ModelMergeSD3_2B - ComfyUI Built-in Node Documentation

Complete documentation for the ModelMergeSD3_2B node in ComfyUI. Learn its inputs, outputs, parameters and usage.

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

The ModelMergeSD3_2B node allows you to merge two Stable Diffusion 3 2B models by blending their components with adjustable weights. It provides individual control over embedding layers and transformer blocks, enabling fine-tuned model combinations for specialized generation tasks.

Inputs

Parameter	Data Type	Required	Range	Description
model1	MODEL	Yes	-	The first model to merge
model2	MODEL	Yes	-	The second model to merge
pos_embed.	FLOAT	Yes	0.0 - 1.0	Position embedding interpolation weight (default: 1.0)
x_embedder.	FLOAT	Yes	0.0 - 1.0	Input embedding interpolation weight (default: 1.0)
context_embedder.	FLOAT	Yes	0.0 - 1.0	Context embedding interpolation weight (default: 1.0)
y_embedder.	FLOAT	Yes	0.0 - 1.0	Y embedding interpolation weight (default: 1.0)
t_embedder.	FLOAT	Yes	0.0 - 1.0	Time embedding interpolation weight (default: 1.0)
joint_blocks.0.	FLOAT	Yes	0.0 - 1.0	Joint block 0 interpolation weight (default: 1.0)
joint_blocks.1.	FLOAT	Yes	0.0 - 1.0	Joint block 1 interpolation weight (default: 1.0)
joint_blocks.2.	FLOAT	Yes	0.0 - 1.0	Joint block 2 interpolation weight (default: 1.0)
joint_blocks.3.	FLOAT	Yes	0.0 - 1.0	Joint block 3 interpolation weight (default: 1.0)
joint_blocks.4.	FLOAT	Yes	0.0 - 1.0	Joint block 4 interpolation weight (default: 1.0)
joint_blocks.5.	FLOAT	Yes	0.0 - 1.0	Joint block 5 interpolation weight (default: 1.0)
joint_blocks.6.	FLOAT	Yes	0.0 - 1.0	Joint block 6 interpolation weight (default: 1.0)
joint_blocks.7.	FLOAT	Yes	0.0 - 1.0	Joint block 7 interpolation weight (default: 1.0)
joint_blocks.8.	FLOAT	Yes	0.0 - 1.0	Joint block 8 interpolation weight (default: 1.0)
joint_blocks.9.	FLOAT	Yes	0.0 - 1.0	Joint block 9 interpolation weight (default: 1.0)
joint_blocks.10.	FLOAT	Yes	0.0 - 1.0	Joint block 10 interpolation weight (default: 1.0)
joint_blocks.11.	FLOAT	Yes	0.0 - 1.0	Joint block 11 interpolation weight (default: 1.0)
joint_blocks.12.	FLOAT	Yes	0.0 - 1.0	Joint block 12 interpolation weight (default: 1.0)
joint_blocks.13.	FLOAT	Yes	0.0 - 1.0	Joint block 13 interpolation weight (default: 1.0)
joint_blocks.14.	FLOAT	Yes	0.0 - 1.0	Joint block 14 interpolation weight (default: 1.0)
joint_blocks.15.	FLOAT	Yes	0.0 - 1.0	Joint block 15 interpolation weight (default: 1.0)
joint_blocks.16.	FLOAT	Yes	0.0 - 1.0	Joint block 16 interpolation weight (default: 1.0)
joint_blocks.17.	FLOAT	Yes	0.0 - 1.0	Joint block 17 interpolation weight (default: 1.0)
joint_blocks.18.	FLOAT	Yes	0.0 - 1.0	Joint block 18 interpolation weight (default: 1.0)
joint_blocks.19.	FLOAT	Yes	0.0 - 1.0	Joint block 19 interpolation weight (default: 1.0)
joint_blocks.20.	FLOAT	Yes	0.0 - 1.0	Joint block 20 interpolation weight (default: 1.0)
joint_blocks.21.	FLOAT	Yes	0.0 - 1.0	Joint block 21 interpolation weight (default: 1.0)
joint_blocks.22.	FLOAT	Yes	0.0 - 1.0	Joint block 22 interpolation weight (default: 1.0)
joint_blocks.23.	FLOAT	Yes	0.0 - 1.0	Joint block 23 interpolation weight (default: 1.0)
final_layer.	FLOAT	Yes	0.0 - 1.0	Final layer interpolation weight (default: 1.0)

Outputs

Output Name	Data Type	Description
model	MODEL	The merged model combining features from both input models