ComfyUI Models & LoRAs
High quality diffusion checkpoints, FP8/FP16 distilled weights, and style LoRA models.
No models match your filter
Try selecting "All" or clearing your search term.
AI Hands XL
Specialized SDXL negative/positive conditioning LoRA fixing extra fingers, fused thumbs, and awkward hand positioning.
Eliminating extra fingers & fused thumbs in SDXL portraits
aidmaNSFWunlock (FLUX Unlock LoRA)
Popular FLUX.1 prompt guidance unlocking LoRA expanding anatomical and figure drawing freedom in FLUX generations.
Unlocking anatomical figure drawing freedom for FLUX.1
ByteDance Hyper-FLUX Acceleration LoRA
ByteDance 8-step / 16-step Hyper-SD acceleration LoRA for FLUX.1 models cutting inference time by 75%.
8-step / 16-step ultra-fast FLUX.1 inference acceleration
ClearHandsXL
Lightweight SDXL hand restoration adapter designed to enforce crisp finger separation and clean fingernails.
Enforcing crisp finger separation & clean fingernails
CLIP-G OpenCLIP BigG Encoder
OpenCLIP ViT-bigG text encoder weights used in SDXL dual encoding and Stable Diffusion 3.5 architectures.
SDXL & SD3.5 dual text conditioning
CLIP ViT-L Text Encoder
Standard OpenAI CLIP ViT-L/14 text encoder required across FLUX.1, SDXL, and Stable Diffusion workflows.
Universal text encoding for FLUX, SDXL & SD1.5
ControlNet Canny LineArt XL
High-precision edge detection ControlNet preserving sharp outline boundaries and sketch-to-image conversion.
Line art coloring & sharp edge structural guidance
ControlNet Depth Anything V2
State-of-the-art depth map conditioning model providing clean spatial geometry and foreground-background separation.
Spatial depth conditioning & 3D layout control
ControlNet OpenPose XL
Human body pose keypoint conditioning model enforcing exact character skeletal postures and hand positions.
Human body skeletal pose & gesture conditioning
CyberRealistic V5.0
Popular SD 1.5 photorealism checkpoint optimized for lifelike human skin textures, expressive eyes, and casual photos.
Lifelike portrait photography & casual selfie generation
Detail Tweaker XL LoRA
Essential SDXL detail enhancer LoRA allowing dynamic weight adjustment (-5.0 to +5.0) to fine-tune image clarity and sharpness.
Fine-tuning SDXL image micro-details & sharpness
Disney Pixar 3D Style LoRA
Transforms characters and scenes into 3D animated film aesthetics with soft subsurface scattering and vibrant colors.
Disney / Pixar 3D animated movie art style
DMD2 SDXL 4-Step Distillation LoRA
Distribution Matching Distillation 2 (DMD2) adapter delivering 4-step ultra-fast generation with high text alignment.
4-step DMD2 distillation for rapid batch generation
DreamShaper 8 (SD1.5)
The definitive SD 1.5 general-purpose checkpoint optimized for fast generation, low VRAM footprint, and artistic versatility.
Low-VRAM fast generation, digital art & fantasy portraits
DreamShaper XL
Versatile SDXL model providing an optimal balance between photorealism, digital art, illustration, and 3D rendering.
Balanced digital illustration, concept art & fantasy scenes
epiCRealism Pure Evolution
Highly detailed SD 1.5 realistic model rendering natural un-retouched photo aesthetics, skin imperfections, and studio lighting.
Un-retouched realistic portrait photography & natural lighting
FLUX.1 [dev] Checkpoint
Black Forest Labs 12B parameter guidance-distilled model offering unmatched prompt adherence and photorealistic details.
Photorealistic text-to-image & complex typography
FLUX.1 [schnell] 4-Step Model
Apache 2.0 open-weight 4-step distilled FLUX model built for commercial production and real-time interactive generation.
