Basic Workflow — v1.0

Basic Workflow is for creating SD 1.5 images from text input, a seed, and selected checkpoint, LoRA, and VAE components, with decoded and saved image outputs.

Sugestão · não verificado

The named stages include text encoding, sampling with KSampler, VAE decoding, and image saving.

Sugestão · não verificado

Use CLIPTextEncode or ttN text for text input, choose a Seed, and select checkpoint, LoRA, and VAE components.

Sugestão · não verificado

The workflow names VAEDecode and SaveImage as stages for decoded and saved images.

Find Basic Workflow on Civitai; it is listed as a Workflows item based on SD 1.5 and published by Alastor 666 1933.

Listed model files are LCM_LoRA_Weights_SD15.safetensors, realisticVisionV60B1_v51HyperVAE.safetensors, and vae-ft-mse-840000-ema-pruned.safetensors.

Named node items include Anything Everywhere, caching_condition, CheckpointLoaderSimple, CLIPTextEncode, EmptyLatentImage, GetNode, KSampler, and Lora Loader Stack (rgthree).

Other named node items include LoraTagLoader, SaveImage, Seed, SetNode, ttN text, VAEDecode, and VAELoader. The workflow also names Caching to not Waste, LoRA Tag Loader for ComfyUI, cg-use-everywhere, rgthree-comfy, tinyterraNodes, and WAS Node Suite.

For shared LoRAs, set them in Lora Loader Stack (rgthree) to load them once during multi-generation runs.

Sugestão · não verificado

The workflow describes Caching to not Waste as caching many parts so future generations can skip many steps and avoid repeated model loading.

The workflow specifies an 8–12 inference-step range for the LCM LoRA.

Before using it, verify that the listed model files and node items are available in your ComfyUI setup.

Sugestão · não verificado

Example images are AI-generated and do not represent real people.