MikeJ opened this issue on Oct 26, 2000 ยท 9 posts
MikeJ posted Thu, 26 October 2000 at 11:22 AM
karlm posted Thu, 26 October 2000 at 4:19 PM
Ahh, very interesting. As far as I can tell (and with some testing) this effect actually has nothing to do with the mapping of the brick material. What you are seeing is due to the fact that the bricks have a sharp repetitive pattern (the dark edges). When you have this perspective where the far away part has the pattern looks so small, the resolution of the image is not fine enough to properly represent the high frequency of this pattern (say thanks to Nyquist - okay, don't shoot the messenger). When the resolution (sampling) is not high enough, a new distorted pattern will emerge....this is essentially a form of aliasing. Okay, now on to the more useful stuff. There's no perfect way to fix it, but the following things will help (also in combination) to varying degrees: 1. Render in a better mode (say Ultra). This supersamples and then antialiases based on this information. It will smooth out the image and suppress this circular pattern. (this is probably the most effective option) 2. Adjust the perspective so you don't see the bricks from so far (i.e. the brick pattern is not so small). 3. Render at a higher resolution (probably at least twice) and then use a program like Photoshop or something to resize it back down using a good scheme like bicubic resampling. This probably has no advantage over option #1 (effectively very similar), so I'd probably just render in Ultra. 4. Do something similar to what the eye would do...add a little focal blur. If you watch the render on screen, since the mosaic process essentially increases the resolution by two at each step, you will actually see the circular pattern start closer to you (in the perspective), and then move further away as the number of render passes increases. On an aside, I find this all very fascinating because I'm studying digital signal processing right now. My undergraduate thesis pertains to wavelet signal analysis with application to image compression in conjunction with models of the human visual system. Long winded, but I hope it helps. -Karl
bloodsong posted Thu, 26 October 2000 at 5:28 PM
umm... build your room so it isn't as big as the world? :) no, really, use a cube for the floor and make it finite, there bud. :) oh, are the bricks on a diagonal because you're using the default camera location? i saved a default file where i have the camera set square to the world, not on an angle, so i'm not used to things being skewed. (makes it MUCH easier to rotate things in view!)
karlm posted Thu, 26 October 2000 at 5:50 PM
If he wants the scene to essentially look the way it does now (without the distortion), then mapping to a finite cube will not help...he'll still have to scale it to the size of the room floor.
MikeJ posted Fri, 27 October 2000 at 2:36 AM
black-canary posted Sun, 05 November 2000 at 11:03 PM
karlm posted Sun, 05 November 2000 at 11:35 PM
While rescaling the map might help/serve the purpose, I think my point was lost. That is, if you want to keep the scale the way it was and have a shallow viewing angle, you will have to resort to the techniques I described above to get rid of the circles.
black-canary posted Mon, 06 November 2000 at 7:04 AM
oooh, I get it...you already DID scale the material to get it to look right...and using a different scale doesn't work...hm. If karl's ideas don't work for you, can you change the scale of the function instead of the material? Or capture a chunk of the brick material as a map and make it a map-based material instead of a function one? And who puts a brick floor in a poured concrete building anyway? ;) MaryCanary
bloodsong posted Mon, 06 November 2000 at 5:31 PM
heyas; that brick texture is a map, not a function. (as i recall.)