Ultra-fast 4-step real-time generation & commercial use
FLUX.1 VAE (ae.safetensors)
Official Black Forest Labs VAE decoder for FLUX.1 [dev] and [schnell] decoding latent samples to crystal-clear RGB images.
FLUX.1 latent decoding to high-res images
FLUX Aesthetic Film Grain LoRA
Cinematic 35mm film grain and color grading LoRA for FLUX.1 models adding organic analog camera texture.
Cinematic 35mm film grain & analog color grading for FLUX
FLUX Kontext [dev]
Multi-reference image editing and context-aware image-to-image generation model maintaining subject consistency.
Multi-reference image editing & subject consistency
FLUX.2-dev Turbo LoRA
Fast 4-to-8 step sampling acceleration LoRA for FLUX [dev] pipelines reducing sampling steps while preserving detail.
Accelerating FLUX [dev] sampling steps
Hands XL & Multi-Model Anatomy Fix LoRA
Comprehensive hand and anatomy refinement LoRA suite compatible across SDXL, SD 1.5, FLUX.1, Pony, and Illustrious checkpoints.
Fixing multi-model finger anatomy, hand gestures, & palm details
HunyuanVideo 1.5
Tencent Hunyuan open-source high-definition video diffusion model generating 720p/1080p cinematics with physics fidelity.
Cinematic 1080p video generation with realistic physics
Ideogram 4.0
State-of-the-art text rendering and typography model for logos, poster design, and graphic composition in NF4/FP8 quantizations.
Typography, logo design, text rendering & poster layout
Illustrious XL V2.0
High-resolution SDXL anime and illustration model providing vibrant line art, anime aesthetics, and character poses.
Anime illustration, manga line art & character design
IP-Adapter FaceID Plus V2
InsightFace embedded facial identity adapter producing high-consistency character replicas from a single photo.
High-consistency character face cloning from single photo
IP-Adapter SDXL Plus Model
Official IP-Adapter SDXL image-prompt adapter weights enabling zero-shot style transfer and reference face cloning.
Zero-shot image prompt conditioning & reference image style transfer
Juggernaut XL (Ragnarok)
The premier photorealistic SDXL checkpoint excelling at cinematic lighting, human anatomy, portraits, and architectural scenery.
Cinematic photorealism, skin textures, and environment art
Krea-2 Turbo
Next-gen 8-step ultra-fast diffusion model for instant real-time rendering, photo-editing, and workflow training.
Real-time rendering, fast generation & LoRA training
LCM-LoRA SD 1.5 Acceleration
Latent Consistency Model (LCM) adapter enabling 4-step generation for low-VRAM SD 1.5 setups.
4-step fast generation for low-VRAM SD1.5 workflows
LCM-LoRA SDXL Fast Sampler
Latent Consistency Model (LCM) acceleration LoRA enabling 4-to-8 step sampling without quality degradation on SDXL.
4-to-8 step real-time sampling acceleration on SDXL
LightX2V Wan 2.2 Video Turbo LoRA
4-to-8 step video sampling acceleration LoRA for Wan2.2 video generation models reducing video rendering times.
4-step fast video rendering for Wan 2.2 models
LTX-Video 2.5
Lightricks LTX-Video 2.5 real-time video diffusion model featuring synchronized audio generation and motion control.
Real-time video generation with synchronized audio
Midjourney Mimic V6 Style LoRA
Replicates Midjourney v6 aesthetic style, dramatic lighting, rich color saturation, and cinematic framing for SDXL.
Midjourney v6 cinematic aesthetic & rich color styling
MiniMax H3 Video Model
Omni-modal high-definition video and audio generation model providing synchronized speech, sound FX, and fluid motion.
Synchronized video-audio generation & fluid camera motion
Pony Diffusion V6 XL
The flagship anime, stylized art, and character generation checkpoint built on SDXL requiring CLIP Skip 2 and score_ tags.
