|
4 | 4 |
|
5 | 5 | Inspired by the Extra Network tabs in Automatic1111's WebUI, Extra Network Browser is a stand-alone take on the concept with additional features.
|
6 | 6 |
|
7 |
| -Easily choose a LoRA, HyperNetwork, Embedding, Checkpoint, or Style visually and copy the trigger, keywords, and suggested weight *to the clipboard* for easy pasting into the application of your choice. |
| 7 | +Easily choose a **LoRA, HyperNetwork, Embedding, Checkpoint, or Style** visually and copy the trigger, keywords, and suggested weight to the clipboard for easy pasting into the application of your choice. |
8 | 8 |
|
9 |
| -Advantages over the Extra Network Tabs: |
| 9 | +Features of Extra Network Browser: |
10 | 10 |
|
11 | 11 | * Great for UI's like ComfyUI when used with nodes like [Lora Tag Loader](https://github.com/badjeff/comfyui_lora_tag_loader/) or [ComfyUI Prompt Control](https://github.com/asagi4/comfyui-prompt-control).
|
12 |
| -* Considerably faster, loads thousands of LoRA. |
| 12 | +* Considerably fast, loads thousands of LoRA easily. |
13 | 13 | * A [Styles](#styles) tab that parses a styles CSV for thumbnail previews just like networks.
|
14 | 14 | * [Keywords](#keywords) in the filename inside brackets [ ]'s are copied along with the LoRA trigger.
|
15 | 15 | * [Weights](#weights) placed in braces { }'s *(eg {1.0} or {0.7-0.8})* in the filename are automatically set in the LoRA's trigger.
|
16 | 16 | * Some characters not compatible with filenames are automatically converted from placeholders, such as ©️ to : *(for [keywords with weights](#weights))*
|
17 | 17 | * Sort by Name, Date Modified, or try Random sort for inspiration.
|
18 | 18 | * Support for multiple images per LoRA/model/etc in a [modal gallery](#modal) (including filename [search](#search)). Hover over a card & click the folder icon.
|
19 | 19 | * Support for [displaying a companion .txt file](#modal) to store descriptions, notes, and prompts. Hover over a card & click the document icon.
|
| 20 | +* [Poses](#poses) tab to visually display OpenPose collections and examples. |
20 | 21 | * [Gallery](#modal) tab for arbitrary image folders, such as saved generation results.
|
21 | 22 |
|
22 | 23 | <br />
|
@@ -53,7 +54,7 @@ cd ../app && npm install
|
53 | 54 |
|
54 | 55 | Then add content:
|
55 | 56 |
|
56 |
| -- Populate the folders in `api/networks/` with your files: `lora`, `checkpoints`, `embeddings`, `hypernets`, `styles`, and `gallery`. |
| 57 | +- Populate the folders in `api/networks/` with your files: `lora`, `checkpoints`, `embeddings`, `hypernets`, `styles`, `poses`, and `gallery`. |
57 | 58 | - Edit `api/networks/styles.csv` with *(only)* `name,prompt` on the first line, and your styles (following the format shown [below](#styles)) on the subsequent lines.
|
58 | 59 |
|
59 | 60 | ***OR***
|
@@ -228,17 +229,49 @@ In this example, your matching image files in `api/networks/styles` would be:
|
228 | 229 |
|
229 | 230 | ---
|
230 | 231 |
|
| 232 | +<a id="poses"></a> |
| 233 | +### Poses: |
| 234 | + |
| 235 | +The Poses tab ("P") is designed to work with OpenPose images used with ControlNet. In this tab, you can manage examples of those poses for easy browing. |
| 236 | + |
| 237 | +This tab works somewhat differently than the model tabs, in that it looks for images rather than model files.<br /> |
| 238 | +I've tried to somewhat match the format of many zipped packs I've seen, and as such I recommend the following folder structure: |
| 239 | + |
| 240 | +``` |
| 241 | +/api/poses/<pose>/ |
| 242 | +/api/poses/<posefolder>/pose1.png |
| 243 | +/api/poses/<posefolder>/pose2.png |
| 244 | +/api/poses/<posefolder>/OpenPose/pose1.png #optional |
| 245 | +/api/poses/<posefolder>/OpenPose/pose2.png #optional |
| 246 | +/api/poses/<posefolder>/Depth/pose1.png #optional |
| 247 | +/api/poses/<posefolder>/Canny/pose2.png #optional |
| 248 | +``` |
| 249 | + |
| 250 | +The structure here is: |
| 251 | +* A general name for the pose folder *(e.g. "Heart Hands" or "T Pose")* |
| 252 | +* Example images with the pose inside that folder |
| 253 | +* OpenPose mannequin files inside a subfolder called OpenPose |
| 254 | +* Optionally any other ControlNet files stored likewise *(e.g. Depth, Canny, or Lineart folders)* |
| 255 | + |
| 256 | +For best compatibility this tab uses **.png** instead of **.jpeg**. |
| 257 | +*This can be changed in `api/index.js` by editing `const ext ... "png" : imgExt`, on line 260 to `jpeg`, etc.* |
| 258 | + |
| 259 | +--- |
| 260 | + |
231 | 261 | <a id="gallery"></a>
|
232 | 262 | ### Gallery:
|
233 | 263 |
|
234 |
| -The Gallery tab *("G")* looks for *subfolders* inside the `api/networks/gallery/` folder and displays a card for each, using a .jpeg matching the folder name in the `/gallery`. |
| 264 | +The Gallery tab ("G") looks for *subfolders* inside the `api/networks/gallery/` folder and displays a card for each, using a .jpeg matching the folder name in the `/gallery`. |
235 | 265 | Examples:
|
236 | 266 | ```
|
237 |
| -api/networks/gallery/saved-images/ # Gallery folder with images |
238 |
| -api/networks/gallery/saved-images.jpeg # Gallery folder thumbnail image |
239 |
| -
|
240 |
| -api/networks/gallery/testing-images/ # Gallery folder with images |
241 |
| -api/networks/gallery/testing-images.jpeg # Gallery folder thumbnail image |
| 267 | +api/networks/gallery/saved-images.jpeg # Gallery folder thumbnail image |
| 268 | +api/networks/gallery/saved-images/one.jpeg # Gallery folder with images |
| 269 | +api/networks/gallery/saved-images/two.jpeg # Gallery folder with images |
| 270 | +api/networks/gallery/saved-images/three.jpeg # Gallery folder with images |
| 271 | +
|
| 272 | +api/networks/gallery/testing-images.jpeg # Gallery folder thumbnail image |
| 273 | +api/networks/gallery/testing-images/cat.jpeg # Gallery folder with images |
| 274 | +api/networks/gallery/testing-images/dog.jpeg # Gallery folder with images |
242 | 275 | ```
|
243 | 276 |
|
244 | 277 | <a id="search"></a>
|
|
0 commit comments