Wan 2.2 14B - Text to Video
ComfyUI in the Cloud
Deploy Wan 2.2 14B - Text to Video, from
$0.60 / hour
Wan 2.2 Text-to-Video ComfyUI Workflow
A ComfyUI text-to-video generation workflow powered by Wan 2.2, featuring dual 14B diffusion models, FP8-scaled precision, UMT5 XXL text conditioning, Wan 2.1 VAE decoding, and optional LightX2V Lightning LoRA acceleration.
Workflow Overview
This workflow utilizes the Wan 2.2 text-to-video architecture for AI-powered video generation within ComfyUI. It combines separate high-noise and low-noise diffusion models, UMT5 XXL text conditioning, and the Wan 2.1 VAE for video latent decoding.
Unlike image-to-video workflows that use an existing image as visual conditioning, the text-to-video configuration is designed to generate video sequences from natural-language descriptions. Text prompts can describe subjects, environments, camera movements, visual styles, lighting, and motion characteristics.
Wan 2.2 uses a dual-stage denoising architecture with dedicated high-noise and low-noise models. The high-noise model handles the initial denoising stages, establishing the overall video structure and motion characteristics. The low-noise model handles the later stages, refining the latent representation and developing finer visual details.
Both diffusion models use FP8-scaled precision to help reduce memory requirements compared with higher-precision model weights. The workflow also includes separate LightX2V Lightning LoRA models for the high-noise and low-noise diffusion stages.
The displayed configuration uses a resolution of 640 × 640 pixels and a duration value of 5.0. Turbo Mode is disabled, meaning the optional Lightning acceleration configuration is not enabled in the displayed workflow.
Generation Settings
Width
640
Height
640
Video Duration
5.0
Resolution
640 × 640
Model Architecture
Dual 14B
Model Precision
FP8 Scaled
Turbo Mode
Disabled
Lightning LoRA
4-Step v1.1
Model Configuration
High-Noise Diffusion Model
14B / FP8wan2.2_t2v_high_noise_14B_fp8_scaled
The high-noise diffusion model handles the initial stages of the text-to-video generation process. During these stages, the latent representation contains relatively high levels of noise. The model performs early denoising operations, helping establish the overall visual composition, temporal structure, and motion characteristics. The selected model contains 14 billion parameters and uses an FP8-scaled configuration. FP8 precision helps reduce memory requirements compared with higher-precision model weights. This model forms the first major component of the Wan 2.2 dual-stage generation architecture.
Low-Noise Diffusion Model
14B / FP8wan2.2_t2v_low_noise_14B_fp8_scaled
The low-noise diffusion model handles the later stages of the text-to-video generation process. After the high-noise model establishes the initial latent representation, the low-noise model continues denoising as the noise level decreases. This stage focuses on refining the generated video representation. It contributes to finer visual details, scene consistency, and the final latent structure. The selected model uses a 14B architecture with FP8-scaled precision. Together, the high-noise and low-noise models form the dual-stage generation configuration used by Wan 2.2.
Text Conditioning Model
UMT5 XXLumt5_xxl_fp8_e4m3fn_scaled
The UMT5 XXL model provides text conditioning for the Wan 2.2 text-to-video generation process. It processes natural-language instructions and converts them into conditioning representations used by the diffusion models. These representations help guide video generation according to the supplied prompt. The model can interpret descriptions involving subjects, environments, visual appearance, and desired motion. The selected text encoder uses an FP8 E4M3FN-scaled configuration. This lower-precision representation helps reduce memory consumption compared with higher-precision text encoder implementations.
Video Variational Autoencoder (VAE)
Wan 2.1 VAEwan_2.1_vae
The workflow uses the Wan 2.1 VAE for processing video latent representations. During decoding, the VAE converts generated latent representations into visible video frames. This component provides the decoding stage required to transform the generated latent sequence into pixel-space output. The decoded frames can then be passed to compatible video preview, processing, or export nodes within ComfyUI.
Dual-Stage Diffusion Architecture
Wan 2.2 uses separate high-noise and low-noise diffusion models instead of relying on a single model throughout the entire denoising process. This allows different model components to handle different noise ranges during video generation.
Stage 1: High-Noise Processing
wan2.2_t2v_high_noise_14B_fp8_scaled
Handles the initial denoising stages when the latent representation contains higher noise levels. This stage helps establish the initial scene structure, visual composition, and motion characteristics of the generated video.
