comfyui-workflow-builder

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

ComfyUI Workflow Builder

ComfyUI 工作流构建器

Translates natural language requests into executable ComfyUI workflow JSON. Always validates against inventory before generating.

将自然语言请求转换为可执行的ComfyUI工作流JSON，生成前始终会基于资产清单进行校验。

Workflow Generation Process

工作流生成流程

Step 1: Understand the Request

步骤1：理解请求

Parse the user's intent into:

Output type: Image, video, or audio
Source material: Text-only, reference image(s), existing video
Identity method: None, zero-shot (InstantID/PuLID), LoRA, Kontext
Quality level: Draft (fast iteration) vs production (maximum quality)
Special requirements: ControlNet, inpainting, upscaling, lip-sync

将用户意图解析为以下维度：

输出类型：图像、视频或音频
源素材：仅文本、参考图像、现有视频
身份保留方法：无、零样本（InstantID/PuLID）、LoRA、Kontext
质量等级：草稿（快速迭代）vs 生产级（最高质量）
特殊需求：ControlNet、局部重绘、超分辨率、唇形同步

Step 2: Check Inventory

步骤2：检查资产清单

Read

state/inventory.json

to determine:

Available checkpoints → select best match for task
Available identity models → determine which methods are possible
Available ControlNet models → enable pose/depth control if available
Custom nodes installed → verify all required nodes exist
VRAM available → optimize settings accordingly

读取

state/inventory.json

确认以下信息：

可用检查点模型 → 选择最匹配任务的选项
可用身份模型 → 确定可使用的实现方法
可用ControlNet模型 → 存在则启用姿态/深度控制能力
已安装自定义节点 → 验证所有所需节点均存在
可用显存 → 对应优化参数设置

Step 3: Select Pipeline Pattern

步骤3：选择流水线模式

Based on request + inventory, choose from:

Pattern	When	Key Nodes
Text-to-Image	Simple generation	Checkpoint → CLIP → KSampler → VAE
Identity-Preserved Image	Character consistency	+ InstantID/PuLID/IP-Adapter
LoRA Character	Trained character	+ LoRA Loader
Image-to-Video (Wan)	High-quality video	Diffusion Model → Wan I2V → Video Combine
Image-to-Video (AnimateDiff)	Fast video, motion control	+ AnimateDiff Loader + Motion LoRAs
Talking Head	Character speaks	Image → Video → Voice → Lip-Sync
Upscale	Enhance resolution	Image → UltimateSDUpscale → Save
Inpainting	Edit regions	Image + Mask → Inpaint Model → KSampler

基于请求内容和资产清单，从以下模式中选择：

模式	适用场景	核心节点
文生图	简单生成场景	Checkpoint → CLIP → KSampler → VAE
身份保留图像生成	角色一致性要求场景	+ InstantID/PuLID/IP-Adapter
LoRA角色生成	训练过的自定义角色生成	+ LoRA Loader
图生视频（Wan）	高质量视频生成	Diffusion Model → Wan I2V → Video Combine
图生视频（AnimateDiff）	快速视频生成、运动控制场景	+ AnimateDiff Loader + Motion LoRAs
数字人生成	角色说话场景	Image → Video → Voice → Lip-Sync
超分辨率	提升分辨率场景	Image → UltimateSDUpscale → Save
局部重绘	编辑图像指定区域	Image + Mask → Inpaint Model → KSampler

Step 4: Generate Workflow JSON

步骤4：生成工作流JSON

ComfyUI workflow format:

json

{
  "{node_id}": {
    "class_type": "{NodeClassName}",
    "inputs": {
      "{param_name}": "{value}",
      "{connected_param}": ["{source_node_id}", {output_index}]
    }
  }
}

Rules:

Node IDs are strings (typically "1", "2", "3"...)
Connected inputs use array format:
```
["source_node_id", output_index]
```
Output index is 0-based integer
Filenames must match exactly what's in inventory
Seed values: use random large integer or fixed for reproducibility

ComfyUI工作流格式：

json

{
  "{node_id}": {
    "class_type": "{NodeClassName}",
    "inputs": {
      "{param_name}": "{value}",
      "{connected_param}": ["{source_node_id}", {output_index}]
    }
  }
}

规则：

节点ID为字符串（通常为 "1", "2", "3"...）
连接型输入使用数组格式：
```
["source_node_id", output_index]
```
输出索引为从0开始的整数
文件名必须与资产清单中的内容完全匹配
种子值：使用随机大整数或固定值保证可复现性

Step 5: Validate

步骤5：校验

Before presenting to user:

Every
```
class_type
```
exists in inventory's node list
Every model filename exists in inventory's model list
All required connections are present (no dangling inputs)
VRAM estimate doesn't exceed available VRAM
Resolution is compatible with chosen model (512 for SD1.5, 1024 for SDXL/FLUX)

提交给用户前需完成以下校验：

所有
```
class_type
```
均存在于资产清单的节点列表中
所有模型文件名均存在于资产清单的模型列表中
所有所需连接均已配置（无悬空输入）
显存估算值不超过可用显存
分辨率与所选模型兼容（SD1.5为512，SDXL/FLUX为1024）

Step 6: Output

步骤6：输出

If online mode: Queue via

comfyui-api

skill If offline mode: Save JSON to

projects/{project}/workflows/

with descriptive name

在线模式：通过

comfyui-api

skill加入执行队列 离线模式：将JSON保存到

projects/{project}/workflows/

路径下，使用描述性文件名

Workflow Templates

工作流模板

Basic Text-to-Image (FLUX)

