.
Love to see a how to make your own LORA tutorial
Great video, really liked that you took it slow. I have this issue that alot of times when i use inpaint, the areas outside of the masked area also gets changed. For example if I mask a mans face, when i press generate his shoes will also change, despite having "inpaint masked" and "only masked" selected.
Batch size to 4 and get the live preview to 1, you can see 4 images generated at real time and you can skip to the next or abort the generation.
👍👍👍
Thank you for your comprehensive guide. Do you have any idea how Hassan Ragabs makes his videos?
Thanks a lot, you explanation is way better than many people, could you please explain about Lora in depth, training it and then how to use a model trained with a face in conjunction with different models? I tried to train it in Lora using my face but this didn't work properly, I couldn't make it work with different models downloaded from Civitai none of that. Thanks!
great video
by the way why every SD tutorial show example with cat?
I want to learn how to train my own model or to train an existing model.
I want to create an illustration for a book, once I generate the first cover art, how do I keep applying the art to different scenarios of the chapters
I just cannot get the big picture. Is Stable Diffusion some sort of engine under the bonnet what these individual entities Automatic 1111, txt2img, inpainting etc. are using in order to function?
hello there, does anyone knows how to get or enable the hires.fix option in SD? Mine doesn't have that option
I never use the vanilla sd 1.5 model to generate images. Can I delete it and just use my other models?
I have a question if anybody can help me. I’m desperately trying to create a full body image starting from a portrait. I have an half body character and I need to keep the exact face and change the “portrait” into a full body pose. Someone can help me?
Would you answer a question if you can? I'm new to AI. I'm not interested in Chat on any AI; however, I am interested in an AI that I can upload my own images and do a SwapFace (not using photos that come from online all the time). Is this AI available?
Continued: For my PC only, please.
But what is this web ui? How do I access it
We're all brand new to AI
This is the most stupidly complicated nonsense in the entire universe of AI content creation