Sign In

Ars Mentalia - ComfyUI Workflow - SD3, XL + 1.5 / i2i / t2i

26
708
5
Updated: Jan 8, 2025
toolartimg2imgcomfyuitxt2imgxl
Type
Workflows
Stats
250
0
Reviews
Published
Jun 15, 2024
Base Model
SD 3.5 Medium
Hash
AutoV2
7A388A7E08
default creator card background decoration
EXDYSA's Avatar
EXDYSA
This Stability AI Model is licensed under the Stability AI Community License, Copyright (c) Stability AI Ltd. All Rights Reserved.
Powered by Stability AI

S D 3 . 0 R E A D Y ! VAE在ckpt內部,使用像這樣內建CLIP的版本 VAE is inside ckpt, CLIP built in is most convenient : https://civitai.com/models/497255?modelVersionId=568278

自動備份和容錯。 強大的流量控制。正确的语法,返回中文: SD3、SDXL和SD1.5之间的交换。 從2GB的VRAM到尖端的發電器。 包括說明。 感謝您的下載。Txt2Img & Img2Img. Automatic backup and fault tolerance. Powerful flow control. Exchange between SD3, SDXL, SD1.5. Scales from 2GB VRAM up to cutting edge hardware. Instructions included. Thank you for your download.
06/14/24 RAM use lowered by 4GB. 6GB RAM or even 4 can now run SD3 with this workflow. Cascade removed in 0.0.59. 3.0 is superior. Documentation++. Intuitive symbols. UniPC works? Still testing other optimizations...

˶𝞢⤬⫒ⵖsᐼ˶ ⠑⠭⠙⠽⠎⠁

Older patch notes :
06/13/24 3.0 Latent isn't backwards compatible with SD and XL 🙃 06/12/24 3.0! Early release, 30% speed boost + A1111 compatability WIP 06/07/24 Bugfix & added dedicated LoRA nodes.
Evaluated on	2017 i7 Windows 11 8GB RAM Nvidia 1050ti 4GB VRAM
				2012 i3 Arch Linux 4GB RAM Nvidia 1030 2GB VRAM

Required nodes are part of Manager and should auto-fetch https://github.com/ltdrdata/ComfyUI-Manager

  • Be sure to update your ComfyUI installation!

Minimum prerequisities for manual installation are :

Some versions also use

Contact for commercial use.

no step on snek (alert!) 🐍