EasyAnimate Sampler for Video to Video

EasyAnimateV2VSampler is video-to-video for the V3/V4 EasyAnimate weights - and it's the most interesting of the three legacy samplers, because it's the one that also handles control inputs. Feed it an existing video and it re-renders it with your prompt; feed it a control video (Canny, pose, depth) and it does controlled generation. On a V5.1 stack you'd use EasyAnimateV5_V2VSampler instead, which adds camera and trajectory support; this one is the older, simpler version.

The inputs that matter

• easyanimate_model - from the loader. The twist: this node works with both Inpaint and Control model types. With Inpaint weights you pass a validation_video; with Control weights you pass a control_video. The node branches internally depending on which pipeline the loader built.
• validation_video (optional) - the video you want to restyle, for Inpaint models. The node samples its resolution and uses it as the source.
• control_video (optional) - for Control models: the conditioning video, e.g. a pose or Canny sequence extracted upstream.
• video_length - 8–144, default 72. Frames are sampled from the source at 8fps.
• base_resolution - 512/768/960/1024, default 768; aspect ratio is derived from the input video like the I2V sampler does.
• denoise_strength - the video-native parameter, 0.05 to 1.0, default 0.7. How much of the source survives. Low = close to the original, high = the model basically redraws it. This is the dial you'll actually tune, because it's the difference between "recolored" and "reimagined."
• seed, steps, cfg, scheduler - the usual, with the five legacy schedulers (no Flow).

The output

images - an IMAGE batch of generated frames. Assemble with VideoHelperSuite.

How it fits

LoadEasyAnimateModel (Inpaint) ──► EasyAnimateV2VSampler (validation_video)

or, for control work:

LoadEasyAnimateModel (Control) ──► EasyAnimateV2VSampler (control_video)

Both paths resolve the output resolution to match the input video (rounded to /16), so you don't think about width/height here - the source defines the canvas.

Notes

denoise_strength at 1.0 is effectively not V2V at all - you've discarded the structure you were trying to preserve. Start around 0.5–0.7 and move down, not up. The control path is genuinely useful for pose-driven animation, but it inherits the same caveat as all the EasyAnimate control work: it's a control signal, not a guarantee, and the community's verdict on EasyAnimate motion was always "conservative." If you're on V5.1, the V5 version of this node gives you the same idea plus camera and trajectory wiring - worth skipping ahead for.