Image-space Occlusion as a Post-process Effect (2009)
Uploader Comments (gtnconcept)
All Comments (7)
-
It looks really cool!!! Nice work!!
-
Even though it's not entirely correct in both physics and standard approximated lighting equations (aka FAKE), i feel u have done a nice exploration on creating perceptively lovely CG images.
Undoubtedly, this technique has many limitations in practice, but overall it's usable in certain cases and therefore it could be nice for other AE users to learn about.
-
yes, but something like radiosity or lightcache or irradiance mapping or pointcloud lighting is faked GI.
This isnt. Its not because there is the usage of the word 'faked' that there doesnt need to be some level of correctness.
Hell, following your reasoning i could draw a fucking line in photoshop and call it faked GI.
-
The track is 5:55am from the album "Otherwake" by Chris Harvey
you can see more of his stuff on myspace - look for chrisharveyuk
you clearly doint know what gi is.
Capeau 2 years ago
Global illumination includes (but is not limited to) ambient shadows, occluded shadows and diffuse color inter-reflection.
I readily admit that the effect is not accurate or complete in my description, but the video does show these 3 features approximated in post, so how is it not *fake* global illumination"?
gtnconcept 2 years ago
because global illumination is just a fancy word for secondary illumination or bounced of light. In none of the renderpasses you use there is info where the bounced of light is going to or even comming from so results arent correct at all and it shows to be honnest.
The fact that theres a few extra blurry passes on top doesnt make it GI.
Try it on something else then a cornell box and it will look totally incorrect.
Capeau 2 years ago
The video title clearly says: *fake* and *post-process*. This is an *experiment* to fake specific aesthetic qualities using Sketchup output (single light source, direct illumination).
It is a simplified "image-space occlusion", similar to SSAO in real-time (games). The depth pass gives the necessary z-buffer information. The split-screen shows no edges in the final render are blurred.
I welcome any input on how this effect could be improved while still only using direct illumination renders.
gtnconcept 2 years ago
Mmmmm interesting... whats the fps difference????
StatusQuo209 2 years ago
The post-process effect is created in After Effects and is not real-time. This test is more to show how the effect can be added after the image has been rendered for software that does not support advanced lighting models.
I do know that Crysis currently does a similar thing in realtime, by using the depth buffer information in place of my crude fogged-based depth map for occlusion calculations.
gtnconcept 2 years ago