Qwen-Image-2.1 Viggle Turbo: 4-Step Ultra-Fast DMD2 Distillation

Full Resolution

Generation Specs RTX 4070

Sampler euler

Scheduler simple

Steps 4

CFG Scale 1

Seed 9283741029

Target VRAM 12GB

VRAM 8GB - 12GB

GPU Hardware Compatibility

Optimal Ground Truth

Native GPU execution at full throughput with zero memory swapping.

Qwen-Image-2.1 Viggle Turbo (DMD2 4-Step) Min 8GB VRAM 1024×1024 RTX 4070 Verified

Qwen-Image-2.1 Viggle Turbo: 4-Step Ultra-Fast DMD2 Distillation

Rapid concept exploration workflow powered by Viggle Turbo DMD2 distilled checkpoint. Generates photorealistic high-resolution images in just 4 sampling steps with 1.8-second generation feedback.

Blueprint Summary RTX 4070 Verified

Reproducible ComfyUI workflow for Qwen-Image-2.1 Viggle Turbo: 4-Step Ultra-Fast DMD2 Distillation using Qwen-Image-2.1 Viggle Turbo (DMD2 4-Step) at 1024×1024 resolution. Requires minimum 8GB VRAM with sampler euler and scheduler simple (4 steps). Includes 1-click terminal model sync and canvas JSON graph.

Node Graph Pipeline 7 Total Nodes

Verify in Resolver

01 Load Models

UnetLoaderGGUF

3 loaders (DiT, CLIP, VAE)

02 Conditioning

TextEncodeQwenImage21

1 prompt encodings

03 KSampler

KSampler

DiT latent denoising

04 Decode & Save

VAEDecode

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/qwen-image-2-1-viggle-turbo-comfyui.sh | bash

Positive Prompt

masterpiece, haute couture editorial photoshoot of an elegant 21-year-old Korean model wearing an emerald green silk satin evening slip gown with an ultra-deep plunging cowl back, standing gracefully in a high-rise luxury penthouse, panoramic twilight blue hour city skyline through floor-to-ceiling glass windows, warm interior tungsten accent lighting, soft rim light sculpting shoulders and spine, cinematic 8k lookbook photography, 85mm f/1.4

Negative Prompt

blurry, cartoon, 3d render, illustration, lowres, oversaturated, deformed fingers, extra limbs, bad anatomy

LoRA Adapter Stack 1 Adapters

RTX 4070 Calibrated Weights

01 Qwen-Image-2.1-viggle-turbo-4step-lora-r64.safetensors +1

Required Models 4 Models

Disk Space Required: 11.6 GB (4 models · DiT/Base: 4.8 GB · Text Encoder: 5.9 GB · LoRAs: 340 MB · VAE: 630 MB)

models/diffusion_models/ 4.8 GB HuggingFace

Qwen-Image-2.1-viggle-turbo-Q4_K_M-HQv3.gguf 4-step DMD2 distilled DiT quantized to GGUF Q4_K_M for ComfyUI

Get

models/text_encoders/ 5.88 GB HuggingFace

qwen3vl_8b_w4a8_heretic.safetensors Refusal-ablated multimodal text encoder (W4A8 quantized)

Get

models/vae/ 630 MB HuggingFace

qwen_image_2.1_vae_bf16.safetensors Official BF16 VAE for Qwen-Image

Get

models/loras/ 340 MB HuggingFace

Qwen-Image-2.1-viggle-turbo-4step-lora-r64.safetensors Weight: 1.0 (Official Viggle rank-64 4-step distillation LoRA adapter)

Get

Field Notes RTX 4070 Benchmark

Benchmark: RTX 4070 (12GB VRAM): ~1.9s per 1024x1024 frame (8.1 GB VRAM peak). RTX 3060: 3.2s.

DMD2 4-Step Distillation Architecture: Trained by Viggle using Distribution Matching Distillation 2 (DMD2) mapped directly from step-400 EMA teacher trajectories, collapsing 40 diffusion steps into 4 discrete leaps. Eliminates Classifier-Free Guidance doubling overhead (CFG must stay locked at 1.0; dialing CFG > 1.0 will cause harsh color posterization and blown-out contrast). Uses nullified scheduler terminal shift (shift_terminal=None) to ensure crisp 4th-step convergence without haze. Delivers ~10x generation speedup on consumer GPUs, making real-time prompt iteration instant.

Frequently Asked Questions FAQ

What GPU and VRAM are required to run Qwen-Image-2.1 Viggle Turbo: 4-Step Ultra-Fast DMD2 Distillation?

This workflow requires a minimum of 8GB VRAM (recommended 12GB 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.