Initial commit: Pose Keypoints To Image node for ComfyUI

2026-01-12 10:21:52 +09:00
commit 2472a51c89
6 changed files with 306 additions and 0 deletions
@@ -0,0 +1,10 @@
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+.env
+.venv
+*.egg-info/
+.installed.cfg
+*.egg
@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2024
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
@@ -0,0 +1,48 @@
+# ComfyUI-PoseKeypointsToImage
+
+POSE_KEYPOINT（DWPose/OpenPose形式）からポーズ画像を描画する ComfyUI カスタムノードです。
+
+[English](README_EN.md)
+
+## 概要
+
+DWPose や OpenPose の JSON 形式のキーポイントデータから、ポーズ画像（IMAGE）を生成するノードです。FaceKeypointsFilter で顔パーツをフィルタリングした後の pose_kps を渡すことで、特定のパーツを除外したポーズ画像を作成できます。
+
+## インストール
+
+### ComfyUI Manager を使用（推奨）
+
+ComfyUI Manager で「PoseKeypointsToImage」を検索してインストール。
+
+### 手動インストール
+
+1. ComfyUI の `custom_nodes` フォルダに移動
+2. このリポジトリをクローン:
+   ```bash
+   git clone https://github.com/smartvain/ComfyUI-PoseKeypointsToImage.git
+   ```
+3. ComfyUI を再起動
+
+## 使い方
+
+1. DWPose Preprocessor（または `POSE_KEYPOINT` を出力するノード）の出力をこのノードに接続
+2. パラメータを設定:
+   - `width` / `height`: 出力画像のサイズ（デフォルト: 512x512）
+   - `point_radius`: キーポイントの描画半径（デフォルト: 2）
+   - `draw_body`: ボディキーポイントを描画するか
+   - `draw_face`: 顔キーポイントを描画するか
+   - `draw_hands`: 手のキーポイントを描画するか
+3. 出力を他のノードに接続
+
+## ノードの場所
+
+`Pose/Face Utils` > `Pose Keypoints → Image`
+
+## 互換性
+
+- `comfyui_controlnet_aux` の DWPose Preprocessor と連携
+- OpenPose JSON フォーマットに対応
+
+## ライセンス
+
+MIT License
@@ -0,0 +1,48 @@
+# ComfyUI-PoseKeypointsToImage
+
+A ComfyUI custom node that renders pose images from POSE_KEYPOINT (DWPose/OpenPose format).
+
+[日本語](README.md)
+
+## Description
+
+This node generates pose images (IMAGE) from DWPose or OpenPose JSON format keypoint data. By passing pose_kps that has been filtered with FaceKeypointsFilter, you can create pose images with specific parts excluded.
+
+## Installation
+
+### Using ComfyUI Manager (Recommended)
+
+Search for "PoseKeypointsToImage" in ComfyUI Manager and install.
+
+### Manual Installation
+
+1. Navigate to your ComfyUI's `custom_nodes` folder
+2. Clone this repository:
+   ```bash
+   git clone https://github.com/smartvain/ComfyUI-PoseKeypointsToImage.git
+   ```
+3. Restart ComfyUI
+
+## Usage
+
+1. Connect the output of DWPose Preprocessor (or any node that outputs `POSE_KEYPOINT`) to this node
+2. Configure the parameters:
+   - `width` / `height`: Output image size (default: 512x512)
+   - `point_radius`: Keypoint drawing radius (default: 2)
+   - `draw_body`: Whether to draw body keypoints
+   - `draw_face`: Whether to draw face keypoints
+   - `draw_hands`: Whether to draw hand keypoints
+3. Connect the output to other nodes
+
+## Node Location
+
+`Pose/Face Utils` > `Pose Keypoints → Image`
+
+## Compatibility
+
+- Works with DWPose Preprocessor from `comfyui_controlnet_aux`
+- Compatible with OpenPose JSON format
+
+## License
+
+MIT License
@@ -0,0 +1,3 @@
+from .pose_keypoints_to_image import NODE_CLASS_MAPPINGS, NODE_DISPLAY_NAME_MAPPINGS
+
+__all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS"]
@@ -0,0 +1,176 @@
+# pose_keypoints_to_image.py
+#
+# POSE_KEYPOINT (DWPose / OpenPose 形式) から
+# ポーズ画像（IMAGE）を描き直すノードです。
+
+import numpy as np
+from PIL import Image, ImageDraw
+import torch
+
+
+class PoseKeypointsToImage:
+    """
+    POSE_KEYPOINT からポーズ画像を生成するノード。
+    DWPose の JSON (frame["people"][0]["pose_keypoints_2d"], etc.) に対応。
+    FaceKeypointsFilter で口を消した後の pose_kps などを渡すことで、
+    「口なし」の OpenPose 風画像を作ることができる。
+    """
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "pose_kps": ("POSE_KEYPOINT",),
+                "width": (
+                    "INT",
+                    {"default": 512, "min": 64, "max": 4096, "step": 8},
+                ),
+                "height": (
+                    "INT",
+                    {"default": 512, "min": 64, "max": 4096, "step": 8},
+                ),
+                "point_radius": (
+                    "INT",
+                    {"default": 2, "min": 1, "max": 10, "step": 1},
+                ),
+                "draw_body": ("BOOLEAN", {"default": True}),
+                "draw_face": ("BOOLEAN", {"default": True}),
+                "draw_hands": ("BOOLEAN", {"default": True}),
+            }
+        }
+
+    RETURN_TYPES = ("IMAGE",)
+    RETURN_NAMES = ("image",)
+    FUNCTION = "render"
+    CATEGORY = "Pose/Face Utils"
+
+    # ------------------ helpers ------------------ #
+
+    def _draw_points(self, draw, flat_list, radius, color=(255, 255, 255)):
+        """
+        flat_list: [x0, y0, s0, x1, y1, s1, ...]
