1. Workflow Overview

This workflow generates high-quality children portraits with:

• Natural lighting (e.g., autumn sunlight, backlighting).

• Custom "Time Tale" LoRA (weight=0.8) for childish aesthetics.

• Baidu Translate integration (Chinese→English prompts).

• Dual-mode (online/local) support.

2. Core Models

• Main Model: 高级质感人像-FLUX_v1.0 (optimized for skin texture).

• LoRA: 时光物语丨儿童写真_D1.04 (download link).

• VAE: ae.sft (color accuracy).

3. Node Details

Key Nodes

• LoraLoader: Applies children portrait LoRA.

• BaiduTranslateNode: Translates Chinese prompts.

• KSampler: Euler sampler (30 steps) at 768x1248 resolution.

Installation

• Required: Install pysssss plugin via ComfyUI Manager.

• Baidu Translate: API key setup required.

4. Workflow Structure

1. Input Group: BaiduTranslateNode, ttN text.

2. Model Loading: CheckpointLoaderSimple, LoraLoader.

3. Generation: KSampler + CLIPTextEncode.

4. Output: VAEDecode + PreviewImage.

5. Inputs & Outputs

• Inputs: Prompt (e.g., "A girl smiling with a cat in autumn light").

• Output: Preview image (PNG).

6. Notes

1. Errors:

   • Missing LoRA files break style rendering.

2. Optimization:

   • Use SamplerCustomAdvanced for finer control in local mode.

3. Compatibility: Requires ComfyUI ≥0.3.15 and NVIDIA GPU.