Stage 2: Low-Noise Processing
wan2.2_t2v_low_noise_14B_fp8_scaled
Handles the later denoising stages when the latent representation contains lower noise levels. This stage refines the generated representation and contributes to the final visual details and temporal consistency.
The transition between high-noise and low-noise processing is controlled by the internal sampling configuration. The exact switching point and sampling schedule are not visible in the screenshot. Both models are selected as part of the displayed Wan 2.2 text-to-video configuration.
LightX2V Lightning LoRA Configuration
The workflow includes separate LightX2V Lightning LoRA models for the high-noise and low-noise diffusion stages. These LoRAs are designed for accelerated text-to-video generation using a 4-step configuration.
Separate Lightning LoRAs are provided because Wan 2.2 uses distinct diffusion models for different noise ranges. Each LoRA is associated with its corresponding diffusion stage. The selected models use the LightX2V version 1.1 configuration.
High-Noise Lightning LoRA
4-Step / v1.1wan2.2_t2v_lightx2v_4steps_lora_v1.1_high_noise
This Lightning LoRA is designed for the high-noise diffusion model. It provides an acceleration configuration intended for the initial denoising stage. The selected model is identified as a 4-step Lightning variant, version 1.1. It can be applied to the high-noise model when the workflow's Turbo Mode is enabled.
Low-Noise Lightning LoRA
4-Step / v1.1wan2.2_t2v_lightx2v_4steps_lora_v1.1_low_noise
This Lightning LoRA is designed for the low-noise diffusion model. It provides an acceleration configuration intended for the later denoising stage. The selected model is identified as a 4-step Lightning variant, version 1.1. It can be applied to the low-noise model when the workflow's Turbo Mode is enabled.
Turbo Mode Configuration
The workflow includes an optional Turbo Mode for accelerated text-to-video generation. The selected Lightning LoRA models are designed for reduced-step inference using their respective high-noise and low-noise diffusion stages.
| Turbo Mode | Disabled |
| Lightning Variant | LightX2V |
| Lightning Step Designation | 4-Step |
| Lightning Version | v1.1 |
| High-Noise LoRA | Selected |
| Low-Noise LoRA | Selected |
Turbo Mode is disabled in the displayed workflow configuration. The Lightning LoRA models remain selected, but the optional acceleration mode is not enabled. The 4-step designation comes from the selected LoRA model names. The actual sampling step count is not visible in the screenshot and should not be assumed from the LoRA names alone.
Text-to-Video Generation Configuration
The Wan 2.2 text-to-video workflow generates video sequences from natural-language descriptions. Text conditioning guides the diffusion models in producing visual content and motion based on the supplied instructions.
Text Prompt Conditioning
umt5_xxl_fp8_e4m3fn_scaled
Natural-language instructions are processed through the UMT5 XXL text conditioning model. Prompts can describe visual subjects, environments, actions, camera movement, and desired cinematic characteristics. The encoded instructions guide the video generation process.
Video Generation
Dual-Stage Diffusion
The high-noise and low-noise diffusion models work across different denoising stages. The high-noise model establishes the initial latent structure. The low-noise model refines the representation before VAE decoding.
The text-to-video configuration does not require an image reference as its primary generation input. Instead, the generated sequence is guided by text conditioning and the internal generation parameters. Exact prompt nodes and conditioning connections are not visible in the supplied screenshot.
Video Generation Settings
The workflow includes dedicated controls for video width, height, and duration. These parameters determine the requested spatial and temporal characteristics of generated video.
Video Width
640 px
Controls the horizontal dimension of the generated video. The displayed width is set to 640 pixels.
Video Height
640 px
Controls the vertical dimension of the generated video. The displayed height is set to 640 pixels.
Video Duration
5.0
Defines the configured duration of the generated video. When interpreted in seconds by the internal workflow, this represents five seconds.
The selected dimensions produce a square 640 × 640 configuration with a 1:1 aspect ratio. The duration value is set to 5.0. The actual frame count and frame rate depend on the internal generation nodes and are not visible in the screenshot.
