Patreon
Get early access to build and test build, be able to try all epochs and test them by yourself on Patreon or contact me for support on Discord
IMPORTANT: Remember to check image sizes and use these proportions or you might not get what you want.
Install Instructions:
open stable diffusion -> extensions -> install from URL -> https://github.com/Mikubill/sd-webui-controlnet.git
click Apply and restart UI
Then download from https://huggingface.co/webui/ControlNet-modules-safetensors/tree/main the file control_depth-fp16.safetensor, control_openpose-fp16.safetensor control_canny-fp16.safetensor
place it inside extensions/sd-webui-controlnet/models
After all of this, you will have a ControlNet v1.1.112 just above Script in txt2image tab
Open it, place the pose (black and white image with depths is depth, black images with colored sticks is openpose, black and white images like drawing is canny, not the example one) you want to replicate by selecting it from your computer and place it in the image selection square box. You can also drag and drop.
Select preprocessor NONE, check Enable Checkbox, select control_depth-fp16 or openpose or canny (it depends on wich poses you downloaded, look at version to see wich kind of pose is it if you don't recognize it in Model list) check Controlnet is more important in Control Mode (or leave balanced).
Then generate your image, don't forget to write a proper prompt for the image and to preserve the proportions of the controlnet image (you can just check proportions in the example images for example).
I uploaded the pose images and 1 example generated image with that pose using the same prompt for all of them. But you can modify the prompt to fit what you want.
Enjoy!
If you'd like to support my efforts and have access to early and bleeding edge builds, please think about joining my Patreon: https://www.patreon.com/betterCuteAsians
Description
This is a depth controlnet poses pack.
Remember to use same proportions of depth image.
I advice to add portrait keyword to your prompt.
FAQ
Comments (15)
can you add json files to your models plz?
Appologies for the noob question, but where do I save these pose files within the stable diffusion folder structure? I've been googling for a while now and I can't find it. TIA!
These poses are made to work with ControlNet - the stick figure files work with Control Net's open pose models. And the white silouethes with ControlNet's Depth models.
To start using control net if you are using Automatic1111 you will need to grab the extenstion which is named - sd-webui-controlnet in the extensions tab.
After you have installed and applied it to your system, you will then need the aformentioned openpose models. To grab those look for Control Net models here on civit ai. Either ControlNet 1.1 Models or the ControlNet Pre-Trained Models contain the models you need. I am not exactly sure about the differences between the two.
You will then need to install these models, and then proceed with controlnet and applying the poses from this post.
I honestly recommend watching an example of someone using openpose for controlnet or a controlnet guide in general since it is a very powerful tool.
Personally I use these:
OpenPose: https://civitai.com/models/38784?modelVersionId=44811
Depth: https://civitai.com/models/38784?modelVersionId=44736
Good luck with your prompting!
PS to give a direct answer to your question the files from this post dont need to be in your stable diffusion folder at all they simply get added as an input to ControlNet, just like you would drop an image in for img2img
@Keyaladron Thank you!
I only see the depth and example images in the download. what am I missing?
this is depth Poses Packs Collection ... read description
sorry for the noob question, but will these work with SDXL?
no, lora's either work with SD 1.5 (or other versions) or SD XL.
yes, these are images, no LoRAs. basically you run it with a preprocessor in controlnet, it behaves the same across SD15 and SDXL.
@NextMeal thx, man!
Just to reiterate for anyone who might be confused by the conflicting answers, these will work with SDXL or FLUX, or probably a bunch of wildly different ai models if they have something like controlnet. THe reason is these are poses not loras. A lora is like a small extra model you use with a model, and it has to match the model.
THIS is just a pack of basically stick figures posed in certain ways. And so controlnet looks at the stick figures and makes an image where the figures line up with the stick figures.
(this explanation is simplified for ease of understanding and thus slightly inaccurate on a technical level)
A noob Q here: after successfully installing controlnet, there's no 'extensions/sd-webui-controlnet/models' folder anywhere, where is it supposed to be located?
In the top directory for Stable Diffusion:
stable-diffusion-webui
@leytewolf752 I use ComfyUI and there's nothing like that.
Good selection of poses. Much appreciated!