powercen
Follow
Stars
Fast and memory-efficient exact attention
Quantized Attention that achieves speedups of 2.1-3.1x and 2.7-5.1x compared to FlashAttention2 and xformers, respectively, without lossing end-to-end metrics across various models.
Official implementation of "FitDiT: Advancing the Authentic Garment Details for High-fidelity Virtual Try-on"
ControlNet scheduling and masking nodes with sliding context support
Accelerate FLUX inferencing speed for ComfyUI.
ComfyUI nodes for the Ultimate Stable Diffusion Upscale script by Coyote-A.
ComfyUI node for background removal, implementing InSPyreNet the best method up to date
[WIP] The all in one inference optimization solution for ComfyUI, universal, flexible, and fast.
A node for ComfyUI that performs GPEN face restoration on the input image(s). Significantly faster than other implementations of GPEN.
Tiled Diffusion, MultiDiffusion, Mixture of Diffusers, and optimized VAE
Extension for Automatic1111 and ComfyUI to automatically create masks for Background/Hair/Body/Face/Clothes in Img2Img
A selection of nodes for Stable Diffusion ComfyUI
A set of nodes for ComfyUI that can composite layer and mask to achieve Photoshop like functionality.
Use RetinaFace to detect and automatically crop faces
a node for comfyui for restore/edit/enchance faces utilizing face recognition
Quick Connections for ComfyUI & litegraph
you can use TRELLIS in comfyUI for img to 3D
Official repo for paper "Structured 3D Latents for Scalable and Versatile 3D Generation".
Fixed Attention Couple, NegPip(negative weights in prompts) for SDXL, FLUX and HunyuanVideo, more CFG++ and SMEA DY samplers, etc.
A set of ComfyUI nodes providing additional control for the LTX Video model