Dreamer Daniel Zamir has programmed a nifty little app that can recognize what is in an image – in particular, a dream image. To try it out,go to Daniel’s GitHub page first to download the app. Next, click on the green “Clone or download” button. If you download the ZIP file, open the file, then open the folder “dist” and click on “app.exe.” Should your computer refuse to open the app because Windows defender does […]
Like thispersondoesnotexist, GauGAN (named after artist Paul Gauguin and GANs, i.e. generative adversarial networks) is another GAN-based tool developed by NVIDIA. This interactive app uses familiar paint program devices such as paint buckets and brushes to suggest a landscape in the form of a segmentation map on the left side of the screen. The GAN simultaneously translates your scribbling into a virtual landscape of your choice on the right side of your screen. What is […]
The links below take you to selected deep-dreaming-related scientific material: Books, Papers, Articles Where We See Shapes, AI Sees Textures (2019) by Jordana Cepelewicz in Quantamagazine Controlling Perceptual Factors in Neural Style Transfer (2017) by Leon A. Gatys et al. Preserving Color in Neural Artistic Style Transfer (2016) by Leon A. Gatys et al. A Neural Algorithm of Artistic Style (2015) by Leon A. Gatys et al. Music Transformer (2018) by Cheng-Thi Anna Huang […]
This waifu does not exist produces an endless stream of potential source images for deep dreaming. Create a new waifu (a fictional character, typically an anime, manga, or video game character that is attractive to the point of being considered a significant other) every time you refresh the website. Click on pause refresh at the bottom of a waifu page to read the little made-up story to the right of the face. Alternatively, call […]
This person does not exist! Every time you refresh the website at two-second intervals, you conjure up a new synthetic face. Most portraits look uncannily real. The website uses code released by GPU inventor NVIDIA using data from generative adversarial networks (GANs) and style transfer to blend images. Software engineer Philip Wang created the thispersondoesnotexist website. To find out about Wang’s method and motives, look at his Facebook post. Also, if you’re interested in relevant […]
Deep Dreamers have found various ways to extend their activities to YouTube. YouTube and Dreaming Music The strongest connection between YouTube and Deep Dreamers is a music channel, DDG Music. All dreamers in the Facebook group are welcome to add tracks under Wendy Bandurski’s FB #music post. Also, some group members use YouTube to display their dreams as videos set to music. See, for example, Ema Vietnam’s series of “dreams for beholders” on her channel. […]
AI expert Cristóbal Valenzuela builds machine learning and AI tools for artists and designers. He programmed a simple text-to-image generator that you can call up under the link above. Open the website and type a text into a window on the left. Next, as you type, the corresponding image takes shape in the window on the right. In the background, the recognition algorithms create images – nightmare images – out of your words. Unfortunately, the […]
NeuralStyler uses AI to convert videos, GIFs or photos into artwork using a handful of styles of famous artists such as Van Gogh. You can create your own styles (train the neural network with styles of your own or contribute styles). Download NeuralStyler for Windows or Linux. Apparently, rendering is very slow (see article at https://www.ghacks.net/2016/09/14/neuralstyler-turn-videos-into-art/).
Painnt is a neural image enhancer app produced by Moonlighting Apps LLC (an apps publisher based in Cordoba, Argentina). Painnt is available in a free and paid version for Windows, Mac, iOS and Android. The free version limits output resolution, has ads and adds a watermark, plus you can choose from a library of some 1,000 styles. Over 2,000 styles are available to subscribers, or you can use custom styles. For the paid version, every […]
From the paper: We introduce Creatism, a deep-learning system for artistic content creation. In our system, we break down aesthetics into multiple aspects, each can be learned individually from a shared dataset of professional examples. Using our system, we mimic the workflow of a landscape photographer, from framing for the best composition to carrying out various post-processing operations. The environment for our virtual photographer is simulated by a collection of panorama images from Google Street […]
At the end of this tutorial you should have a basic car, navigating and avoiding walls. It will start by driving randomly into walls, but learn over time.