Have a ComfyUI workflow? Get a readable Workflow Doc in your browser.Make a workflow readable.Try it now ↗
Lightx2v is awesome, seriously awesome! 4-step LoRA is amazing. Now, using this workflow, we can get results in 8 steps that are as good as the 20-step workflow.
Runtime profile
Source description
4step:
Lightx2v is awesome, seriously awesome! 4-step LoRA is amazing. Now, using this workflow, we can get results in 8 steps that are as good as the 20-step workflow.
Adding shift nodes, It makes a huge difference to the image quality when using the Lighting LoRA.
8step:
We can shave off 60% of the generation time now.
AI-generated commentary
AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.
Use Qwen-image-workflow_gguf_Lowvram to create images from text prompts, including intricate text covers and memes; its description highlights complex text generation and prompt-following.
The described input is a text prompt; examples of generated images include text covers and memes.
Before setup, confirm which qwen_image_vae.safetensors file to use; only the filename is listed.
Suggestion · not verified
Named model files include ae.safetensors, qwen_2.5_vl_7b_fp8_scaled.safetensors, qwen_2.5_vl_fp16.safetensors, qwen_image_vae.safetensors, and qwen-image-Q4_K_S.gguf.
Listed node names include CFGZeroStar, CLIPLoader, CLIPTextEncode, EmptySD3LatentImage, KSampler, and ModelSamplingAuraFlow.
Other listed node names are Power Lora Loader (rgthree), SaveImage, UnetLoaderGGUF, UNetTemporalAttentionMultiply, VAEDecode, and VAELoader.
The description links Qwen-Image-Lightning-4steps-V1.0.safetensors and Qwen-Image-Lightning-8steps-V1.0.safetensors.
When using the Lighting LoRA, the author suggests adding shift nodes.
Suggestion · not verified
The Civitai listing does not state a usage license.
Check the listing's terms before using the workflow or sharing generated images.
Suggestion
Does this need to be edited?
Sign in to send an edit request.
Sources
1 sourceSource excerpts
1 excerptSource context: 1395 downloads · Type Workflows · Base model Qwen
V1:
This is the basic workflow for Qwen-Image.
Qwen-Image is good at complex text generation and prompt-following . You can even use it to create intricate text covers, memes, etc.
Official examples: https://qwenlm.github.io/blog/qwen-image/
Glad to see city96's fresh Qwen-Image-gguf checkpoint has arrived!
Checkpoint gguf https://huggingface.co/city96/Qwen-Image-gguf/tree/main "
Estimated VRAM requirement
Estimate unavailable
11.3 GB across 1 of 8 model files. Model file total + 25% loading overhead + 2 GB execution buffer, rounded up.
Requirements
17 requirementsVAE · Hugging Face · Comfy-Org/Lumina_Image_2.0_Repackaged · Lumina Image 2.0 Repackaged
CLIPLoader
PossibleText encoder · Unknown
qwen_2.5_vl_7b_fp8_scaled.safetensors
File unverifiedText encoder · Unknown
Text encoder · Hugging Face · Comfy-Org/Omnigen2_ComfyUI_repackaged
qwen_image_vae.safetensors
PossibleVAE · Unknown
Checkpoint · 11.3 GB · GGUF · Unknown
VAEDecode
PossibleVAE · Unknown
VAELoader
PossibleVAE · Unknown
CFGZeroStar
Not resolvedNode pack · Unknown
CLIPTextEncode
PossibleNode pack · Unknown
EmptySD3LatentImage
Not resolvedNode pack · Unknown
KSampler
Node pack · Unknown
Node pack · Unknown
Power Lora Loader (rgthree)
Not resolvedNode pack · Unknown
SaveImage
PossibleNode pack · Unknown
UnetLoaderGGUF
PossibleNode pack · Unknown
UNetTemporalAttentionMultiply
Not resolvedNode pack · Unknown
