webp: no embedded information found.

version: f2.0.1v1.10.1-previous-665-gae278f79 python: 3.10.6 torch: 2.3.1+cu121 gradio: 4.40.0

If the directory of embedded models contains their images in any format, then **Forge rescans each image each time a checkpoint or VAE changes**. Such delays are especially significant if there are many embedded files in the directories with their corresponding pics, and the checkpoint model or VAE change occurs during the workflow cycle itself—for example, a different checkpoint model in HiRes-Fix, Refiner, or ADetailer. Embedded models are used from a single location for different systems (Forge, Comfy, A1111); LoRA Manager is quite popular for Comfy, which automatically creates preview images for embedded models in .webp format.
**Is it possible to exclude the webp format from such frequent scanning, given that it does not contain scannable data?**
To eliminate such delays, we currently had to delete all previews in the embeddings directories and sub-directories.

for example:

Loading Model: {'checkpoint_info': {'filename': 'H:\\SD\\webui\\models\\Stable-diffusion\\01iL\\BSSEquinoxILSemi_v25.safetensors', 'hash': '16d44f2d'}, 'additional_modules': ['H:\\SD\\webui\\models\\VAE\\illustriousXLV20_v10.safetensors'], 'unet_storage_dtype': None}
[Unload] Trying to free all memory for cuda:0 with 0 models keep loaded ... Current free memory is 14096.53 MB ... Unload model JointTextEncoder Current free memory is 15855.76 MB ... Unload model KModel Current free memory is 20852.95 MB ... Unload model IntegratedAutoencoderKL Done.
StateDict Keys: {'unet': 1680, 'vae': 248, 'text_encoder': 197, 'text_encoder_2': 518, 'ignore': 0}
Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
K-Model Created: {'storage_dtype': torch.float16, 'computation_dtype': torch.float16}
**lazydn_pony.webp: no embedded information found.
lazyhand.webp: no embedded information found.
lazyhandV2.webp: no embedded information found.
lazympos.webp: no embedded information found.
lazynegV1.webp: no embedded information found.
lazynegV2.webp: no embedded information found.
lazynegV3.webp: no embedded information found.
lazyposV1.webp: no embedded information found.
lazyposV2.webp: no embedded information found.
lazyreal.webp: no embedded information found.
lazyup6.webp: no embedded information found.
lazyup7.webp: no embedded information found.
lazyup_pony.webp: no embedded information found.
lazydn_pony.webp: no embedded information found.
lazyhand.webp: no embedded information found.
lazyhandV2.webp: no embedded information found.
lazympos.webp: no embedded information found.
lazynegV1.webp: no embedded information found.
lazynegV2.webp: no embedded information found.
lazynegV3.webp: no embedded information found.
lazyposV1.webp: no embedded information found.
lazyposV2.webp: no embedded information found.
lazyreal.webp: no embedded information found.
lazyup6.webp: no embedded information found.
lazyup7.webp: no embedded information found.
lazyup_pony.webp: no embedded information found.**
Model loaded in 6.2s (unload existing model: 3.0s, forge model load: 3.2s).
[Unload] Trying to free 3051.58 MB for cuda:0 with 0 models keep loaded ... Done. | 0/34 [00:00

No source file or test is named. Start by tracing the embedded-model scan that emits the repeated “no embedded information found” messages, then inspect how image extensions are handled when checkpoints or VAEs change. Done means .webp previews are excluded from repeated scanning without affecting supported embedded-model metadata detection.

Written by the indexing model from the issue text.