![Teddy 01](https://www.xndre.com/wp-content/uploads/2022/05/Teddy01.png)
![Teddy 02](https://www.xndre.com/wp-content/uploads/2022/05/Teddy02.png)
![Teddy 03](https://www.xndre.com/wp-content/uploads/2022/05/Teddy03.png)
![Teddy 04](https://www.xndre.com/wp-content/uploads/2022/05/Teddy04.png)
![Teddy 05](https://www.xndre.com/wp-content/uploads/2022/05/Teddy05.png)
![Teddy 06](https://www.xndre.com/wp-content/uploads/2022/05/Teddy06.png)
![Teddy 07](https://www.xndre.com/wp-content/uploads/2022/05/Teddy07.png)
![Teddy 08](https://www.xndre.com/wp-content/uploads/2022/05/Teddy08.png)
Another experiment in using faces recognition software (see the Bella series) to crop source images to make pixel portraits. What if I did not copy over pixels, but larger pieces of the source image?
I used images of Teddy Quinlivan to create this series. To test what would work best I let the program size the areas randomly, between 5 x 5 px (1/100 of the source) to 250 x 250 (1/4). And as before I let the program run and create a 1000 images, from which I selected the 8 that I liked best.
In the end every size of the blocks generated interesting and convincing results, so I decided to select 8 with all different block size.
![Teddy 01](https://www.xndre.com/wp-content/uploads/2022/05/Teddy01.png)
![Teddy 02](https://www.xndre.com/wp-content/uploads/2022/05/Teddy02.png)
![Teddy 03](https://www.xndre.com/wp-content/uploads/2022/05/Teddy03.png)
![Teddy 04](https://www.xndre.com/wp-content/uploads/2022/05/Teddy04.png)
![Teddy 05](https://www.xndre.com/wp-content/uploads/2022/05/Teddy05.png)
![Teddy 06](https://www.xndre.com/wp-content/uploads/2022/05/Teddy06.png)
![Teddy 07](https://www.xndre.com/wp-content/uploads/2022/05/Teddy07.png)
![Teddy 08](https://www.xndre.com/wp-content/uploads/2022/05/Teddy08.png)