¿Sueñan los androides con obras de Banksy?
Creatividad, arte e inteligencia artificial
![Edward Belamy](./static/images/edward_belamy.png)
![Edward Belamy](./static/images/edward_belamy.png)
Portrait of Edmond De Belamy
Ink on canvas. 70 cm × 70 cm
![](./static/images/family_belamy_1.png)
![](./static/images/family_belamy_2.png)
![](./static/images/family_belamy_3.png)
![](./static/images/family_belamy_4.png)
![](./static/images/family_belamy_5.png)
![](./static/images/family_belamy_6.png)
![Edward Belamy](./static/images/edward_belamy.png)
![Signature](./static/images/belamy_signature.png)
![Edward Belamy](./static/images/edward_belamy.png)
Creativity the ability to produce ideas or artefacts that are new, surprising, and valuable.
Margaret Boden, Professor of Cognitive Science
Intentaremos responder..
- ¿Cuáles son los algoritmos?
- ¿Cómo se aplica al arte?
- ¿Quienes son los artistas?
- ¿Cómo puedo hacerlo yo?
Los algoritmos
Los algoritmos
Red neuronal artificial
f(x) = y
Red neuronal artificial
![](./static/images/base_what_is_ANN.png)
Red neuronal artificial
![](./static/videos/what_is_ANN_1.gif)
Red neuronal artificial
![](./static/videos/what_is_ANN_2.gif)
Red neuronal artificial
![](./static/videos/what_is_ANN_3.gif)
Conjunto de datos MNIST
![](./static/images/mnist_3digit.png)
Dada una imagen de un dígito manuscrito ¿cuál es ese dígito?
![](./static/images/ANN_D_7_frame0.png)
![](./static/images/ANN_D_2_frame0.png)
![](./static/images/MNIST_instances.png)
![](./static/images/mnist_test_all.png)
![](./static/videos/base_CNN.png)
Red neuronal discriminativa
![](./static/images/GANs_discriminativo_black.png)
Red neuronal generativa
![](./static/images/generativa_a.png)
Red neuronal generativa
![](./static/images/generativa_b.png)
Red neuronal generativa
![](./static/images/generativa_c.png)
![](./static/images/evaluate_GAN.png)
Red neuronal generativa
![](./static/images/GANs_generativo_black.png)
Red neuronal generativa + discriminativa
![](./static/images/GANs_completo_before_black.png)
Red generativa antagónica (GAN)
![](./static/images/GANs_completo_black.png)
![](./static/images/GAN_MNIST_base.png)
![](./static/images/base_NVIDIA_GAN_dataset.png)
![](./static/images/eigenfaces.png)
![](./static/images/base_GAN_faces_learning.png)
thispersondoesnotexist.com
![](./static/images/base_face_style.png)
![](./static/images/base_deepfake_a.png)
![](./static/videos/deepfake_a.gif)
![](./static/videos/deepfake_a2.gif)
![](./static/videos/deepfake_b.gif)
El arte y sus protagonistas
![](./static/images/wikiart_portraits.jpg)
![](./static/images/portraitnet_waifu2x.png)
![Edward Belamy](./static/images/edward_belamy.png)
Portrait of Edmond of Belamy
![](./static/images/wikiart_landscape.png)
![](./static/images/landscapenet_waifu2x.png)
![](./static/images/ts_a_1.png)
![](./static/images/ts_c_1.png)
![](./static/images/ts_d_1.png)
GANism (the specific look and feel of seemingly GAN-generated images) may yet become a significant modern art trend.
![](./static/videos/balenciaga_dataset.gif)
![](./static/images/balenciaga_results.png)
![](./static/images/balenciaga_image1.jpg)
![](./static/images/balenciaga_image2.jpg)
![](./static/images/balenciaga_image3.jpg)
![](./static/images/balenciaga_image4.jpg)
![](./static/images/RB_infinite_skulls.png)
Neural zoo
![](./static/images/sc_neural_zoo_a.png)
![](./static/images/sc_neural_zoo_b.jpeg)
![](./static/images/sc_neural_zoo_c.jpeg)
Perception engines
![](./static/images/tw_perception_engines.jpeg)
![Twitter](./static/images/AR_avatar.jpeg)
Fall of the House of Usher
![](./static/images/AR_fall_house_usher.png)
![](./static/images/obviuous.png)
Sander Dieleman
bennanne.github.io
![](./static/videos/audio_a.gif)
![](./static/videos/audio_b.gif)
![](./static/images/audio_c.png)
Articulations
![](./static/images/poema_esp.png)
En la piel del artista
Transferencia de estilo
github.com/cysmith/neural-style-tf
github.com/CompVis/adaptive-style-transfer
1. Configurarlo (╯°□°)╯︵ ┻━┻
1. 1 Conseguir ejecutar comandos por consola
1. 2 Python, git y tensorflow instalado
2. git clone https://github.com/cysmith/neural-style-tf.git
3. Copiar imágenes a transformar en la carpeta './image_input/'
4. Copiar las imágenes con los estilos en la carpeta './styles'
5. bash stylize_image.sh <ruta_a_image_input> <ruta_a_styles>
6. Ver si funciona ¯\_(ツ)_/¯
7. Maldecir e ir a https://deepart.io/
![](./static/images/okuda_slide.jpg)
Red generativa
github.com/robbiebarrat/art-DCGAN
github.com/mlberkeley/Creative-Adversarial-Networks
1. Configurarlo (╯°□°)╯︵ ┻━┻
1. 1 Conseguir ejecutar comandos por consola
1. 2 Torch, git, Lua, CUDA, CuDNN instalado
2. Buscar en Google como instalar cuDNN (┛◉Д◉) ┛彡┻━┻
3. https://github.com/robbiebarrat/art-DCGAN.git
4. Copiar las imágenes con los en la carpeta './images'
5. DATA_ROOT=myimages dataset=folder ndf=50 ngf=150 th main.lua
ndf = número de capas en el discriminador
ngf = número de capas en el generador
6. Ver lo difícil que es que converja ┻━┻︵└(՞▃՞ └)
7. DATA_ROOT=myimages dataset=folder
netD=discriminator.t7
netG=generator.t7
th main.lua
8. net=your_generator_net.t7 th generate.lua
Y para terminar
The creative act does not reside in the output generated by the machine, but in its initial human programming.
The art generated by AI is not creative, because it is unintentional.
![](./static/images/game_a.png)
![](./static/images/game_b.png)
Art, Creativity, and the Potential of Artificial Intelligence
![](./static/images/AICAN1.png)
![](./static/images/AICAN_results1.png)
![](./static/images/GAN_Evol.png)
Bibliografía y referencias
- Tensorspace.org
- But what *is* a Neural Network? 3Blue1Brown
- Progressive Growing of GANs for Improved Quality, Stability, and Variation, Karras et al ,2017
- A Style-Based Generator Architecture for Generative Adversarial Networks
- Face2Face: Real-time Face Capture and Reenactment of RGB Videos
- Do androids dream of Balenciaga SS29?
- Abvious art
- The challenge of realistic music generation: modelling raw audio at scale
- CAN: Creative Adversarial Networks, Generating Art by Learning About Styles and Deviating from Style Norms, Elgammal et al, 2017
- Can you guess the airwork?
- AI’s Growing Role in Musical Composition
- AI can produce pictures, but can it create art for itself?
- GANGogh: Creating Art with GANs