Skip to main content
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
The ModelMergeQwenImage node merges two AI models by combining their components with adjustable weights. It allows you to blend specific parts of Qwen image models, including transformer blocks, positional embeddings, and text processing components. You can control how much influence each model has on different sections of the merged result.

Inputs

ParameterData TypeRequiredRangeDescription
model1MODELYes-The first model to merge (default: none)
model2MODELYes-The second model to merge (default: none)
pos_embeds.FLOATYes0.0 to 1.0Weight for positional embeddings blending (default: 1.0)
img_in.FLOATYes0.0 to 1.0Weight for image input processing blending (default: 1.0)
txt_norm.FLOATYes0.0 to 1.0Weight for text normalization blending (default: 1.0)
txt_in.FLOATYes0.0 to 1.0Weight for text input processing blending (default: 1.0)
time_text_embed.FLOATYes0.0 to 1.0Weight for time and text embedding blending (default: 1.0)
transformer_blocks.0. to transformer_blocks.59.FLOATYes0.0 to 1.0Weight for each transformer block blending (default: 1.0)
proj_out.FLOATYes0.0 to 1.0Weight for output projection blending (default: 1.0)

Outputs

Output NameData TypeDescription
modelMODELThe merged model combining components from both input models with the specified weights