🚀Edit& Run it online (RunningHub):
https://www.runninghub.ai/post/1846119782923038721
[Workflow introduction]
Open the workflow. First, upload a style reference image at the leftmost loading image of the screen, and upload a multi-angle view at the white loading image below. Use ipadapter to transfer the image, Joy_caption to reverse the prompt words of the image, and then use Controlnet to control the image and give it to the sampler for raw image. HD Ultimate SD Upscale is used to enlarge and optimize the image later.
[Usage scenario]
The convenience of this workflow lies in generating multi-views of corresponding styles through style images, which brings creativity and ideas to friends who make games, and facilitates the rapid implementation of work.
[Key nodes]
ipadapter
[Model version]
SDXL
Model name: CHEYENNE v16safetensors
[LoRA model]
None
[ControlNet application]
OpenPoseXL2 loader
Openpose pose preprocessor
Start time: 0
End time: 1
Intensity: 1
Depth Anything V2
control-lora-depth-rank128
Start time: 0
End time: 0.34
Intensity: 0.34
[K sampler]
CFG: 8
Sampling method: dpmpp_2m
Scheduler: karras
Noise reduction: 1
Description
FAQ
Comments (4)
It looks good, how can I run it locally?
you can download it and try to run locally,but it will need more time to setup
I ran the cloud version and its quite impressive. I was wondering can you make a workflow that takes in image as input, user describes a pose and expression and then it can change the original image based on that ? this is achievable on app.runwayml.com but its not free.
stay idle in cloud , not working