Wan2.2 Fun and RealismBoost LoRA for V2V
Enhancer
LoRA
Video2Video
Wan
1
965
Video Detail Enhancer is an advanced workflow powered by the Wan2.2 Fun model and RealismBoost LoRA, designed to fix common issues in AI-generated videos such as blurry faces, hands, and textures, while dramatically upgrading the realism, sharpness, and consistency of animated content. By combining the superior Mixture-of-Experts (MoE) architecture of Wan2.2 Fun with a second-pass video-to-video (V2V) refinement workflow and the precision guidance of RealismBoost LoRA, this solution turns “average” AI video outputs into professional-grade cinematic sequences.
What is RealismBoost LoRA?
RealismBoost LoRA is a LoRA fine-tune specifically trained to correct facial and hand artifacts, enhance eye gleam, mouth expressions, and micro-details lost in most fast synthesis pipelines.
Use Cases
Anime, influencer, or character animation with clean identity, clear expressions, and detailed motion.
Marketing/product videos needing high-fidelity visuals and smooth transitions.
Film, storyboard, and music video concept prototyping with camera moves and scene progression.
Social media content that stands out with professional polish on both facial features and background elements.
Read more
0
Reply
Nodes & Models
Note
MarkdownNote
Fast Groups Bypasser (rgthree)
CLIPTextEncode
TorchCompileModel
Reroute
VAEEncode
LoraLoaderModelOnly
4steps-lora-250928-low_noise_model.safetensors
Wan14B_RealismBoost.safetensors
Wan2.2-Fun-A14B-InP-low-noise-HPS2.1.safetensors
ModelSamplingSD3
ContextWindowsManual
KSampler
VAEDecode
INTConstant
INTConstant
ImageResizeKJv2
PathchSageAttentionKJ
WanVideoEnhanceAVideoKJ
ImageConcanate
VHS_LoadVideo
VHS_VideoInfoLoaded
VHS_VideoCombine
ImageResizeKJv2
ImageConcanate
easy cleanGpuUsed
Video Detail Enhancer is an advanced workflow powered by the Wan2.2 Fun model and RealismBoost LoRA, designed to fix common issues in AI-generated videos such as blurry faces, hands, and textures, while dramatically upgrading the realism, sharpness, and consistency of animated content. By combining the superior Mixture-of-Experts (MoE) architecture of Wan2.2 Fun with a second-pass video-to-video (V2V) refinement workflow and the precision guidance of RealismBoost LoRA, this solution turns “average” AI video outputs into professional-grade cinematic sequences.
What is RealismBoost LoRA?
RealismBoost LoRA is a LoRA fine-tune specifically trained to correct facial and hand artifacts, enhance eye gleam, mouth expressions, and micro-details lost in most fast synthesis pipelines.
Use Cases
Anime, influencer, or character animation with clean identity, clear expressions, and detailed motion.
Marketing/product videos needing high-fidelity visuals and smooth transitions.
Film, storyboard, and music video concept prototyping with camera moves and scene progression.
Social media content that stands out with professional polish on both facial features and background elements.
Read more
0
Reply