10 really next-level amazing online nueral websites I’m sure you’ve never heard of!

№1 DALL-E 2

American filmmaker made a music video for Nina Simone’s song “Feeling Good” using DALL-E 2 neural network

Filmmaker Karenxcheng posted a clip on her social media accounts created with DALL-E 2, an updated version of a neural network that generates images based only on descriptions.

It can also edit photos to add new subjects and create new drawings based on existing ones as well as make them in the spirit of the original.

№2 Magic Eraser

Magic Eraser — remove excess in a photo. No need to know how to build rockets or work as a designer at Google to remove unnecessary things from your photos. All you need is to have a couple of neural network sites to do it for you. Magic Eraser is one of those. Upload a photo, delete objects, download. Keep in mind that the free version compresses pictures to 600 pixels.

Test it: https://www.magiceraser.io/

№3 Unshorten.link

Tired of getting rickrolled via shortened links? Well,

Find out what a short link hides. Short links should be a supposed threat indicator for everyone, as they can contain ip-identification, advertising, phishing sites and so on. The Unshortern.link extension(Click on extension or https://chrome.google.com/webstore/detail/unshortenlink/gbobdaaeaihkghbokihkofcbndhmbdpd, when you click on a link, will tell you if the end site is safe and if it is linked to OSINT services.

You will also learn in advance to which address the short link will take you.

№4 ThisPersonDoesNotExist (Generates a fake human’s photo)

Well, The person you’re seeing below doesn’t actually exist and this image was made by NVIDIA’s nueral capabilities

The neural network generates a realistic image of a human face. A new image appears each time the page is opened or refreshed. The algorithm is based on the generative neural network StyleGAN from Nvidia.

№5 Quick, Draw!

A truly amusing neural game. The user creates drawings and asks Google algorithms to guess what he meant. The learning model improves as the number of guessed images increases. Speaking truly it’s really amazing to play this and yes after the game ends you can actually see what other people have drawn and the nueral network has learnt them!

№6 Physiology via Video!

Neural network trained to determine physiological indicators from video recording of a patient’s face

A neural network can determine heart rate and blood oxygen saturation levels without the use of medical equipment or a fitness tracker. The technology simplifies the process of medical diagnosis and makes it completely non-contact.

№7 This one can delete the person!

“Removing” a person from a video image in real time

Can you recall when cameras in movies were fed a static image to keep intruders out of sight? Well, they never even dreamed that someday you could just insert an algorithm into the camera code that erases you from the video, everything else would be real.

It is a rare case, when even in science fiction movies they haven’t thought of such a thing. Also Note: the system works directly in the browser, on the client side.

№8 Anime characters from sketches!

A Japanese developer creates a neural network that generates anime characters based on a sketch

Over the last 5–7 years we’ve seen all sorts of neural networks capable of generating anime characters. With each year, the quality of these algorithms has only grown, so now some of them are even used to quickly fill in content or generate basic images for further manual refinement.

One of the most impressive samples today was created by a Japanese developer known on Twitter under the nickname t_takasaka. His experimental model generates not only an illustration based on a rough sketch, but also a three-dimensional model.

Using this algorithm, you can instantly change the style, shape, shadows, facial orientation and other parameters.

If the developer continues active development, that in a few years the program could become the basis for creating anime characters in a variety of areas and without the need to know 3D modeling.

№9 Photo UpScaler

Neuralinka “stretches” the image by a factor of two or four without reducing its quality — it’s really cool. The service supports .png, .jpeg, .jpg and .webp formats.

During the process you can also remove noise, but this option doesn’t always work (sometimes it makes the image look unnatural and “oily”). But the main thing is that everything (in general) is free.

Website of the project - https://www.upscale.media/

№10 Doodle to Image

Artificial Intelligence from Nvidia for semi-automated picture generation has quadrupled the resolution of the resulting images, as well as adding more materials to the generation. The result is in the video:
On the left is what the author draws, silently … just a sketch/idea; on the right is how the AI helps to finish it off.

You can download the Canvas beta — http://gaugan.org/gaugan2/

WHOO!

We made it to the the bottom! If you liked this post then a clap would be nueralistic!