BXR - I2V Upscaler & Interpolation - WAN 2.2 (V2 Refresh) 👑

Thanks to everyone who has supported the page recently and helped hit the 1k milestone! To give back, I’m releasing BXR I2V ULTRA v2.0.

This is my dedicated ComfyUI workflow for taking Image-to-Video generations to the absolute maximum quality. It’s designed to be powerful right out of the gate, but modular enough for you to customize it to your own style.

Key Features:

• The "Get Started" Node: I've built in a node that will directly download everything you need to run this on my personal standards. No hunting for the baseline models.

• Total Flexibility: Want to bring your own models, LoRAs, or upscalers to the table? The workflow is cleanly sectioned so you can easily swap in your preferred assets.

• Built-in Enhancement: Integrated RIFE frame interpolation and high-fidelity upscaling (configured for Remacri) to ensure your final outputs are buttery smooth and crisp.

How to Install & Use:

1. Download the .json file attached to this post.

2. Open ComfyUI.

3. Simply drag and drop the .json file directly onto your ComfyUI workspace to load the workflow.

4. Run the "Get Started" node to pull the required models, or hook up your own.

⚠️ IMPORTANT WARNING: ComfyUI V2 Desktop / Node 2.0 This workflow is NOT built on Node 2.0 yet. I have not tested it within the new V2 mode, so please use it with caution or stick to the standard ComfyUI interface to ensure stability.

Enjoy the workflow, and drop your best generations in the reviews!

Dependencies & Models

Custom Node Suites Identified:

• WanImageToVideo: Required for the core Wan2.1 I2V generation (WanImageToVideo node).

• ComfyUI-Easy-Use: Required for the VRAM management (easy cleanGpuUsed node).

• ComfyUI-Frame-Interpolation (or similar FL RIFE suite): Required for the frame smoothing (FL_RIFE node).

• ComfyUI-VideoHelperSuite: Required for processing and exporting the final media (CreateVideo and SaveVideo nodes).

Models, LoRAs, and Checkpoints Referenced:

These are included in the "Get Started" Node.

• Diffusion Models: DasiwaWAN22I2V14BLightspeed_snatchkissHighV11.safetensors and DasiwaWAN22I2V14BLightspeed_snatchkissLowV11.safetensors

• LoRAs: DR34ML4Y_I2V_14B_HIGH_V2.safetensors and DR34ML4Y_I2V_14B_LOW_V2.safetensors

• CLIP/Text Encoder: umt5_xxl_fp8_e4m3fn_scaled.safetensors

• VAE: wan_2.1_vae.safetensors

• Upscale Model: 4x_foolhardy_Remacri.pth

• Interpolation Model: rife49