i adjusted the contrast to make it more easily to see the edges and pixels better.
thank you.
- Rudi
What filtering technique is used?
Author
Hello, I wonder if someone with a good eye for this kind of thing can tell me what type of filtering is used on this image zoom? (Its taken from a software rendered demo) I havent made a working implementation of trilinear-filtering and therefore I am uncertain if it is that technique, can someone please tell me if it is or what they think it might be? the image can be seen here:
i adjusted the contrast to make it more easily to see the edges and pixels better.
thank you.
- Rudi
i adjusted the contrast to make it more easily to see the edges and pixels better.
thank you.
- Rudi
It doesn't look like filtering..
More like a fractal pattern of some sort. It's hard to tell because we don't have the original image. You can make any image from any other image so really - one image doesn't say much about the algorithm.
More like a fractal pattern of some sort. It's hard to tell because we don't have the original image. You can make any image from any other image so really - one image doesn't say much about the algorithm.
Looks like a bug. As if someone got the bilinear filtering weights backwards.
That, or it is supposed to be some kind of ugly dithering effect.
No, it isn't. A bilinear filtered image (or trilinear if volumes or mipmaps are involved) is supposed to be entirely smooth without jaggies. That's the whole point of the filter. Unless you accidentally have your weights inverted, then you can get a picture somewhat similar to that one above.
That, or it is supposed to be some kind of ugly dithering effect.
Quote:
havent made a working implementation of trilinear-filtering and therefore I am uncertain if it is that technique,
No, it isn't. A bilinear filtered image (or trilinear if volumes or mipmaps are involved) is supposed to be entirely smooth without jaggies. That's the whole point of the filter. Unless you accidentally have your weights inverted, then you can get a picture somewhat similar to that one above.
Author
I see.
If is may help i've made two more screenshots, one with the original and the other with adjusted contrast and brightness. since the original image is quite dark i had to just adjust the pixels, it doesnt change how you can see the algorithm at work (whatever algorithm that is though).
if you want to see the demo in realtime with this technique used then you can download it here.
the demo is from 1999, so its quite old. but it looked so damn nice at that time, still is kind of. the resolution is low, but with that technique the zoom is just still amazing to look at.
if you have no idea what it is, then thanks for your time anyway.
- Rudi
If is may help i've made two more screenshots, one with the original and the other with adjusted contrast and brightness. since the original image is quite dark i had to just adjust the pixels, it doesnt change how you can see the algorithm at work (whatever algorithm that is though).
if you want to see the demo in realtime with this technique used then you can download it here.
the demo is from 1999, so its quite old. but it looked so damn nice at that time, still is kind of. the resolution is low, but with that technique the zoom is just still amazing to look at.
if you have no idea what it is, then thanks for your time anyway.
- Rudi
Author
Quote:Original post by Yann L
Looks like a bug. As if someone got the bilinear filtering weights backwards.
That, or it is supposed to be some kind of ugly dithering effect.
I see. well the bug or the dithering looks kinda cool when zoomed in. :P
They're probably dithering. Also notice the heavy banding and discolourations, due to additive blending in 15 or 16 bit colour mode. Those are not voluntary effects, they're visual artifacts due to the limitations of the hardware at that time.
Author
Quote:Original post by Yann L
They're probably dithering. Also notice the heavy banding and discolourations, due to additive blending in 15 or 16 bit colour mode. Those are not voluntary effects, they're visual artifacts due to the limitations of the hardware at that time.
so you're telling me that the demo runs in 15 or 16 bit colour mode? that's cool. the demo has lots of colours. maybe they used a palette in 16-bit colour mode, so you are probably right. i didnt imagine that!
have you any idea what kind of dithering algorithm it is, or is it too hard to tell?
never mind. i think i got it now.
thanks for the help.
That could never be a kind of filtering. Notice the triange pattern! Inverted bilinear filtering would give quads.
This image shows it better:
Looks like 3 blended fractals with different colors to me.
This image shows it better:
Looks like 3 blended fractals with different colors to me.
Author
Quote:Original post by snake5
That could never be a kind of filtering. Notice the triange pattern! Inverted bilinear filtering would give quads.
i think its probably bilinear filtering and artefacts due to a 15 or 16 bit colour mode like Yann L said. i think it causes the triangles because of dithering caused by the pallette. else its some subdivision scheme that uses some weirdo trick i guess.
That is neither a bug nor dithering. If it would actually be a bug, you'd see it on all the triangles drawn in the demo, the same goes with dithering. And fractals actually are a kind of "subdivision scheme".
Here are some examplesGoogle: "fractals triangle"
Here are some examplesGoogle: "fractals triangle"
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
