Can AI SuperPower enhance my artistic skills? – Make photo-realistic images with NVIDIA RTX technologies, NVIDIA Canvas and IGEL OS

By Fredrik Brattstig @virtualbrat

AI has been around for quite a while. AI is a powerful thing, but it has been like a scientific thing, until now 🙂
NVIDIA has released a beta software called NVIDIA Canvas and I have played with the software in my lab a bit. To be very upfront, my artistic skills are a bit limited so I’m very thankful to get some AI SuperPower to enhance the outcome! Here is a sample screenshot of my work:

Left: Me – Right: Me with AI SuperPower

A quick look at what I’ve set up:
– IGEL OS endpoint
-Virtual Windows 10 machine powered by NVIDIA A16 GPU with NVIDIA Canvas + Citrix VDA

Why am I using the NVIDIA A16? NVIDIA Canvas is using the RTX GPUs to process the AI, which means that you will need to use an RTX-enabled GPU to benefit from the AI super power. The NVIDIA A16 is a GPU that I really like, as it contains four GPU’s on one single GPU board. Each of the GPUs on the board has its own reserved 16GB of frame buffer, 64 GB in total. This allows me to segment the GPU for my virtual machines more dynamically. I can have multiple 1GB frame buffer (Graphics RAM) VMs combined with multiple VMs with 2GB frame buffer and also combined with multiple VMs with 4GB frame buffer, and yes, you guessed it, even VMs with 8GB frame buffer. And still using one single physical GPU board! I’ll have a blog coming up shortly about the NVIDIA A16 GPU board, stay tuned for that.

The NVIDIA Canvas application comes with a set of “materials” which you draw on a canvas, the AI that is built into the software and executed by the RTX-enabled graphics card will then translate the simple drawing into a photorealistic image.
Nvidia Canvas seems to be requiring more than 2 GB frame buffer, so I had to connect to a VM with 4 GB frame buffer to be able to run the application in session.

For the IGEL OS endpoint, I basically just enable Hardware Accelerated Citrix HDX h264 in the session configuration, as that will give me a “better than local” user experience.

Now with AI enhancing the average user to create a photorealistic image, I believe that we can agree that even if it still is in semi- early stages of adding AI to consumers, the avalanche has just begun..

I added a video of where I use an IGEL OS endpoint connecting to my Citrix VM with an NVIDIA A16 4GB profile and using NVIDIA Canvas. Play the video below

Thank you for your attention. AI for everyone!


Read more about NVIDIA Canvas here:
Read more about NVIDIA 16 here:

Read more about IGEL here:

Last but not least, more about Citrix here: