My personal basic HiDream Full + Flux.Dev refiner workflow — v1.0

This HiDream Full + Flux.Dev refiner workflow is for turning a text prompt and an empty latent image into a saved image through text encoding, Flux guidance, sampling, VAE decoding, and image saving.

Its named stages include model loading, text-encoder loading, empty-latent setup, Flux guidance, sampling, VAE decoding, and image saving.

CLIPTextEncode is the named stage for text input, and EmptySD3LatentImage is the named stage for an empty latent image.

VAEDecode and SaveImage are the named stages for decoding and saving the image output.

It is listed as a Workflows entry based on HiDream.

DigitalZombieAI is listed as the publisher, and the source page is https://civitai.com/models/1487982.

The listed model files are ae.safetensors, clip_g_hidream.safetensors, clip_l_hidream.safetensors, clip_l.safetensors, and flux1-dev.safetensors.

The list also includes hidream_i1_full_fp16.safetensors, llama_3.1_8b_instruct_fp8_scaled.safetensors, t5xxl_fp16.safetensors, and t5xxl_fp8_e4m3fn_scaled.safetensors.

The named components include CLIPTextEncode, DualCLIPLoader, EmptySD3LatentImage, FluxGuidance, KSampler, and ModelSamplingSD3.

They also include QuadrupleCLIPLoader, SaveImage, UNETLoader, VAEDecode, VAEEncode, and VAELoader.

Before using it, verify that each listed model file is available and that its filename matches exactly.

Suggestion · non vérifié

Also verify that your setup includes the named components, including CLIPTextEncode, KSampler, and VAELoader.

Suggestion · non vérifié