Deep Learning and Computer Graphics

 Our previous mini review of some selected Siggraph papers from this years Siggraph conference really brought home the message that AI research in deep learning is really driving new developments in computer graphics rendering.


We going to continue that theme in this post by checking out 2 Nvidia podcasts on this topic. Augmenting the generation of computer graphics using deep learning networks.


The first is a talk with Nvidia’s David Luebke, who is vice president of graphics research at Nvidia.  David talks about a lot of fascinating state of the art topics, including how deep learning is speeding up the real time render pipeline for generating computer graphics, the use of GAns (generative adversarial networks) to directly synthesize graphical imagery, and augmented reality displays, including using deep learning to help build holographic displays for AR.

You can access the podcast here.


The second is a talk with Nvidia’s Arron Lefohn.  Arron continues the discussion on how deep learning is revolutionizing computer graphics.  More info on GANs here, as well as other aspects of the computer graphics pipeline that can be enhanced or rethought using deep learning..

You can access the podcast here.


After listening to the 2 talks, I hope you can feel some of the excitement about new developments in this field.  It’s kind of astonishing how much and how quickly things are changing.


One big take home message is the work being done on using GANs (generative adversarial networks) to directly synthesize computer graphics imagery.  This is a rapidly changing research area


David in his talk specifically mentions this, and discusses it in some detail for a few minutes of his talk. The notion that you can explore the latent semantic space of the GAN to help understand what the GAN is actually representing is fascinating.

You may remember that we discussed latent semantic maps in an earlier post. 

If you step back from a deep learning neural network and think about what is is actually doing, it’s building a multi-dimensional transformational mapping.  The latent semantic space is a way of representing that non-linear multi-dimensional mapping manifold. Of exploring what the network is representing within different parts of the multi-dimensional mapping.


Nvidia’s GAn researchers (and others) have found that in addition to the appearance of objects within a photo, a GAN may also be building representation of lighting and rotation properties of these objects.








Comments

Post a Comment

Popular posts from this blog

CycleGAN: a GAN architecture for learning unpaired image to image transformations

Image
 Like the Doublemint Twins touting the joys of Doublemint gum, 2 GANs are surely better than 1 GAN.  Especially if we package them together inside of one meta GAN module.  And this is exactly what the CycleGAN architecture does. Have you ever harbored dark secrets of turning a horse into a zebra?  The CycleGAN was developed to do just that. Learn how to turn a horse into a zebra.  And more. Now right away you can notice a difference between the image to image transformation GAN architectures we've been discussing over the last few posts.  Those last few posts described systems that learn from a database of matched input-output image pairs. And if your goal is to turn an edge representation into a nicely filled in continuous tone image, it's easy to build your database of matched input-output image pairs that your GAN system can then learn off of.  Take a continuous tone image (which will be the output of the database pair entry), then run it through an edge detector algorithm. 
Read more

Pix2Pix: a GAN architecture for image to image transformation

Image
 I thought following up yesterday's TraVelGAN post with a Pix2Pix GAN post would be useful to compare what is going on in the 2 architectures.  Two different approaches to the same problem. I stole this Pix2Pix Overview slide below from an excellent deeplearning.ai GAN course (note that they borrowed it from the original paper) because it gives you a good feel for what is going on inside of the Pix2Pix architecture.    Note how the Generator part is very much like an auto-encoder architecture, but rebuilt using the U-Net architecture features (based on skip-connections) that fastai has been discussing in their courses for several years before it became more widely known to the deep learning community at large  (and which originally came from an obscure medical image segmentation paper) . So the Generator in this Pix2Pix GAN is really pretty sophisticated, consisting of a whole image to image auto-encoder network with U-Net skip connections to generate better image quality at highe
Read more

Smart Fabrics

Image
"What makes smart fabrics revolutionary is that they have the ability to do many things that traditional fabrics cannot, including communicate, transform, conduct energy and even grow."  This is a quote from Rebeccah Pailes-Friedman, who is a professor at Pratt with a research focus on developing wearable technology and smart textiles. Enhancing textiles can be viewed through different lens.  Enhancements can be aesthetic, or performance boosting, or durability boosting. 
Read more