United3dArtists Archive

In Blender you have the choice between several Antialias methods. The range goes from Gaussian, Mitchell Netavalli and Catmull Rom up to cubic, quadratic, box and tent.

In trueSpace there`s only the choice to activate Antialias at all. What Antialias method is in use in trueSpace? Does somebody know that?

Method as Gaussian and other are made to speed up the rendering (not a real supersampling but an algorythm to calculate the antialiasing).
TS use only :
supersampling: This process means calculating a virtual image at a higher spatial resolution than the frame store resolution and then averaging down to the final resolution. the method results in a lot of computations but the quality is the best.
Adaptative sampling: the subdivision can be repeated to an arbitrary level only when the render engine think it is necessary. Adaptive sampling works very well on boundary edges. In certain scenes aliased artifacts can occur in the final image.

supersampling 2x is enought for most of render. It need about 4 time more to render the image.
supersampling 3x and 4x are used to avoid moire effect on certain scene.

if you render a picture in hd definition and no AA then reduce the size in a photo editor (50% with interpolation) you have the same result than to render a half image with AAx2. That show what TS do with AA.

It's sometimes called 'The Roman Method'.

There is also the feature following antialiasing on model side. It is on the flyout with glows/lens flares. That's all I know about it -- it exists.

Thanks guys. That gives me more questions than answers, hehe. But i think i understand it a bit better now

I`m interested in which method in trueSpace is comparable with the methods in Blender. Thus my question at all. But having a look at the render results shows me that it must be one of the three methods called catmull rom, box or tent

Finis wrote:There is also the feature following antialiasing on model side. It is on the flyout with glows/lens flares. That's all I know about it -- it exists.

its a pretty cool feature. If you have a tiny line in a render and AA will cause that it doesnt look as one line anymore, the FFAA can connect the line to a smooth nice antialiased line. try it ;)

For a still image i prefer to render an hd picture (more than i need) then reduce it in a photo editor with a factor of 50% or 25% (similar to a AA2x or 4x). Some algorythm on photo editor are more powerful and i can test the best without the need to render again my image. Why? Because if i need to change the contrast, it is better to do that before reduce the image to enhance the edge on a contrasted and small image.

for animation, it is different: AA is interesting to avoid animated artefact or moire when a detailed texture move. Especially if the texture is bigger than it appear on the render. If the texture is smaller than the render, i combine AA (for the edge) with filter option ( to smooth the texture). instead of AA i use also blur motion. In fact, the method used depend of the kind of work i do.

Maybe it is possible to find out the method of TS, if you make the same line in TS and Blender.
Then render to a BMP-File, load it in your image view program then zoom until you can see the pixels.
Now you can compare the different methods. (maybe. It is just an idea )