Which resize method ?

[Rainy]

I have a DTV capture which has a resolution of 480x576, I would like to resize it to 512x384 and encode to xvid, which resize method should I use: bilinear, bicubic neutral, soft-, simple resize ?

Thanks in advance

[int 21h]

From VirtualDub help file
· Nearest neighbor, or point sampling, picks the closest pixel in the source image. This is by far the fastest mode but also looks lousy, especially if the new dimensions are not integral multiples or divisions of the old ones. In this mode, VirtualDub basically drops and duplicates pixels.

· Bilinear uses a linear approximation in both directions using a 2x2 kernel. This is the second fastest mode, and produces very good results. It can reduce down to about 66% and enlarge up to about 8x without visible stairstepping. Most video cards with smooth video overlays have bilinear filtering in hardware. Bilinear filtering tends to produce a slightly blurry image and at very high magnification results in trapezoidal patterns. Because this mode only uses a 2x2 kernel, at high reduction it results in pixellation much like the nearest neighbor mode.


· Bicubic is the best choice for enlarging a picture, producing slightly sharper images at moderate levels and curves instead of trapezoids at high zooms. It is significantly slower than bilinear but produces much better at high magnfications such as 20x. Because bicubic mode treats detail better, very noisy video can appear worse under bicubic filtering than bilinear. Because bicubic mode only uses a 4x4 kernel, it too results in pixellation at high reduction.


· Precise bilinear uses a full bilinear filtering kernel rather than a 2x2 and works for both reducing and enlarging pictures. When enlarging, this mode works similarly to bilinear, but when shrinking this filter averages many pixels together. This results in a much more faithful image and effectively squashes noise, even better than the 2:1 filter. Unfortunately, this filter mode is also slightly slower than bilinear mode when enlarging and signfiicantly slower when shrinking.


· Precise bicubic uses a full bicubic filtering kernel rather than a 4x4. It works very similarly to precise bilinear, but the bicubic kernel produces sharper images, when either shrinking or enlarging. Shrinking a noisy 320x240 video with this filter can result in a sharp, clean 160x120 video. This filter is also the slowest of them all.


SimpleResize differs from these in that it has no noise filtering properties.

AFAIK, the Avisynth filters are very very similar to the Virtualdub filters. Its generally agreed however that 0.65 is a pleasant comprimise between sharpness and softness. However, your mileage may vary, and the best advice that can be offered is try them and find out which you like best.

[DJ Bobo]

Better resize to 480x352. Here you'll be resizing in only one direction, that's why you won't need more than Bilinear Resizing.

For any other size, better go for Bicubic Resizing.

[Manao]

Why do you want to increase the horizontal size ? Keep 480 * YYY

And you could also try lanczos3reisze, since you have a little area, the loss of compressibility will come with a slightly better quality.

[Edit:] @ DJ Bobo : with only one direction of resizing, all methods are the same ?

[DJ Bobo]

@ Manao
Almost, that's why one can take bilinear (there won't be a great difference in sharpness, the differences will be minor and very hard to distinguish)