• Advertisement
Sign in to follow this  

Pixel samples.

This topic is 681 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I have ported SSAO from Nvidia’s DeinterleavedTexturing example.

In this example SSAO is generated for each pixel’s sample.

I can see quality improvements in SSAO from 1 sample to 4 samples. And it worth it.

But for 8 samples, performance is very poor. Also I can’t see quality improvements on an image.

So I want to reduce SSAO to max 4 samples, even GBuffer is 8-sampled.

The only thing to understand is: how map 8-sampled pixel to 4-sampled, and 2-sampled?


Question 1: which samples from 8-sampled GBuffer should I take to get best image?

Should I take {0,1,2,3}/{0,1} or {0,2,4,6}/{0,4} samples, or some other order?

(I would like to avoid picking samples from only one half of a pixel)


Question 2: the sample mapping order between Intel, AMD, and NVidia is the same?


Question 3: for vendor-specific “quality” modes like CSAA, would this mapping be the same or I should implement for each vendor different algorithm of samples picking?


Question 4: (I don’t have access to new Intel hardware): does Intel has something like CSAA or just plain MSAA?

In other words, does Intel use DXGI_SAMPLE_DESC::Quality at all?


Thanks in advance!

Edited by Happy SDE

Share this post

Link to post
Share on other sites
Sign in to follow this  

  • Advertisement