Upload your photo and instantly make your favorite photo look like a popular style of artwork. Get a professional artist to turn it into a real physically hand-made piece of art shipped to you for as little as $49!
Get matched with an independent artist to make a canvas painting
Our paint effect is generated by our AI painter, which uses state of the art deep neural network powered convolutional NN models. If you like the preview, we can get a portrait painter to paint it for real and ship it to your doorstep for as little as $49 and in as little as 2 weeks.
Surprise someone with the gift of a real, textured, physical work of art instead of merely a digital painting.
This is the most basic technique and building block for all other techniques for turning a photo into a painting style. The technology usually applies a basic computational convolutional effect, where a small grid of pixels, called a kernel, modifies the underlying pixel values in a static way that is defined by the programmer. A common convolution is the gaussian blur, which applies a blur effect. A small grid of pixels modifies the RGB pixel values to achieve a blur, and this is applied across the whole image.
Most image filters apply a series of image convolution algorithms with manually hand-tuned parameters. Because they are hand-tuned, they result in the basic and fixed effect on most images.
This is a much newer technique that is data driven. A neural network model is trained on images in general so that features can be extracted from the underlying image. Features are encoded grouping of pixels that match an abstract concept of some object that has been learned from the statistical model. For example, with enough training data, a model of images could identify the concept of a person, hand, or stop sign, from pixel values in an image because it can learn the invariant characteristics of those objects (the "big picture" so to speak). Features are extracted from the photo you upload as well as the style image that is being targeted. The convolutional network essentially applies a series of image convolution filters, much like Photoshop filters, and tunes the parameters of each filter automatically to automatically try to match the features extracted from the style image.
This is arguably the most advanced technique. A master artist can interpret what makes up a certain style used in a painting style and apply the same style to another photo reference. The downsides is that it does cost some money to do and they are a bit slower.
At Instapainting, we combine Convolutional Neural Networks and human artists to give the best experience. You can preview your artwork in select styles instantly using our AI Painter, and then get a master human artist to interpret and make it real and physical, and clean up the shortcomings of the AI Painter.
General Questions: support@instapainting.com
Existing Customers
Born in San Francisco.