+        score(s) > 0 の点だけを描画する
+        """
+        if flat_list is None:
+            return
+        if len(flat_list) % 3 != 0:
+            return
+
+        for i in range(0, len(flat_list), 3):
+            x, y, score = flat_list[i : i + 3]
+            if score is None:
+                continue
+            if score <= 0:
+                continue
+            if np.isnan(x) or np.isnan(y):
+                continue
+            x = float(x)
+            y = float(y)
+            draw.ellipse(
+                (x - radius, y - radius, x + radius, y + radius),
+                fill=color,
+            )
+
+    def _get_people_list(self, frame):
+        """
+        frame["people"] があればそれを返し、
+        無ければ [frame] を 1人分として扱う。
+        """
+        if not isinstance(frame, dict):
+            return []
+
+        if "people" in frame and isinstance(frame["people"], list) and frame["people"]:
+            return frame["people"]
+
+        # 旧形式: frame 自体に keypoints がいる想定
+        return [frame]
+
+    # ------------------ main ------------------ #
+
+    def render(
+        self,
+        pose_kps,
+        width,
+        height,
+        point_radius,
+        draw_body,
+        draw_face,
+        draw_hands,
+    ):
+        """
+        pose_kps: DWPose Estimator などからの POSE_KEYPOINT
+        """
+
+        # pose_kps は通常「各フレーム dict のリスト」
+        if isinstance(pose_kps, dict):
+            frames = [pose_kps]
+        elif isinstance(pose_kps, list):
+            frames = pose_kps
+        else:
+            blank_np = np.zeros((1, height, width, 3), dtype=np.float32)
+            blank = torch.from_numpy(blank_np)
+            return (blank,)
+
+        # 可能なら 1フレーム目の canvas_width/height で上書き
+        if frames and isinstance(frames[0], dict):
+            cw = frames[0].get("canvas_width", width)
+            ch = frames[0].get("canvas_height", height)
+            try:
+                width = int(cw)
+                height = int(ch)
+            except Exception:
+                pass
+
+        images = []
+
+        for frame in frames:
+            img = Image.new("RGB", (width, height), (0, 0, 0))
+            draw = ImageDraw.Draw(img)
+
+            if not isinstance(frame, dict):
+                images.append(np.array(img).astype(np.float32) / 255.0)
+                continue
+
+            people = self._get_people_list(frame)
+
+            for person in people:
+                if not isinstance(person, dict):
+                    continue
+
+                if draw_body and "pose_keypoints_2d" in person:
+                    self._draw_points(draw, person["pose_keypoints_2d"], point_radius)
+
+                if draw_face and "face_keypoints_2d" in person:
+                    self._draw_points(draw, person["face_keypoints_2d"], point_radius)
+
+                if draw_hands:
+                    if "hand_left_keypoints_2d" in person:
+                        self._draw_points(
+                            draw, person["hand_left_keypoints_2d"], point_radius
+                        )
+                    if "hand_right_keypoints_2d" in person:
+                        self._draw_points(
+                            draw, person["hand_right_keypoints_2d"], point_radius
+                        )
+
+            img_arr = np.array(img).astype(np.float32) / 255.0  # H,W,3
+            images.append(img_arr)
+
+        if not images:
+            blank_np = np.zeros((1, height, width, 3), dtype=np.float32)
+            blank = torch.from_numpy(blank_np)
+            return (blank,)
+
+        batch_np = np.stack(images, axis=0)   # (B,H,W,3)
+        batch = torch.from_numpy(batch_np)    # torch.Tensor に変換
+
+        return (batch,)
+
+
+NODE_CLASS_MAPPINGS = {
+    "PoseKeypointsToImage": PoseKeypointsToImage,
+}
+
+NODE_DISPLAY_NAME_MAPPINGS = {
+    "PoseKeypointsToImage": "Pose Keypoints → Image",
+}