Real-Time Hair Rendering With Deep Opacity Maps | Two Minute Papers #171

TL;DR

A new technique called Deep Opacity Maps simplifies the computation of self-shadowing effects in hair and fur, achieving higher quality results with only 3 layers and running in real time.

Transcript

Dear Fellow Scholars, this is Two Minute Papers with Károly Zsolnai-Fehér. In earlier episodes, we've seen plenty of video footage about hair simulations and rendering. And today we're going to look at a cool new technique that produces self-shadowing effects for hair and fur. In this image pair, you can see this drastic difference that shows how p... Read More

Key Insights

• 🤳 Self-shadowing effects greatly enhance the visual appearance of hair and fur.
• 🤳 Computing self-shadowing effects for dense hair geometry is computationally expensive.
• ✈️ Previous techniques used multiple planes to simplify the problem but had limitations.
• 🙂 Deep Opacity Maps chooses layers more wisely based on the distance from the light source, improving quality and efficiency.
• 🏃 Deep Opacity Maps runs in real time and is more memory efficient compared to previous techniques.
• 👶 The new technique eliminates layering artifacts and provides more accurate and realistic images.
• 👻 Deep Opacity Maps allows for better occlusion information and adapts to the specific scene being rendered.

Questions & Answers

Q: Why is computing self-shadowing effects for hair and fur costly?

Computing self-shadowing effects requires determining how each hair occludes the others, which becomes extremely time-consuming when dealing with dense geometry and hundreds of thousands of hair strands.

Q: How did previous techniques simplify the problem of computing self-shadowing effects?

Previous techniques used a few planes that cut the hair volume into layers, making it easier to work with. However, using a higher number of layers like 128 resulted in better quality but slower processing and unrealistic images.

Q: What is the key idea behind Deep Opacity Maps?

Deep Opacity Maps records the distance of different parts of the hair geometry from the light source. Layers are then created based on this distance, adapting to the scene and providing more useful occlusion information.

Q: How does Deep Opacity Maps compare to previous techniques?

Deep Opacity Maps achieves higher quality results with only 3 layers, runs in real time, and is more memory efficient compared to previous techniques. It eliminates layering artifacts and produces more accurate images.

Summary & Key Takeaways

• Self-shadowing effects in hair and fur significantly enhance the visual appearance.

• Computing these effects is costly due to the dense geometry and occlusion calculation.

• Previous techniques used multiple planes to simplify the problem, but resulted in unrealistic images and limited processing speed.

• Deep Opacity Maps chooses layers more wisely, adapting to the scene and providing more useful occlusion information, resulting in higher quality results with just 3 layers in real time.