Skip to main content
The ModelMergeWAN2_1 node merges two WAN2.1 models by blending their components using weighted averages. It supports different model sizes, including 1.3B models with 30 blocks and 14B models with 40 blocks, with special handling for image-to-video models that include an extra image embedding component. Each component can be individually weighted to control the blending ratio between the two input models.

​
Inputs

Note: All weight parameters use a range from 0.0 to 1.0 with 0.01 step increments. The node provides up to 40 block weight inputs to accommodate different model sizes: 1.3B models use 30 blocks (blocks.0. through blocks.29.), while 14B models use 40 blocks (blocks.0. through blocks.39.). The img_emb. parameter is used by image-to-video models.

​
Outputs

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
Source fingerprint (SHA-256): 6a17defa25b1ef045b85af4a73e00d3a64c1948c0c47f355d1d488a75b09f224