AUTOMATIC1111/stable-diffusion-webui · error · FileNotFoundError

DAT data missing: {scaler.local_data_path}

Error message

DAT data missing: {scaler.local_data_path}

What it means

FileNotFoundError raised by DatUpscaler.load_model after the scaler's URL was (or was not) downloaded via load_file_from_url but scaler.local_data_path still does not exist on disk. DAT models ship as .pth URLs on GitHub; when the download fails silently or writes elsewhere, the os.path.exists check trips before the scaler is returned.

Source

Thrown at modules/dat_model.py:54

            expected_architecture="DAT",
        )
        return upscale_with_model(
            model_descriptor,
            img,
            tile_size=opts.DAT_tile,
            tile_overlap=opts.DAT_tile_overlap,
        )

    def load_model(self, path):
        for scaler in self.scalers:
            if scaler.data_path == path:
                if scaler.local_data_path.startswith("http"):
                    scaler.local_data_path = modelloader.load_file_from_url(
                        scaler.data_path,
                        model_dir=self.model_download_path,
                    )
                if not os.path.exists(scaler.local_data_path):
                    raise FileNotFoundError(f"DAT data missing: {scaler.local_data_path}")
                return scaler
        raise ValueError(f"Unable to find model info: {path}")


def get_dat_models(scaler):
    return [
        UpscalerData(
            name="DAT x2",
            path="https://github.com/n0kovo/dat_upscaler_models/raw/main/DAT/DAT_x2.pth",
            scale=2,
            upscaler=scaler,
        ),
        UpscalerData(
            name="DAT x3",
            path="https://github.com/n0kovo/dat_upscaler_models/raw/main/DAT/DAT_x3.pth",
            scale=3,
            upscaler=scaler,
        ),

View on GitHub (pinned to 82a973c043)

Solutions

  1. Manually download the required DAT_x{2,3,4}.pth from the GitHub repo and place it in models/DAT/
  2. Verify models/DAT exists and is writable; remove partial files then retry so load_file_from_url re-downloads
  3. Check the server can reach github.com (proxy env vars HTTP(S)_PROXY set for the webui process)

Example fix

# before: selecting DAT x2 in Extras -> FileNotFoundError

# after
mkdir -p models/DAT
curl -L -o models/DAT/DAT_x2.pth https://github.com/n0kovo/dat_upscaler_models/raw/main/DAT/DAT_x2.pth
Defensive patterns

Strategy: validation

Validate before calling

import os
need = {'DAT x2':'DAT_x2.pth','DAT x3':'DAT_x3.pth','DAT x4':'DAT_x4.pth'}
missing = [f for f in need.values() if not os.path.isfile(os.path.join('models','DAT',f))]
if missing:
    raise FileNotFoundError(f'DAT weights missing: {missing}; download into models/DAT/')

Type guard

def dat_weights_present(name: str, dat_dir='models/DAT') -> bool:
    return os.path.isfile(os.path.join(dat_dir, f"DAT_x{name.split()[-1][1:]}.pth"))

Try / catch

try:
    scaler.load_model(path)
except FileNotFoundError as e:
    if 'DAT data missing' in str(e):
        download_dat_weights(path); return scaler.load_model(path)
    raise

Prevention

When it happens

Trigger: Selecting the DAT upscaler in Extras while models/DAT/DAT_x2.pth (or x3/x4) is missing because the HTTPS download from github.com/n0kovo/dat_upscaler_models was blocked/failed; model dir redirected (custom --dat-models-path mismatch); file deleted between listing and load; partial download that saved to a temp name.

Common situations: Offline or China-region deployments where raw.githubusercontent/github downloads stall; interrupted first use leaving a corrupted cache entry; permission errors on the models directory so the file is never written.

Related errors

AI-assisted analysis of AUTOMATIC1111/stable-diffusion-webui@82a973c043 (2026-08-14). Data as JSON: /api/errors/a1d0b6f8ab9fddb1. Report an issue: GitHub.