基础文生图（FLUX）

json

{
  "1": {
    "class_type": "LoadCheckpoint",
    "inputs": {"ckpt_name": "flux1-dev.safetensors"}
  },
  "2": {
    "class_type": "CLIPTextEncode",
    "inputs": {"text": "{positive_prompt}", "clip": ["1", 1]}
  },
  "3": {
    "class_type": "CLIPTextEncode",
    "inputs": {"text": "{negative_prompt}", "clip": ["1", 1]}
  },
  "4": {
    "class_type": "EmptyLatentImage",
    "inputs": {"width": 1024, "height": 1024, "batch_size": 1}
  },
  "5": {
    "class_type": "KSampler",
    "inputs": {
      "seed": 42,
      "steps": 25,
      "cfg": 3.5,
      "sampler_name": "euler",
      "scheduler": "normal",
      "denoise": 1.0,
      "model": ["1", 0],
      "positive": ["2", 0],
      "negative": ["3", 0],
      "latent_image": ["4", 0]
    }
  },
  "6": {
    "class_type": "VAEDecode",
    "inputs": {"samples": ["5", 0], "vae": ["1", 2]}
  },
  "7": {
    "class_type": "SaveImage",
    "inputs": {"filename_prefix": "output", "images": ["6", 0]}
  }
}

json

{
  "1": {
    "class_type": "LoadCheckpoint",
    "inputs": {"ckpt_name": "flux1-dev.safetensors"}
  },
  "2": {
    "class_type": "CLIPTextEncode",
    "inputs": {"text": "{positive_prompt}", "clip": ["1", 1]}
  },
  "3": {
    "class_type": "CLIPTextEncode",
    "inputs": {"text": "{negative_prompt}", "clip": ["1", 1]}
  },
  "4": {
    "class_type": "EmptyLatentImage",
    "inputs": {"width": 1024, "height": 1024, "batch_size": 1}
  },
  "5": {
    "class_type": "KSampler",
    "inputs": {
      "seed": 42,
      "steps": 25,
      "cfg": 3.5,
      "sampler_name": "euler",
      "scheduler": "normal",
      "denoise": 1.0,
      "model": ["1", 0],
      "positive": ["2", 0],
      "negative": ["3", 0],
      "latent_image": ["4", 0]
    }
  },
  "6": {
    "class_type": "VAEDecode",
    "inputs": {"samples": ["5", 0], "vae": ["1", 2]}
  },
  "7": {
    "class_type": "SaveImage",
    "inputs": {"filename_prefix": "output", "images": ["6", 0]}
  }
}

With Identity Preservation (InstantID + IP-Adapter)

带身份保留（InstantID + IP-Adapter）

Extends basic template by adding:

Load reference image node
InstantID Model Loader + Apply InstantID
IPAdapter Unified Loader + Apply IPAdapter
FaceDetailer post-processing

See

references/workflows.md

for complete node settings.

在基础模板上扩展添加以下内容：

加载参考图像节点
InstantID Model Loader + Apply InstantID
IPAdapter Unified Loader + Apply IPAdapter
FaceDetailer 后处理

完整节点设置请参考

references/workflows.md

。

Video Generation (Wan I2V)

视频生成（Wan I2V）

Uses different loader chain:

Load Diffusion Model (not LoadCheckpoint)
Wan I2V Conditioning
EmptySD3LatentImage (with frame count)
Video Combine (VHS)

See

references/workflows.md

Workflow 4 for complete settings.

使用不同的加载链路：

加载扩散模型（而非LoadCheckpoint）
Wan I2V 条件处理
EmptySD3LatentImage（支持帧数配置）
视频合并（VHS）

完整设置请参考

references/workflows.md

的工作流4。

VRAM Estimation

显存估算

Component	Approximate VRAM
FLUX FP16	16GB
FLUX FP8	8GB
SDXL	6GB
SD1.5	4GB
InstantID	+4GB
IP-Adapter	+2GB
ControlNet (each)	+1.5GB
Wan 14B	20GB
Wan 1.3B	5GB
AnimateDiff	+3GB
FaceDetailer	+2GB

组件	预估显存占用
FLUX FP16	16GB
FLUX FP8	8GB
SDXL	6GB
SD1.5	4GB
InstantID	+4GB
IP-Adapter	+2GB
ControlNet（单模型）	+1.5GB
Wan 14B	20GB
Wan 1.3B	5GB
AnimateDiff	+3GB
FaceDetailer	+2GB

Common Mistakes to Avoid

需避免的常见错误

Wrong output index: CheckpointLoader outputs
```
[model, clip, vae]
```
at indices
```
[0, 1, 2]
```
CFG too high for InstantID: Use 4-5, not default 7-8
Wrong resolution for model: FLUX/SDXL=1024, SD1.5=512
Missing VAE: FLUX needs explicit VAE (
```
ae.safetensors
```
)
Wrong model in wrong loader: Diffusion models need
```
LoadDiffusionModel
```
, not
```
LoadCheckpoint
```

输出索引错误：CheckpointLoader的输出
```
[model, clip, vae]
```
对应索引为
```
[0, 1, 2]
```
InstantID的CFG值过高：建议使用4-5，而非默认的7-8
模型分辨率不匹配：FLUX/SDXL适配1024分辨率，SD1.5适配512分辨率
缺失VAE：FLUX需要显式指定VAE（
```
ae.safetensors
```
）
模型与加载器不匹配：扩散模型需使用
```
LoadDiffusionModel
```
，而非
```
LoadCheckpoint
```

Reference Files

参考文件

```
references/workflows.md
```
- Detailed node-by-node templates
```
references/models.md
```
- Model files and paths
```
references/prompt-templates.md
```
- Model-specific prompts
```
state/inventory.json
```
- Current inventory cache

```
references/workflows.md
```
- 逐节点详细模板
```
references/models.md
```
- 模型文件与路径
```
references/prompt-templates.md
```
- 模型专属提示词模板
```
state/inventory.json
```
- 当前资产清单缓存