1. Workflow Overview

This workflow, titled "Ultimate Landscape Plant Generator", specializes in creating high-fidelity plant images for garden design, virtual scenes, etc. Key features:

• Customizable Plants: Swap plant names in prompts (e.g., change Bellis perennis to Lavandula).

• Professional Rendering: Lumion-style lighting and textures.

• Auto Watermarking: Add text watermarks (e.g., "WUDAO DESIGN") for commercial use.

• HD Output: Default 960x1280 resolution (adjustable).

Core Models:

• Stable Diffusion XL: Base model (基础算法_XL).

• VAE: Built-in (enhances colors/details).

2. Node Details

Key Nodes:

1. CheckpointLoaderSimple: Loads base model + VAE.

   • Path: models/checkpoints/.

2. CLIPTextEncode:

   • Positive Prompt:

     Bellis perennis 'Ruby Star', plant, in a garden... rendered in lumion  

   • Negative Prompt: Filters artifacts/watermarks.

3. KSampler:

   • Sampler: dpmpp_3m_sde (30 steps, CFG=7).

4. CR Simple Text Watermark:

   • Adds customizable watermarks (text/position/font).

   • Install: Via ComfyUI Manager (search "CR Watermark").

5. PreviewImage: Real-time preview.

3. Workflow Structure

1. Model & Conditioning:

• Load 基础算法_XL and encode prompts.

2. Image Generation:

• Starts from empty latent (960x1280).

• KSampler generates plant landscape.

3. Post-Processing:

• VAEDecode: Converts latent to image.

• Watermarking: Two-layer text overlay.

4. Output:

• Saves to ComfyUI/output/.

4. Inputs & Outputs

• Inputs:

  • Resolution: Adjust in EmptyLatentImage.

  • Plant Name: Edit prompt (e.g., Lavandula for lavender).

  • Watermark: Customize text/font/position.

• Output:

  • HD plant image with watermark (PNG).

5. Notes

1. Dependencies:

   • Place 基础算法_XL.safetensors in models/checkpoints/.

2. Fonts:

   • Default uses Quicksand-Bold.ttf (install or replace).

3. Optimization:

   • ≥8GB VRAM for high-res (e.g., 1920x2560).

   • Reduce KSampler steps (e.g., 20) for speed.

4. Extensions:

   • Add LoRA nodes for more styles (e.g., cartoon/realistic).