Interactive Graphics 23 - Ambient Occlusion & Soft Shadows

Interactive Computer Graphics.
School of Computing, University of Utah.
Full Playlist: • Interactive Computer G...
Course website: graphics.cs.utah.edu/courses/...

[@mohitsethi99]

It was really well explained, infact thank you for the whole series.

[@alex-shortt]

thank you for making these public!!

[@yoyonel1808]

Other solutions/researchs on this topic: - Voxelisation [sort of Raytrace] in an octree in realtime (full GPU support), very good/interesting work done by Cyril Crassin (working in NVidia Research) - Soft Shadow Volume : geometric extension of Shadow Volume for soft shadows, performed initially by Ulf Assarsson very nice paper on this topic - Soft Shadow Maps (Efficient Sampling of Light Source Visibility) : (one proposed) extension of Shadow Map for soft shadows, proposed initially by myself ^^ A geometrical extension of pixel from shadow mapping considering/transforming in this method into "micro-facet" (aligned with light camera view), this transformation help to transform pixel/point shadow to micro-face/quad shadow 🙂 - Temporal shadows : using the historic of shadow rendering to increase the precision/resolution with temporal "smart" blending frames after frames. Very interesting topic ! - and so many others ^^ Hard topic, it's a good try to introduce the problematic !

[@kubatpav]

Thank you. Great video!

[@fatemermg7490]

You are awesome!! Thank you for uploading all these great lectures.

[@caiyueli2235]

love all your lessons, they are pretty helpful for me ! thank you professor !

[@skyheart_dev]

I like the way you explain this. Subscribed. It would be great to see more videos on simulating softbodies/rigidbodies - more softbodies because they are cool :)

[@DeanEdis]

Such a cool explanation! I create raymarched scenes a lot, where the signed distance field can be used to quickly generate an AO approximation. So effective!

[@mobilemusic1633]

You are SO epic. Gonna do my Masters in Utah just to take your course!

[@PaulEmsley]

12:37 "an example of something" - it's the large ribosomal subunit of an Archaeon. Tom Steitz was the PI for this work and he won a Nobel Prize for it.

[@cem_yuksel]

I thought that it was a cool example, but I had no idea about its significance. Thanks for letting me know!

[@PaulEmsley]

@@cem_yuksel Thank you for your lectures. Along with the Cherno, ThinMatrix and the BennyBox you have helped me write my own molecular graphics program. When I post a video demoing the SSAO, I'll edit this comment so you can see it.

[@Koffiato]

Çok teşekkür ederim, harika açıklamışsınız hocam

[@emreerkan94]

Hocam harika anlatmışsınız

[@chaver28]

very informal video :D you got my sub

[@thominio]

1:04:10 unfortuantly the KIT is focused on soft shadows it goes over multiple sessions just over this topic ...

[@motbus3]

thanks for the classes :)

[@jordanz6592]

Amazing content! struggling to find freely available content on this topic and specifically the equations, is there any more resources potentially available for the public? Im doing a small study on Ambient Occlusion and would just like to have more information about the function and possibly some reliable sources

[@cem_yuksel]

This is a good start: en.wikipedia.org/wiki/Ambient_occlusion

[@32zim32]

Don't get it how you replace part of equation which depends of light direction just by Ka constant

[@augustvctjuh8423]

SSAO sounds like a very nice technique. Would definitely add if I wrote a realtime renderer But imagine if the windows were all blocked off in that classroom (7:24). No light should enter the scene. It might be interesting to take a few thousand samples scattered throughout the viewport to get an average for ambient light, before you do AO. Then you can scale your AO contributions based on it. Or maybe more sophisticated methods already exist :)

[@kubatpav]

Hello Cem. What do you think, does it make sense to keep AO texture separate from diffuse texture or there are pretty much no benefits so pre-baking it to diffuse texture is absolutely fine? (Unity engine)

[@cem_yuksel]

Illumination, which affects diffuse reflections varies, varies with moving lights/objects. Ambient illumination can be assumed constant, but the diffuse won't be. I would not combine/pre-bake them.

[@kubatpav]

@@cem_yuksel Thank you for reply Cem.

[@ostrich_dog]

about soft shadows: what if, instead of sampling multiple pixels from the shadow map, i used mip-maps to apply different levels of occlusion? could something like that work?

[@cem_yuksel]

Perhaps. My student Yang Shen and I worked on this idea some years ago. We could solve many of its problems, but the results were not satisfactory for reasons that would take too long to explain. If you are curious, you can read her thesis: collections.lib.utah.edu/details?id=197818

[@ostrich_dog]

@@cem_yuksel thank you very much for your answer

[@pokojopl8759]

Honey, I've been following you for a long time and I was wondering if you'd be interested in working on a project at my company