-
Notifications
You must be signed in to change notification settings - Fork 14
Dataset Organization
After the entire process, you will get the following structure in /path/to/dataset_dir
if you use the default configuration files and run booru
and screenshots
pipelines in parallel.
.
├── intermediate
│ ├── booru
│ │ ├── classified
│ │ ├── cropped
│ │ └── raw
│ └── screenshots
│ ├── animes
│ ├── classified
│ ├── cropped
│ └── raw
└── training
├── booru
│ ├── 1_character
│ ├── 2+_characters
│ └── emb_init.json
├── screenshots
│ ├── 0_characters
│ ├── 1_character
│ ├── 2_characters
│ ├── 3+_characters
│ └── emb_init.json
├── core_tag.json
├── emb_init.json
└── wildcard.txt
💡 If --remove_intermediate
is specified the folders classified
and cropped
are removed during the process.
The folder that should be used for training is /path/to/dataset_dir/training
. Besides the training data, tt contains two important files.
-
emb_init.json
provides information for embedding initialization to be used for pivotal tuning (emb_init.json
in the subfolders can be ignored). -
wildcard.txt
provide the wildcard to be used with sd-dynamic-prompts.
You can put other folders, such as your regularization images in the training folder before launching the process so that they will be taken into account as well when we compute the repeat to balance the concept at the end.
Each folder /path/to/dataset_dir/training/{image_type}
is organized in the following way if --arrange_format
is set to n_characters/character
(the default value).
Level 1
├── ./0_characters
├── ./1_character
├── ./2_characters
├── ./3_characters
├── ./4+_characters
💡 Use --max_character_number n
so that images containing more than n
characters are all put together. If you don't want them to be included in the dataset. You can remove it manually.
Level 2
├── ./1_character
│ ├── ./1_character/AobaKokona
│ ├── ./1_character/AobaMai
│ ├── ./1_character/KuraueHinata
│ ├── ./1_character/KuraueHinata Hairdown
│ ├── ./1_character/KuraueKenichi
│ ├── ./1_character/KuraueMai
│ ├── ./1_character/KurosakiHonoka
│ ├── ./1_character/KurosakiTaiki
...
💡 Use --min_images_per_combination m
so that character combinations with fewer than m
images are all put in the folder character_others
.
TODO: Add add an argument to optionally remove them.
The hierarchical organization allows to auto-balance between different concepts without too much need of worrying about the number of images in each class.
You can pass the argument --extra_path_component
to replace {image_type}
with {extra_path_component}/{image_type}
in the aforementioned paths. This allows you for example to have a good organization when processing multiple animes in parallel.
Note that you will need to set --compute_core_tag_up_levels
to 2 (or even higher number if --extra_path_component
contains path separators) if you want to have a single wildcard and embedding initialization file for the entire dataset. Similarly, you may want to increase --rearrange_up_levels
or --compute_multiply_up_levels
to make sure that dataset balancing is computed from the root training folder.
- Home
- Dataset Organization
- Main Arguments
- Organization of the Character Reference Directory
- Start Training
- Conversion Scripts
- Anime and fanart downloading
- Frame extraction and similar image removal
- Character detection and cropping
- Character classification
- Image selection and resizing
- Tagging, captioning, and generating wildcards and embedding initialization information
- Dataset arrangement
- Repeat computation for concept balancing