We explore in this video how to use LCM (Latent Consistency Model) Lora, which promises to speed up image and animation generation by 10 times.
Discord: / discord
#animatediff #comfyui #stablediffusion
============================================================
💪 Support this channel with a Super Thanks or a ko-fi! https://ko-fi.com/koalanation
☕ Amazing ComfyUI workflows: https://tinyurl.com/y9v2776r
🚨 Use Runpod and access powerful GPUs for best ComfyUI experience at a fraction of the price. https://tinyurl.com/58x2bpp5 🤗
☁️ Starting in ComfyUI? Run it on the cloud without installation, very easy! ☁️
👉 RunDiffusion: https://tinyurl.com/ypp84xjp 👉15% off first month with code 'koala15'
👉 ThinkDiffusion: https://tinyurl.com/4nh2yyen
============================================================
While LCM, in combination with AnimateDiff, does it, the detail quality is not great. However, by just adding a 2nd K sampler with few steps, but can generate an amazing animation, as good as without LCM. The extra step is, of course, at the expense of some additional rendering time. However, we can increase the speed by 2 or 3 times.
I also show some comparison of critical parameters to use in the KSamplers, to optimize a find a better balance of generation time vs. detail. All these, using the Instant Lora method and the conditional masking method showed in previous videos.
The workflow and frames to test this workflow are found in this Civit.AI article: https://tinyurl.com/w2xr3hj4
The LCM models
LCM LORA SD1.5: https://tinyurl.com/yc3x5rda
LCM LORA SDXL: https://tinyurl.com/ke27fxvh
Basic requirements:
ComfyUI: https://tinyurl.com/24srsvb3
ComfyUI Manager: https://tinyurl.com/ycvm4e29
Vast.ai: https://tinyurl.com/5n972ran
Runpod: https://tinyurl.com/mvbh46hk
Custom nodes:
AnimateDiff Evolved: https://tinyurl.com/yrwz576p
Advanced ControlNet custom node: https://tinyurl.com/yc3szuuf
VideoHelper Suite: https://tinyurl.com/47hka2nn
ControlNet Auxiliary preprocessors: https://tinyurl.com/3j3p6bjw
IP Adapter: https://tinyurl.com/3x3f2rfw
ComfyUI Impact pack: https://tinyurl.com/4jsmf8va
ComfyUI Inspire pack: https://tinyurl.com/2wkzezxm
KJ Nodes: https://github.com/kijai/ComfyUI-KJNodes
WAS node Suite: https://tinyurl.com/2ajuh2mx
Models:
DreamShaper v8 (SD1.5): https://tinyurl.com/3rka67pa
ControlNet v1.1: https://tinyurl.com/je85785u
vae-ft-mse-840000-ema-pruned VAE: https://tinyurl.com/c9t6wntc
ClipVision model for IP-Adapter: https://tinyurl.com/2wrtvnx4
IP Adapter plus SD1.5: https://tinyurl.com/2p8ykxf6
Motion Lora mm-stabilized-mid: https://tinyurl.com/mr42m5hp
Upscale RealESRGANx2: https://tinyurl.com/2frvcyca
LCM LORA SD1.5: https://tinyurl.com/yc3x5rda
LCM LORA SDXL: https://tinyurl.com/ke27fxvh
Tracklist:
[TBD]
My other tutorials:
Animatediff + ControlNet + conditional masking.i: • Animatediff perfect scenes. Any backg...
AnimateDiff and Instant Lora: • ComfyUI Instant Lora Animatediff ulti...
ComfyUI animation tutorial: • Stable Diffusion ComfyUI Animations ...
Vast.ai: • ComfyUI - Vast.ai: tutorial - how to ...
TrackAnything: • ComfyUI animation with segmentation: ...
Videos: Pexels
Music: Youtube Music Library
Edited with Canva, and ClipChamp. I record the stuff in powerpoint.
Subscribe to Koala Nation Channel: https://cutt.ly/OZF0UhT
© 2023 Koala Nation
#comfyui #animatediff #stablediffusion #lcm
Смотрите видео LCM + AnimateDiff High Definition (ComfyUI) - Turbo generation with high quality онлайн без регистрации, длительностью часов минут секунд в хорошем качестве. Это видео добавил пользователь Koala Nation 27 Ноябрь 2023, не забудьте поделиться им ссылкой с друзьями и знакомыми, на нашем сайте его посмотрели 9,756 раз и оно понравилось 256 людям.