⭐ Star Qwen Image Start(t) Settings

Qwen Image Star(t) Settings - All-in-one starter node for Qwen diffusion models

By Starnodes2024·Created 2 years ago·Updated 7 days ago· 103

◄Positive_Prompt—►

◄Negative_Prompt—►

◄Diffusion_ModelDefault►

◄VAEDefault►

◄CLIPDefault►

◄CLIP_Typeqwen_image►

◄CLIP_Devicecpu►

◄Latent_Ratio1:1 (1328x1328)►

◄Latent_Width1328►

◄Latent_Height1328►

◄Batch_Size1►

◄use_nearest_image_ratiofalse►

Category⭐StarNodes/Starters

Inputs (15)

Name	Type	Default	Description
Positive_Prompt	STRING		Positive prompt for image generation. If empty, generates a fluffy purple monster.
Negative_Prompt	STRING		Negative prompt for image generation. If empty, creates a zero-out condition.
Diffusion_Model	COMBO	Default	Select diffusion model from models/diffusion_models or models/unet folders
VAE	COMBO	Default	Select VAE model from models/vae folder
CLIP	COMBO	Default	Select CLIP model from models/clip folder
CLIP_Type	COMBO	qwen_image	CLIP model type for loading. For Qwen models, use 'qwen_image'
CLIP_Device	COMBO	cpu	Device to load CLIP model on
Latent_Ratio	COMBO	1:1 (1328x1328)	Predefined aspect ratios for Qwen Image models
Latent_Width	INT	132816–8192	Custom width when using Free Ratio
Latent_Height	INT	132816–8192	Custom height when using Free Ratio
Batch_Size	INT	11–4096	Number of images to generate in batch
use_nearest_image_ratio	BOOLEAN	false	When enabled with image input, automatically selects the closest aspect ratio
imageopt	IMAGE		Optional image input for automatic aspect ratio detection
model_overrideopt	MODEL		Optional model input. When connected, bypasses the Diffusion_Model selector and uses this model directly.
LoRA_Stackopt	LORA_STACK		Optional stack of LoRAs to apply to the model and internal conditioning.