WAN 2.1 FLF2V: Simple ComfyUI Workflow for First-Last Frame Video Generation (Low VRAM Friendly, Laptop) — v2.0-FirstFrameLastFrame

Use this ComfyUI workflow to generate first-last-frame video from a first-frame image and a last-frame image; it produces generated video and upscaled images.

It loads and resizes images, generates first-last-frame video, upscales images, applies RIFE VFI, and combines video frames.

Inputs are a first-frame image and a last-frame image.

Outputs include generated video and upscaled images; video frames are combined after generation and frame interpolation.

The page says it is outdated and points to the Ultimate Workflow: https://civitai.com/models/1309674?modelVersionId=1496495.

Listed model files include clip_vision_h_fp8_e4m3fn.safetensors, rife49.pth, and RealESRGAN_x4.pth.

Additional listed model files are t5xxl_um_fp8_e4m3fn_scaled.safetensors, wan_2.1_vae.safetensors, and wan2.1-flf2v-720p-14b-q2_k.gguf.

Listed node packs are comfyui-frame-interpolation, comfyui-kjnodes, comfyui-videohelpersuite, gguf, rgthree-comfy, and tiled_ksampler.

It uses the Wan Video 14B i2v 720p base model and version v2.0-FirstFrameLastFrame.

Review the linked Ultimate Workflow before use.

Suggestion · non vérifié

Before use, verify the intended files for clip_vision_h_fp8_e4m3fn.safetensors, RealESRGAN_x4.pth, wan2.1-flf2v-720p-14b-q2_k.gguf, and t5xxl_um_fp8_e4m3fn_scaled.safetensors.

Suggestion · non vérifié