Additional Parameters
| Workflow Type | Text-to-Video |
| Width | 640 |
| Height | 640 |
| Resolution | 640 × 640 |
| Aspect Ratio | 1:1 |
| Duration | 5.0 |
| High-Noise Model | Wan 2.2 T2V 14B FP8 |
| Low-Noise Model | Wan 2.2 T2V 14B FP8 |
| Text Encoder | UMT5 XXL FP8 |
| VAE | Wan 2.1 VAE |
| Turbo Mode | Disabled |
| Lightning LoRA | LightX2V 4-Step v1.1 |
| Sampling Steps | Not visible in the screenshot |
| Noise Seed | Not visible in the screenshot |
Surrounding Nodes and Workflow Integration
The Wan 2.2 text-to-video subgraph consolidates high-noise and low-noise diffusion models, text conditioning, video decoding, generation dimensions, and optional Lightning acceleration settings.
In a typical text-to-video workflow, the generation process begins with a natural-language prompt. The prompt may describe a scene, subject movement, camera behavior, lighting conditions, and visual style. Compatible text-conditioning nodes process the supplied prompt before video generation.
The UMT5 XXL text encoder converts natural-language instructions into conditioning representations. These representations can guide the high-noise and low-noise diffusion models during the generation process.
The high-noise diffusion model handles the initial generation stage. It processes the latent representation during the higher-noise portion of the denoising schedule. The low-noise diffusion model then handles the later denoising stage, refining the generated video representation.
The workflow includes separate Lightning LoRA configurations for both diffusion stages. When Turbo Mode is enabled, compatible workflow logic can apply the high-noise Lightning LoRA to the high-noise model and the low-noise Lightning LoRA to the low-noise model. Turbo Mode is disabled in the displayed configuration.
The width and height parameters determine the configured spatial dimensions of the generated video. The duration parameter defines the requested sequence length. Noise initialization and sampling behavior depend on the internal workflow configuration.
After the diffusion process, the selected Wan 2.1 VAE decodes the generated latent representations into visible video frames. These frames can then be passed to compatible video combination, preview, or export nodes.
The screenshot shows the subgraph's model selections and generation parameters. Exact prompt connections, sampler settings, frame rate, noise seed, internal denoising schedule, and final output nodes are not visible in the supplied screenshot.
Workflow Highlights
- ✓ Wan 2.2 Text-to-Video: Uses the Wan 2.2 architecture for generating video sequences from text-based conditioning.
- ✓ Dual 14B Models: Includes separate high-noise and low-noise diffusion models.
- ✓ FP8 Quantization: Uses FP8-scaled diffusion model configurations to help reduce memory requirements.
- ✓ High-Noise Processing: Uses a dedicated diffusion model for the initial denoising stages.
- ✓ Low-Noise Refinement: Uses a separate diffusion model for the later denoising stages.
- ✓ UMT5 XXL Conditioning: Provides text conditioning using the FP8 E4M3FN-scaled encoder.
- ✓ Wan 2.1 VAE: Includes a dedicated VAE for decoding video latent representations into frames.
- ✓ Lightning LoRA Support: Includes separate LightX2V 4-step LoRA v1.1 models for high-noise and low-noise stages.
- ✓ Optional Turbo Mode: Provides an acceleration option using the selected Lightning LoRAs, currently disabled.
- ✓ 640 × 640 Resolution: Configured for square video generation with a 1:1 aspect ratio.
- ✓ Configurable Duration: Includes a duration parameter currently set to 5.0.
- ✓ Text-Guided Generation: Generates video sequences using natural-language conditioning without requiring an input image.
- ✓ Reusable Subgraph: Organizes dual-stage diffusion, model selection, Lightning LoRAs, and video parameters into a reusable ComfyUI component.
Workflow Summary
This Wan 2.2 text-to-video ComfyUI workflow combines two 14B FP8-scaled diffusion models for high-noise and low-noise denoising stages. It uses UMT5 XXL text conditioning and the Wan 2.1 VAE for processing and decoding video latent representations.
The workflow is configured with a resolution of 640 × 640 pixels and a duration value of 5.0. The dual-stage diffusion configuration separates initial denoising from later refinement operations.
The configuration also includes separate LightX2V Lightning LoRA models for the high-noise and low-noise stages. Both selected LoRAs are identified as 4-step version 1.1 variants. Turbo Mode is currently disabled, so the optional Lightning acceleration configuration is not enabled.
Unlike image-to-video workflows, this text-to-video configuration uses natural-language instructions to guide the generation of video content. The UMT5 XXL text encoder provides conditioning representations, while the two diffusion models perform the generation and refinement stages.
The subgraph organizes model selection, dual-stage diffusion, text conditioning, VAE decoding, generation dimensions, and optional acceleration settings. This configuration can be integrated into larger ComfyUI text-to-video workflows using compatible prompt, conditioning, sampling, and output nodes.