tag:blogger.com,1999:blog-6829904489614710740.post2894681793003001760..comments2023-04-05T17:57:40.716+08:00Comments on My Pseudocode Life: Light Prepass Cascaded Shadow Mapping, continuedJohn David Bonifacio Uyhttp://www.blogger.com/profile/04108888915485048942noreply@blogger.comBlogger6125tag:blogger.com,1999:blog-6829904489614710740.post-46484696327606648682009-07-19T17:26:45.638+08:002009-07-19T17:26:45.638+08:00bd,
sorry for the late reply.
Yes, you are corre...bd,<br /><br />sorry for the late reply.<br /><br />Yes, you are correct. The first four offsetted then the 2nd for diagonal to the first.<br /><br />As you can see, it is self-Optimized. I don't know if you can, but if its possible post your results. I appreciate any feed back on this. Thanks!John David Bonifacio Uyhttps://www.blogger.com/profile/04108888915485048942noreply@blogger.comtag:blogger.com,1999:blog-6829904489614710740.post-23344816954892982272009-07-15T08:26:54.852+08:002009-07-15T08:26:54.852+08:00Argh - I meant the following as the first four sam...Argh - I meant the following as the first four samples:<br /><br />[1.5, 1.5], [2.5, 1.5], [1.5, 2.5], and finally [2.5, 2.5] (I mistakenly listed [1.5, 1.5] twice!)Anonymoushttps://www.blogger.com/profile/08767674539524387141noreply@blogger.comtag:blogger.com,1999:blog-6829904489614710740.post-34290847802021916962009-07-15T06:30:49.165+08:002009-07-15T06:30:49.165+08:00Hmmm - I'm starting to get the pixture, but le...Hmmm - I'm starting to get the pixture, but let me ask a couple more questions. Let's saw I have the following 5x5 texel shadow map:<br /><br />[0,0][1,0][2,0][3,0][4,0]<br />[0,1][1,1][2,1][3,1][4,1]<br />[0,2][1,2][2,2][3,2][4,2]<br />[0,3][1,3][2,3][3,3][4,3]<br />[0,4][1,4][2,4][3,4][4,4]<br /><br />And if I PCF sample exactly at [2,2] I would get only the center pixel's shadow value, but if I sampled at [2.5,2.5] I would get a quarter of [2,2], [3,2], [2,3], and [3,3].<br /><br />So, if I understand, you are suggesting I first take four PCF samples, at [1.5, 1.5], [2.5, 1.5], [1.5, 1.5], [1.5, 2.5], and finally [2.5, 2.5]?? <br /><br />And then, based on if I find any shadow coverage, I then add four more samples?<br /><br />Thanks!!Anonymoushttps://www.blogger.com/profile/08767674539524387141noreply@blogger.comtag:blogger.com,1999:blog-6829904489614710740.post-56302098184916374992009-07-14T10:45:46.913+08:002009-07-14T10:45:46.913+08:00O Btw, I dont shadow-test each samples. Only after...O Btw, I dont shadow-test each samples. Only after the 4 prime samples (multiply by four when shadow testing). So this is even faster than 4 Tap PCF in terms of logic tests.John David Bonifacio Uyhttps://www.blogger.com/profile/04108888915485048942noreply@blogger.comtag:blogger.com,1999:blog-6829904489614710740.post-63213572816792053912009-07-14T10:42:58.943+08:002009-07-14T10:42:58.943+08:00Hi bd,
I'll post a deeper explanation about t...Hi bd,<br /><br />I'll post a deeper explanation about this eventually.<br /><br />But the overview explanation is...<br /><br />Sample the by 1/3 of its size on 8 directions of a texel(North, South, East, West, NW, NE, SW, SE) but not decreasing its sample size.<br /><br />The O-Tap there is do the diagonal direction first(NW, NE, SW,SE) (or you could use the axis direction N, S, E, W). You do this first for your common shadow test in normal PCF. If PASS, then you do the next 4. If you notice, I ignore the center sample, it is because the the total 8 taps have enough sampling to have the center.<br /><br />Anyway, I hope to write a short paper on this. Maybe after all of my pending tasks are done.<br /><br />Cheers!John David Bonifacio Uyhttps://www.blogger.com/profile/04108888915485048942noreply@blogger.comtag:blogger.com,1999:blog-6829904489614710740.post-59273730212831971932009-07-14T09:54:18.310+08:002009-07-14T09:54:18.310+08:00Hey - it (O-tap) looks very interesting. Do you h...Hey - it (O-tap) looks very interesting. Do you have any more information on where to place the samples? I didn't quite understand your description and how it compares to regular PCF.<br /><br />Thanks!!Anonymoushttps://www.blogger.com/profile/08767674539524387141noreply@blogger.com