Anime illustration, character art & specialized score tagging
Pony Style & Character LoRA Adapter
Universal style and character adapter framework for Pony Diffusion V6 XL enabling precise costume and art style control.
Pony V6 XL character clothing & art style adaptation
Portrait Engine FLUX / FLUX 2 KLEIN (Detailed Skin)
Next-gen skin texture and facial pore engine LoRA for FLUX.1, FLUX 2 Klein, and Z-Image DiT checkpoints.
Ultra-detailed skin pores, micro-lighting, & portrait skin rendering
Proper Hands for SDXL
Anatomy correction LoRA focused on natural wrist proportions, realistic hand poses, and hand-held object interaction.
Natural wrist proportions & realistic hand object interaction
Qwen-Image Edit
Alibaba Qwen multimodal image editing model supporting bilingual Chinese/English prompts and natural image modifications.
Bilingual Chinese/English image editing & text-based photo manipulation
Qwen3-VL Vision Text Encoder
Multimodal Qwen3-VL text and vision encoder for Krea-2 and next-gen DiT model conditioning.
Krea-2 & multimodal DiT model conditioning
Realism LoRA by Stable Yogi (Illustrious)
Photorealistic style adapter specifically tailored for Illustrious XL anime checkpoints.
Semi-realistic rendering for Illustrious XL artwork
Realism LoRA by Stable Yogi (Pony)
Photorealistic rendering adapter for Pony Diffusion models blending 2D/3D anime styles into realistic human photography.
Transforming Pony anime models into photorealistic 3D portraits
Realism LoRA by Stable Yogi (SDXL)
High-fidelity realism LoRA for SDXL enhancing skin micro-texture, natural lighting, and photographic depth.
Enhancing portrait realism & camera depth on SDXL
Realistic Vision V6.0 (SD1.5)
The iconic low-VRAM photorealism checkpoint for SD 1.5 delivering natural portraits, lighting, and textures.
Low-VRAM realistic portrait photography & natural skin tones
RealVisXL V5.0
State-of-the-art hyper-realistic SDXL model specialized in ultra-detailed portraits, natural skin pores, and studio lighting.
Studio photography, realistic skin textures & portraits
SDXL Base 1.0
Stability AI flagship base model featuring dual text encoders (OpenCLIP & CLIP ViT-L) for native 1024x1024 generation.
Base model fine-tuning & general 1024x1024 generation
SDXL VAE FP16 Fix
Fixed SDXL VAE weights by MadeByOllin preventing NaN black screen artifacts during half-precision FP16 VAE decoding.
Preventing NaN black images in SDXL FP16 workflows
Skin Realism XL (Skin Details & Imperfections)
Specialized SDXL skin detail LoRA adding subtle pores, natural freckles, blemishes, and realistic skin specular highlights.
Adding realistic pores, freckles, & skin specular highlights
T5XXL Text Encoder (FLUX / SD3)
Essential Google T5-XXL text encoder weights required for FLUX.1 and SD3 prompt understanding in FP16 and FP8.
FLUX.1 & SD3 text prompt encoding
UMT5-XXL Text Encoder (Wan2.2)
Multilingual UMT5-XXL text encoder weights required for Wan2.1/2.2 video model prompt processing.
Wan2.2 video model multilingual text encoding
Wan 2.2 (T2V / I2V 14B)
Tencent Wan-AI 14B MoE open-weight video model supporting high-definition Text-to-Video and Image-to-Video generation.
HD text-to-video & image-to-video motion generation
Z-Image Base Model
Full-precision Z-Image Diffusion Transformer (DiT) base checkpoint for maximum generation quality and fine-tuning.
High-fidelity DiT generation & base model fine-tuning
Z-Image Turbo (DiT)
Next-generation 8-step Diffusion Transformer (DiT) optimized for ultra-low VRAM GPUs (8-16GB) with fast generation times.
Fast 8-step generation on consumer GPUs with minimal VRAM
Showing 1–24 of4 models