RealityScan 2.2 + NerfStudio
数据转换排错记录

D:\temp\ReasonixChat\3\
├── orin-images\          # 原始图像(手机拍摄):IMG_20260702_165925.jpg ~ IMG_20260702_170007.jpg,共 16 张
├── images\               # RealityScan 导出的 COLMAP 配套图像:00000.png ~ 00015.png(已重命名,共 16 张)
├── sparse\0\             # RealityScan 导出的 COLMAP 稀疏重建(文本格式)
│   ├── cameras.txt       # 16 个相机,PINHOLE 模型(无畸变)
│   ├── images.txt        # 16 张图像的位姿(qvec + tvec)+ 2D 特征点
│   └── points3D.txt      # 23513 个稀疏 3D 点
└── 3.csv                 # RealityScan 2.2 "Internal/External camera parameters" 导出,16 行

ns-process-data realitycapture --data {data directory} --csv {csv file} --output-dir {output directory}

UnicodeEncodeError: 'gbk' codec can't encode character '\U0001f389' in position 0: illegal multibyte sequence

PYTHONIOENCODING=utf-8 ns-process-data realitycapture --data orin-images --csv 3.csv --output-dir out_test

Missing image data for 16 cameras.
Missing camera data for 16 frames.
Final dataset is 0 frames.

File "...\nerfstudio\process_data\realitycapture_utils.py", line 82, in realitycapture_to_json
    frame["fl_x"] = float(cameras["f"][i]) * scale / 36.0
KeyError: 'f'

for i, name in enumerate(cameras["#name"]):          # 行 65
    ...
    frame["fl_x"] = float(cameras["f"][i]) * scale / 36.0        # 行 82 → KeyError: 'f'
    frame["cx"] = float(cameras["px"][i]) * scale + width / 2.0  # 行 84
    frame["cy"] = float(cameras["py"][i]) * scale + height / 2.0 # 行 85
    ...
    rot = _get_rotation_matrix(-float(cameras["heading"][i]),   # 行 94
                               float(cameras["pitch"][i]),
                               float(cameras["roll"][i]))

"""Map RealityScan 2.2 CSV column names to the names nerfstudio 1.1.5 expects.

Usage:
    python fix_rc_csv.py <input.csv> [output.csv]

nerfstudio 1.1.5 realitycapture_utils.py expects:
    #name, x, y, alt, heading, pitch, roll, f, px, py, k1, k2, k3, k4, t1, t2
RealityScan 2.2 exports:
    #name, x, y, alt, yaw, pitch, roll, f_35mm, px_norm, py_norm, k1, k2, k3, k4, t1, t2
"""
import csv
import sys

COLUMN_MAP = {
    "yaw": "heading",
    "f_35mm": "f",
    "px_norm": "px",
    "py_norm": "py",
}

def main() -> None:
    if len(sys.argv) < 2:
        print(__doc__)
        sys.exit(1)
    src = sys.argv[1]
    dst = sys.argv[2] if len(sys.argv) > 2 else src.replace(".csv", "_nerfstudio.csv")
    with open(src, encoding="utf-8-sig", newline="") as fin, open(dst, "w", encoding="utf-8", newline="") as fout:
        reader = csv.DictReader(fin)
        fieldnames = [COLUMN_MAP.get(c, c) for c in reader.fieldnames]
        writer = csv.DictWriter(fout, fieldnames=fieldnames)
        writer.writeheader()
        for row in reader:
            writer.writerow({COLUMN_MAP.get(k, k): v for k, v in row.items()})
    print(f"written: {dst}")
    print("header:", fieldnames)

if __name__ == "__main__":
    main()

- #name,x,y,alt,yaw,pitch,roll,f_35mm,px_norm,py_norm,k1,k2,k3,k4,t1,t2
+ #name,x,y,alt,heading,pitch,roll,f,px,py,k1,k2,k3,k4,t1,t2

# 1) 列名映射(生成 3_nerfstudio.csv)
python fix_rc_csv.py 3.csv

# 2) 格式转换(--data 必须指向【原始】图像目录;加 PYTHONIOENCODING 规避 GBK 报错)
PYTHONIOENCODING=utf-8 ns-process-data realitycapture \
  --data orin-images \
  --csv 3_nerfstudio.csv \
  --output-dir output

Started with 16 images
Final dataset is 16 frames.

PYTHONIOENCODING=utf-8 ns-train splatfacto \
  --data output \
  --output-dir ./runs \
  --experiment-name myscan \
  --viewer.quit-on-train-completion True \
  nerfstudio-data --downscale-factor 2

# 导出训练完成的模型为点云 / 网格等
ns-export gaussian-splat --load-config runs/myscan/splatfacto/2025-xxxx/xxxx/config.yml --output-dir exports

ns-viewer --load-config runs/myscan/splatfacto/2025-xxxx/xxxx/config.yml

python fix_rc_csv.py 3.csv
PYTHONIOENCODING=utf-8 ns-process-data realitycapture --data orin-images --csv 3_nerfstudio.csv --output-dir output

# gaussian-splatting(main)+ 三个 submodule + glm
git clone 失败时改用:curl -L -o gs.zip "https://ghfast.top/https://github.com/graphdeco-inria/gaussian-splatting/archive/refs/heads/main.zip"
# 注意 diff-gaussian-rasterization 必须用 dr_aa 分支(主仓库代码调用 antialiasing 参数):
curl -L -o dgr.zip "https://ghfast.top/https://github.com/graphdeco-inria/diff-gaussian-rasterization/archive/refs/heads/dr_aa.zip"
# simple-knn 在 INRIA GitLab:https://gitlab.inria.fr/bkerbl/simple-knn
# fused-ssim(metrics 用):https://github.com/rahul-goel/fused-ssim
# glm(光栅化依赖):https://github.com/g-truc/glm → submodules/diff-gaussian-rasterization/third_party/glm

@echo off
call "C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Auxiliary\Build\vcvars64.bat"
set DISTUTILS_USE_SDK=1
cd /d D:\temp\gaussian-splatting\submodules\diff-gaussian-rasterization
python setup.py install
cd /d D:\temp\gaussian-splatting\submodules\simple-knn
python setup.py install
cd /d D:\temp\gaussian-splatting\submodules\fused-ssim
python setup.py install

"-allow-unsupported-compiler", "-D_ALLOW_COMPILER_AND_STL_VERSION_MISMATCH"

# txt → bin(colmap 官方二进制格式;qvec/tvec 是 float64,容易写错)
python tmp/colmap_txt2bin.py
# 图像+内参同步缩放到 0.5x(P104 显存/速度考虑;3DGS 无内置缩放)
python tmp/prepare_gs_data.py 0.5
# 产物:D:\temp\ReasonixChat\3\gs_data(images/ + sparse/0/{cameras,images,points3D}.bin)

cd D:\temp\gaussian-splatting
python train.py -s D:/temp/ReasonixChat/3/gs_data -m D:/temp/ReasonixChat/3/gs_output --iterations 7000
python render.py -m D:/temp/ReasonixChat/3/gs_output -s D:/temp/ReasonixChat/3/gs_data --iteration 7000

python train.py -s D:/temp/ReasonixChat/3/gs_data -m D:/temp/ReasonixChat/3/gs_output_full --iterations 30000