What are NPZ files? about taggui HOT 7 CLOSED

I have added .npz to the list of excluded file types when loading a directory. I think it is safer to specify file types to exclude rather than file types to include in order not to miss any obscure image formats.

from taggui.

Haha, this is indeed a difficult choice.
On the one hand, it is impossible to determine which file types of images the user will use, and on the other hand, it is impossible to determine whether the user will put other types of files and images in the same folder.
However, because I only use jpg and png file, I made such modifications to the code.

from taggui.

Thank you for the suggestions. I will keep them in mind, but I will close the issue for now.

from taggui.

Are you sure TagGUI is creating the .npz files? I haven't seen them being created when I use TagGUI.

.npz seems to be a file format used by NumPy to save arrays (docs), but there's no code in TagGUI that calls NumPy like that.

from taggui.

The .npz files seem to be created by kohya_ss when saving latents to disk (https://github.com/kohya-ss/sd-scripts/blob/main/library/train_util.py#L2007).

from taggui.

Yes, kohya_ss will create .npz files if you use the save_latent_to_disk parameter during training. But the .npz file will not cause TagGUI to report an error, it will just appear in the ImageList together with the image (this was the case in the previous version). In response, I modified the code here from excluding json files to only accepting image type files.
From if path.suffix.lower() not in ('.json', '.jsonl') to if path.suffix.lower() in (".png", ".jpg", ".jpeg").

from taggui.

Man, you guys are amazing! Wanna take over some of my other favorite repos? Anyways, I must've made the NPZs when I ran Kohya_SS and received the error relating to the text captioning file format. I apologize for writing up a bogie, fellas.

I formatted the title, rather than deleting the issue, but I can nuke it if you guys want, just lmk.

I also left the issue open, for you to decide if you want to want to attempt any PRs to address the issue. If you decide to do something about the default, even though no one uses it, you could do one of the following:

• Add an option to save files as *.caption during generation.
• Add buttons that rename all caption file extensions in the folder to whichever format, so a 'Make files TXT' button and a 'Make files *.CAPTION' perhaps.
• Add it as a setting somewhere.

Thanks for the lightning fast response, sorry my own was less so...

from taggui.