kohya-ss / kohya-ss/sd-scripts

A massive NPZ file when using alpha mask

Open
#1,739 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
7.2k
Forks
1.2k
Avg merge
11m
Merged PRs (30d)
2

Description

![image](https://github.com/user-attachments/assets/4bfc7ad3-c1c9-4c4b-b744-c547dff1dc50)
When using --alpha_mask, the size of the SDXL NPZ file skyrockets from 25X KB to 4 MB.
Since it’s just a 1 channel mask, the storage requirement is excessively large.

Contributor guide

No contributing guide indexed for this repository

Research direction

Start by reproducing an SDXL run with --alpha_mask and compare the resulting NPZ file size and stored mask data with a run that does not use the option. Trace where the alpha mask is written, then verify that the one-channel mask no longer causes excessive storage growth while preserving the mask data.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.