floyo logo
Powered by
ThinkDiffusion
⚡️Nano Banana 2 ⚡️ just landed. Start creating now.
floyo logo
Powered by
ThinkDiffusion
⚡️Nano Banana 2 ⚡️ just landed. Start creating now.

Vertical Video Character Face & Actor Swap (Wan 2.2 Animate)

432

Generates in about -- secs

Nodes & Models

PrimitiveInt
CLIPLoader
umt5_xxl_fp8_e4m3fn_scaled.safetensors
CLIPVisionLoader
clip_vision_h.safetensors
MarkdownNote
Note
VAELoader
wan_2.1_vae.safetensors
LoadImage
UNETLoader
Wan2_2-Animate-14B_fp8_e4m3fn_scaled_KJ.safetensors
CLIPTextEncode
CLIPVisionEncode
ModelSamplingSD3
GrowMask
PreviewImage
WanAnimateToVideo
KSamplerAdvanced
VAEDecode
CreateVideo
SaveVideo
DownloadAndLoadSAM2Model
sam2_hiera_base_plus.safetensors
Sam2Segmentation
DownloadAndLoadSAM2Model
sam2_hiera_base_plus.safetensors
Sam2Segmentation
LoRA Stacker
wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors
WanAnimate_relight_lora_fp16.safetensors
LoRA Stacker
wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors
WanAnimate_relight_lora_fp16.safetensors
LoRA Stacker
wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors
WanAnimate_relight_lora_fp16.safetensors
VHS_LoadVideo
easy loraStackApply
PointsEditor
BlockifyMask
DrawMaskOnImage
PixelPerfectResolution
DWPreprocessor
PixelPerfectResolution
PixelPerfectResolution
DWPreprocessor
DWPreprocessor
DrawMaskOnImage
MaskPreview+

What this workflow does
Turn a talking character in your vertical video (9:16) into a new person using just one reference image. The background, camera move and lighting stay the same — only the actor’s face/character changes. Perfect for Reels/Shorts/TikTok style shots.

Basic inputs

  • Reference Image – Clear front-facing image of the new character/actor.

  • Source Video – Vertical clip (9:16) where the main person is clearly visible.

  • Masking – Use the green points to mark the face/character you want to replace. Keep masks tight to the head/face so the body and scene stay consistent.

How it behaves

  • Keeps original camera motion, timing and background.

  • Swaps only the character’s face/identity to match your reference image.

  • Designed for short social clips so you can preview quickly and iterate.

Recommended settings (safe starting point)

  • Resolution: Vertical 9:16 (for example 1080×1920), values in multiples of 16.

  • Clip length: Short social-friendly shots (a few seconds) give faster, cleaner results.

Read more

N
l
lhs1978
1 month ago
Gastei 20 min e nada foi fetio e ainda de erro... Qdo acontece isso tem que devolv er os crreditos.

Reply