Freya Cherry Kimono: ReActor Face Swap & GFPGAN Facial Restoration

Full Resolution

Generation Specs RTX 4070

Sampler reactor

Scheduler insightface

Steps 1

CFG Scale 1

Seed 42

Target VRAM 8GB

VRAM 6GB - 8GB

GPU Hardware Compatibility

Optimal Ground Truth

Native GPU execution at full throughput with zero memory swapping.

ReActor Face Swap (InsightFace InSwapper 128 + GFPGAN) Min 6GB VRAM 1024×1024 RTX 4070 Verified

Freya Cherry Kimono: ReActor Face Swap & GFPGAN Facial Restoration

High-fidelity face-swap workflow combining InsightFace InSwapper 128 with GFPGAN v1.4 restoration. Seamlessly maps Freya's identity onto a traditional cherry blossom silk kimono portrait with authentic skin texture and zero boundary artifacts.

Blueprint Summary RTX 4070 Verified

Reproducible ComfyUI workflow for Freya Cherry Kimono: ReActor Face Swap & GFPGAN Facial Restoration using ReActor Face Swap (InsightFace InSwapper 128 + GFPGAN) at 1024×1024 resolution. Requires minimum 6GB VRAM with sampler reactor and scheduler insightface (1 steps). Includes 1-click terminal model sync and canvas JSON graph.

Node Graph Pipeline 4 Total Nodes

Verify in Resolver

01 Decode & Save

SaveImage

Latent to pixel space

Execution DAG Topology Interactive Visualizer

Drag to pan · Scroll to zoom · Hover wires

Model & Asset Setup 1-Click Script

Run in your ComfyUI root:

curl -fsSL https://decomfy.com/api/scripts/freya-cherry-kimono-reactor-faceswap.sh | bash

Positive Prompt

Freya, 1girl, 20yo Scandinavian blonde, high cheekbones, delicate features, wearing traditional Japanese silk kimono with delicate cherry blossom floral patterns, Kyoto spring garden background, soft natural daylight, photorealistic 8k, sharp focus.

Negative Prompt

blurry, low resolution, bad eyes, distorted face, plastic skin, artifacts, double chin

Required Models 2 Models

Disk Space Required: 902 MB (2 models · DiT/Base: 902 MB)

models/insightface/ 554 MB HuggingFace

inswapper_128.onnx InsightFace InSwapper 128 ONNX checkpoint for facial feature embedding transfer

Get

models/facerestore_models/ 348 MB HuggingFace

GFPGANv1.4.pth GFPGAN v1.4 face restoration checkpoint for realistic micro-texture recovery

Get

Field Notes RTX 4070 Benchmark

Benchmark: 2.1s on NVIDIA GeForce RTX 4070 (6.2 GB VRAM peak, InSwapper 128 + GFPGAN v1.4, 1024x1024)

Tested and benchmarked live on dedicated RTX 4070. Uses InsightFace 512D facial embedding extraction and InSwapper 128 ONNX with RetinaFace ResNet50 detection. Enhanced with GFPGAN v1.4 face restoration at 100% visibility to recover pore-level skin micro-textures and eliminate unnatural plastic blurring. Executes in 2.1 seconds on consumer hardware.

Frequently Asked Questions FAQ

What GPU and VRAM are required to run Freya Cherry Kimono: ReActor Face Swap & GFPGAN Facial Restoration?

This workflow requires a minimum of 6GB VRAM (recommended 8GB VRAM). Tested and verified on NVIDIA GeForce RTX 4070 (12GB VRAM) at 1024x1024 resolution.

How do I resolve missing custom nodes for this workflow?

You can drop the workflow JSON into our client-side Missing Node Auto-Resolver at https://decomfy.com/resolve/ to detect missing nodes and generate install commands, or run the 1-click terminal setup script provided below.

How does Qwen-Image-2.1 Viggle Turbo achieve sub-6 second render times?

It uses DMD2 (Distribution Matching Distillation) 4-step distilled LoRA weights at CFG 1.0. This cuts sampling from 20+ steps down to 4 steps while preserving high structural consistency and avoiding over-saturation.