Complete Controlnet Guide for Stable diffusion and batch image generation in A1111

#stablediffusion #A1111 #AI #Lora #koyass #sd #sdxl #controlnet #canny #depthmap #openpose #batchimage #batch #ai #imagegeneration #generativeart #generativeai
#characte
This video will be hopefully a comprehensive explanation of Controlnet use in Stable diffusion
The Video will include the following:
00:00:00 introducing samples results based on controlnet usage in A
00:00:38 installation and setup tips
00:01:02 Contorlnet user interface explanation
00:01:28 downloading controlnet models
00:03:07 start testing Controlnets
00:03:57 canny controlnet for strict edge adherence
00:04:11 depth controlnet and important settings
00:07:36 normal maps controlnet for 3D like recognition
00:08:25 open pose controlnet for human position
00:10:42 canny controlnet for text output in SD 1.5
00:11:00 MLSD for straight lines and architecture
00:11:43 lineart for anime or real pictures like hand drawing
00:12:06 scribble controlnet for simple hand drawing outlines
00:12:40 softedge softer than canny and lineart
00:13:10 seg segmentation of the image to objects
00:13:55 shuffle controlnet for shuffling (reordering structure) image elements
00:14:57 tile controlnet for details changes of an image
00:15:44 inpaint controlnet for inpainting or outpainting
00:16:35 IP2P adding elements to existing pexel image, like burn it, make it ice
00:17:19 Reference controlnet to change image style based on reference image
00:18:17 multiple controlnet usage
00:19:36 batch controlnet for applying same prompt to series of reference images
00:20:28 batch open pose book application using one prompt
Update: IP Adapter is really impressive with style application
Openpose DW is better than open pose
SDXL control nets maybe used with SDXL models.
officila controlnet 1.1 models repositry
huggingface.co/lllyasviel/Con...
A1111 for stable diffusion
github.com/AUTOMATIC1111/stab...

[@fokeyjo]

Really handy way of showing off the features of controlnet.. I was a bit lost as to what it does. Thanks!

[@AI-HowTo]

Glad it was helpful

[@DanielSchweinert]

Thank you very much! Keep up the good work! Finally someone who is explaining all these ControlNets. The official info about them are very sparse.

[@AI-HowTo]

Much appreciated! hopefully it is useful and clear enough video.

[@CabbageGarbage123]

Thank you very much for showcasing everything ! And thank you for showing Adetailer, I really never knew about it now I don't need to inpaint every image after generation !

[@AI-HowTo]

You are welcome, after detailer should indeed by integral part of A1111.

[@ricardopicchi1509]

Great content! Thank you for sharing and keep the great work. Just found your channel, really good 👍🏻

[@AI-HowTo]

Thank you for your supporting words, best of luck to you as well.

[@widitruth]

Amazing, thanks

[@jag9022]

great explanations thansk !

[@AI-HowTo]

You are welcome!

[@rikimarudonchelie1229]

Thanks so much for this...really makes sense & clear the confusion of its powerful function of each buttons & perform actions & render...hope to see more explanations... just subscribe.

[@AI-HowTo]

Glad it was helpful

[@AI-HowTo]

and welcome

[@Lell19862010]

Is there any possibility to use batch with openpose, making each image with a different seed?

[@proddeouf]

Hey ! thanks for the great tuto ! How to generate only depth map or pose pictures with CN, and use each of them as a mask for batch apply after ? I want to use it as mask for an image sequence for doing videos.

[@AI-HowTo]

You are welcome, in Img2Img if you apply controlnet in Batch image processing, it will automatically create the mask for the current input image, as i display in this video kzbin.info/www/bejne/oWW9iH16a7xrmpI so you can use controlnet for openpose/depth map combination for instance and it will apply that .... this will not however save the openpose or depthmap pictures indepenently .... if you want them independent for other purposes one can possibly use ComfyUI and save the preprocessor mask output directly using save image node ... or in A1111 one may use for instance Rembg script in Extras tab and check Return Mask option (this requires Rembg extension installation), similarly there should extension for openpose such as openpose editor which may allow you to generate poses and save them as images if that is what you mean.

[@proddeouf]

@@AI-HowTo thanks for sharing the tip ! I will do some experiments and hope it’s functioning ! I will try the extensions and give a shot to ComfyUI, btw. Maybe it will be better. I want to create a mask as pre production to light the process;

[@gustavosuarez7945]

Great tutorial! Can this be achieved with Comfyui ?

[@AI-HowTo]

Yes, absolutely, as both use the same Controlnet models, and the same principles, the application methodology changes... what matters is to understand what these models do then discover how they can be used in the target tool such as A1111/ComfyUI

[@camspider9887]

That is a truly awesome guide; I have just subscribed. If I may ask, what should I do if I want to exclude a specific part from the Canny model? Here's my scenario: I have a custom model, and I want to mimic images, such as the pose and environment. To achieve this, I've created two instances in ControlNet 1- OpenPose 2- Canny. The issue I'm encountering is that Canny is replicating the overall shape of my head from the source image. How can I exclude the head part from Canny? I hope you understand my question, and a big thanks in advance.

[@AI-HowTo]

thanks, canny can be very strict, this is why we can use depth maps for example to reduce this strickness and get the overall shape of the body, or change the edge detection settings to make it weaker .... not sure however if we can exclude a direct part of the canny processor automatically in A1111, I dont this this feature exsits at the time being, more complicated workflows can be performed in ComfyUI but that requires more work.... if you are working on a single image then you can use inpaint afterwards to fix it, but for automation ComfyUI is the way to go in complicated scenarios.

[@camspider9887]

definitely gonna use inpaint afterwards, thatnks in advance @@AI-HowTo

[@StorylineYoutube]

Great one! Could I maybe have you pose folder?

[@AI-HowTo]

I got them in the past from civitai, not sure from which source, so I give you a better link civitai.com/tag/poses which has over 700 collections of different poses/collections with their previews that you can download any .

[@AI-HowTo]

some very good pose collections are: civitai.com/api/download/models/33869 civitai.com/api/download/models/22375 civitai.com/api/download/models/21847 civitai.com/api/download/models/22265 civitai.com/api/download/models/93697 civitai.com/api/download/models/59119 etc.

[@StorylineYoutube]

you're the best! keep going@@AI-HowTo

[@NYUB]

Noob here. What SD interface are you using?

[@AI-HowTo]

A1111, Automatic 1111, You may consider watching this kzbin.info/www/bejne/iKXNdaatl7h4e7s if you are new to this free tool.

[@Andonchoto]

i need to downgrade CN, how do i do that please?

[@AI-HowTo]

I am not sure if you can! if you check this video studio.kzbin.infoPdkBmDoUIy8/edit for downgrading A1111 there was a release history with hascode allowing us to go back to each previous release, but in Controlnet there is not on github, so probably you cannt downgrade since info doesnt exist online, there might be a way to rollback, but i am not sure how .... there is only an older archieved version which is v1 github.com/lllyasviel/webui-controlnet-v1-archived which can be installed by uninstalling current contorlnet extension, then installing this from the link i just provided.