If you’ve ever wanted to be the next Andy Warhol, but just didn’t have the artistic talent, now AI can do it for you. Style transfer neural networks enable you to apply an artistic style to an image. They are created by taking an existing image classification network and using it as a loss function in training the transfer network, comparing transformation output to training images and an example style image, such as a painting. Intuitively, we are using a pre-trained network to train a new network to mix semantic features of a target image with textures from a style image.

We make use of JC Johnson’s fast-neural-style implementation of this process, using his pre-trained VGG16 network to calculate the loss during training. We use K80 GPUs on Rescale to efficiently train the transformation network and then apply the network to new images.

Style transfer happens in two steps: 1) training the style transfer network itself and 2) applying the transfer model to new images. Below, we show how to set up a Rescale job to perform each step.

Below are four of Rescale’s team members styled to various artworks, including Jackson Pollock, Takashi Murakami, and photo of a skyscraper in Hong Kong. See if you can guess which is which.  


Style Transfer can be easily achieved on Rescale’s cloud computing platform with our new K80 GPU hardware. We’ve included a tutorial along with sample files for you to generate your own neural network artistically rendered images. In this tutorial, you will be able to style Jackson Pollock’s “On Arches” onto a portrait of Clint Eastwood, Darth Vader, David Bowie, and a corgi.

Performing Style Transfer on Rescale
We will now show you how to run your own style transfer job on Rescale. We will create a Torch network training job and run jcjohnson’s fast-neural-style package. This job can then be cloned to train new styles or style new images. You can clone this job on Rescale here.

To start, we upload the following files as input:

fast-neural-style.tar.gz: tarball with jcjohnson’s software downloaded from GitHub.

coco2014.h5.gz: Microsoft COCO 2014 image-in-context training dataset in hdf5 format (this repository is actually a random selection of 1/10 of the original images in COCO 2014).

clint.jpg: image we will style

pollock.jpg: image (artwork) used to style image

vgg16.t7.gz: trained VGG16 model in Torch’s t7 format

Next, we specify that we will be running Torch and which script to run. We first run train.lua. to build the style transfer model using VGG16 as the base model, COCO 2014 as the training dataset, and pollock.jpg as the style. Once the style transfer model, model.t7 is created, we call fast_neural_style.lua to style clint.jpg. After styling is complete, we clean up files we do not need to keep in Rescale storage.


We are almost ready to run. Now, select the Obsidian core type, 2 CPU cores (which corresponds to 1 K80 GPU), then hit Submit.


Once our cluster starts and our inputs are synced to the cluster, the initial style model training begins and we see a view like below: 


After training and image styling completes, the result files are synced back and we can view our styled Clint Eastwood:


Notice that we also have model.t7 in our output files. This is the Pollock style transfer neural network which we use to style further images. Let’s first make this model file available as an input file:


Now we create a new job with the same code repository, our new model file, and some more images to style. You can clone this job on Rescale here.


This time, we have uploaded our images as an archive, which will get unpacked after upload and we run fast-neural-style on the resulting input directory. Note how we no longer need to run the more compute-intensive training process.


Once the training completes, the styled images appear in output_imgs in the results:


*Special thanks to Mark Whitney for setting up the technical and instructional aspects for this blog post. 

This article was written by Daphne Su.


We recently updated the design of 1) our pricing page and 2) the hardware selection page in order to offer more visibility on pricing, as well as to create a smoother user experience when users are setting up jobs.

Pricing Page

Our pricing page now includes a detailed, dynamic chart where you can filter for hardware specs and pricing.



For example, if you wanted to compare Prepaid pricing and Instant/On-Demand pricing for Windows, while referencing Memory and Processor info for the cores, you can select only those four columns in the filter and toggle between Linux and Windows options on the top right.  You can also download the chart as a PDF directly onto our Rescale template to use as a hard-copy handout or to email to colleagues.


To access the pricing page: Log into your account, click the dropdown at your email address at the very top right corner, click Account, and then click Pricing on the left sidebar.

Hardware Selection Page

The hardware selection page, which you will encounter when setting up a job, has undergone a minor restructuring, along with added pricing options.  We moved the hardware summary to the side navigation as a fixed element so you can keep track of your settings throughout the job setup process and so that it doesn’t get hidden on smaller screens.  You can also easily compare Prepaid, Instant/On-Demand, and Low Priority pricing and availability side by side while setting up your job.


Below the hardware selection chart, we changed the core number sliders (and for DOE jobs, the task slot sliders as well) into a simple number entry field to be more streamlined and to better accommodate large numbers of cores.  Just type in the number of cores you want or use the + – button to view the next appropriate number of cores above or below the number you input.  For example, if you entered 5 cores but the core type you selected only offers 4 or 8, then the + and – button will automatically skip to the next suitable number of cores.


At Rescale, we follow a users first approach to our product and design, not just by incorporating data from user-focused research, but by leveraging feedback from existing team activities as well.  Compiling information from many channels gives us a holistic and many-angled view on our users’ behavior and needs.

These include:

  • User feedback through support channels: Our support team goes over user issues with our whole dev and design team on a weekly basis.
  • Sales and marketing bringing back info from talking to customers and prospective customers on their needs and workflows.  
  • User testing and customer product surveys to pinpoint user challenges and keep up-to-date on customer satisfaction.

Stay tuned for more improvements, or send your feedback to us at  

This article was written by Daphne Su.