HiDream-I1 | T2I

HiDream-I1 | Text-to-Image AI Generation

HiDream-I1 is a cutting-edge open-source text-to-image foundation model with 17 billion parameters, designed for ultra-fast, high-quality image generation. This ComfyUI-based workflow enables users to convert natural language prompts into stunning visuals—ideal for concept art, storytelling, ideation, or professional-grade image synthesis with HiDream-I1.

Whether you're illustrating a sci-fi world, sketching product ideas, or imagining surreal environments, HiDream-I1 delivers remarkable fidelity, speed, and ease of use.

Why Use HiDream-I1 Text-to-Image?

The HiDream-I1 workflow provides an intuitive and expressive platform to generate high-quality images from text alone:

• Turn text prompts into images in seconds with minimal setup using HiDream-I1
• Backed by a 17B parameter model for detailed, coherent visual generation with HiDream-I1
• Supports high resolution output and customizable sampling options in the HiDream-I1 workflow
• Great for concept design, digital art, visual storytelling, and ideation with HiDream-I1
• Runs seamlessly inside ComfyUI with easy-to-modify HiDream-I1 workflow components

From single-sentence prompts to layered descriptive passages, HiDream-I1 captures your intent and renders it with clarity and style.

How to Use the HiDream-I1 Text-to-Image Workflow

HiDream-I1 Text-to-Image Overview

• Load HiDream-I1 Model : Load the official HiDream-I1 17B model
• Enter Prompt Text : Input your text prompt for HiDream-I1 (positive and optional negative prompts)
• Set Image Parameters : Adjust resolution, sampling method, and steps for your HiDream-I1 generation
• Generate Image : The workflow will render your result using HiDream-I1's fast inference
• Save Image : Output image is automatically saved and viewable in the save node

Quick Start Steps with HiDream-I1:

1. Load the HiDream-I1 model using the provided model loader node
2. Enter a positive prompt describing what you want to see with HiDream-I1
3. Optionally add a negative prompt to filter out unwanted features in your HiDream-I1 generation
4. Set the desired image resolution and sampling configuration for optimal HiDream-I1 results
5. Click the Queue Prompt button in ComfyUI to generate with HiDream-I1
6. View and download your HiDream-I1 generated image from the Save Image node

Example Prompts for HiDream-I1

• "A futuristic cityscape at sunset, neon lights, flying cars, high detail, cinematic composition"
• "A serene forest in autumn, golden leaves, misty atmosphere, photorealistic"
• "A wizard casting a spell in a stormy mountain landscape, dramatic lighting, fantasy art style"
• "Cyberpunk samurai walking through a rainy alley, glowing signage, cinematic angle"
• "Elegant product mockup of a smartwatch on a marble pedestal, studio lighting, minimal background"

1 - Load HiDream-I1 Model

Choose the right HiDream-I1 model variant for your task:

• HiDream-I1-Full: Full version model for more detailed inference and longer sequences
• HiDream-I1-Dev: Development version, optimized for slightly faster performance with fewer steps
• HiDream-I1-Fast: Optimized for quick results with the fastest generation speed

2 - Enter Prompts for HiDream-I1

In the Text-to-Image group, enter your positive and negative prompts to guide image generation with HiDream-I1.

• Positive Prompt for HiDream-I1:
  • Describe the scene, subject, style, and mood you want HiDream-I1 to generate.
  • Use detailed, creative, and descriptive language to leverage the full capability of the HiDream-I1 17B parameter model.
  • Example for HiDream-I1:

    A futuristic city skyline at dusk, glowing neon lights, cinematic lighting, highly detailed, ultra realistic, photo style
    

• Negative Prompt for HiDream-I1:
  • Specify what you want to exclude from the HiDream-I1 generated image.
  • Helps improve clarity and avoid artifacts or unwanted features in your HiDream-I1 creation.
  • Example:

    blurry, extra limbs, grainy, distorted faces, watermark, text, low quality, poorly drawn
    

💡 Tip: The more specific and vivid your positive prompt, the better the HiDream-I1 output. Use negative prompts to avoid common generative flaws in HiDream-I1 images.

3 - HiDream-I1 Sampler Settings

• HiDream-I1 Full
  • hidream_i1_full_fp16.safetensors
  • shift: 3.0
  • steps: 50
  • sampler: uni_pc
  • scheduler: simple
  • cfg: 5.0
• HiDream-I1 Dev
  • hidream_i1_dev_bf16.safetensors
  • shift: 6.0
  • steps: 28
  • sampler: lcm
  • scheduler: normal
  • cfg: 1.0 (no negative prompt)
• HiDream-I1 Fast
  • hidream_i1_fast_bf16.safetensors
  • shift: 3.0
  • steps: 16
  • sampler: lcm
  • scheduler: normal
  • cfg: 1.0 (no negative prompt)

Acknowledgement

The HiDream-I1 Text-to-Image workflow is made possible through the powerful and modular architecture of ComfyUI. We extend our deep appreciation to the ComfyUI development team and open-source contributors for building a flexible, node-based framework that allows seamless integration of advanced generative models like HiDream-I1.

Their dedication to accessible and extensible tooling empowers creators and researchers to explore state-of-the-art image generation with HiDream-I1 with minimal effort and maximum control.