1580158
/
batch-face-swap
mirror of https://github.com/kex0/batch-face-swap.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

update guide and examples

alwayson_wip
kex0 2023-05-05 05:15:17 +02:00
parent e1ebb22b70
commit c7c04b9f27
1 changed files with 44 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

57
README.md
View File

@ -1,7 +1,7 @@
## Batch Face Swap extension for https://github.com/AUTOMATIC1111/stable-diffusion-webui
Automaticaly detects faces and replaces them.
![example1](https://user-images.githubusercontent.com/46696708/211933260-7a27cc13-e33a-4bf1-911f-43e0aa97b96c.png)
![preview](https://user-images.githubusercontent.com/46696708/236370022-1e243f62-a59c-437a-a841-d9a3d37778aa.png)
## Installation
### Automatic:
@ -14,20 +14,52 @@
2. Open `requirements_versions.txt` in the main SD web UI folder and add `mediapipe`.
3. Start or reload SD web UI.
## Guide
1. Open `img2img` tab.
2. Select `Batch Face Swap` script.
3. Paste a path of the folder containing your images in the `Images directory` textbox.
4. (Optional) It may sometimes fail to find a face if the face is very small in comparison to the size of the image.
So, you can tell it to split the image and look at the smaller portions of the image by using the `How many images to divide into` slider.
(don't worry it will stitch the image back together)
5. Click `Generate`
## txt2img Guide
1. Expand the `Batch Face Swap` tab in the lower left corner.
<details>
<summary>Image</summary>
<img src="https://user-images.githubusercontent.com/46696708/236360445-7391d68b-4973-4b43-aa75-f8a8782f6c4e.png">
</details>
2. Click the checkbox to enable it.
<details>
<summary>Image</summary>
<img src="https://user-images.githubusercontent.com/46696708/236361252-5a1d05c8-e216-4685-a7ef-80733e08a08a.png">
</details>
3. Click `Generate`
If you want to adjust `Denoising strength` or `Mask blur` you have to disable the override checkboxes.
## img2img Guide
1. Expand the `Batch Face Swap` tab in the lower left corner.
<details>
<summary>Image</summary>
<img src="https://user-images.githubusercontent.com/46696708/236361645-84519cfe-d6a1-492f-adab-3baca037b6de.png">
</details>
2. Click the checkbox to enable it.
<details>
<summary>Image</summary>
<img src="https://user-images.githubusercontent.com/46696708/236361252-5a1d05c8-e216-4685-a7ef-80733e08a08a.png">
</details>
3. You can process either 1 image at a time by uploading your image at the top of the page.
<details>
<summary>Image</summary>
<img src="https://user-images.githubusercontent.com/46696708/236361988-78cfe787-d17a-46a1-bb41-4865a57dcdda.png">
</details>
Or you can give it path to a folder containing your images.
<details>
<summary>Image</summary>
<img src="https://user-images.githubusercontent.com/46696708/236362301-53ce2315-9fa1-46e3-9698-d9164e1354be.png">
</details>
4. Click `Generate`
For more information, activate the `Show tips` checkbox at the very bottom.
Override options only affect face generation so for example in `txt2img` you can generate the initial image with one prompt and face swap with another. Or generate the initial image with one model and faceswap with another.
<details>
<summary>Example</summary>
![chrome_BfZn7JEqVu](https://user-images.githubusercontent.com/46696708/213899772-36498c5f-d8ea-4b15-8b45-d1d11576654c.png)
Left 'young woman in red dress' using `chilloutMix`
Right 'Emma Watson in red dress' using `realisticVision`
<img src="https://user-images.githubusercontent.com/46696708/236363435-07e1cc38-062b-4696-9ce3-11239812f898.png">
</details>
![chrome_XSjamNtABV](https://user-images.githubusercontent.com/46696708/236360114-bd902f03-73cb-4836-a647-27f5371e3197.png)
## Example
![example](https://user-images.githubusercontent.com/46696708/211818536-7d3bd06e-f6b1-40e9-854e-9cb44be3b2f8.png)
@ -36,5 +68,4 @@ Prompt:
```ShellSession
detailed closeup photo of Emma Watson, 35mm, dslr
Negative prompt: (painting:1.3), (concept art:1.2), artstation, sketch, illustration, drawing, blender, octane, 3d, render, blur, smooth, low-res, grain, cartoon, watermark, text, out of focus
Steps: 50, Sampler: Euler a, CFG scale: 7, Seed: 4052732944, Size: 512x512, Model hash: a9263745, Batch size: 8, Batch pos: 1, Denoising strength: 0.5, Mask blur: 4
```