Downsampling for web and its complexities

[Chrisbie]

I reduce my pictures down to typically something in the region of 1000 to 1500 pixels wide for web use but take pictures at as high a resolution as my camera can take for the possibility of printing later.
with a Camera that takes pictures at twice this width in pixels the maths needs to reduce it in size requires a reduction based upon the square of the ratio such as 2 squared or 4. ( a reduction of say 2056 wide to 1028)
Now with higher megapixels producing 6000 plus wide pictures we are talking reductions of a ration of 6 squared or 36 (if my maths is correct).

This raises some questions
Is the current typical software capable of this sort of downsampling or do we need to look at software with more complex algorithms to do this such as Savitsky Golay sampling.

Is downsampling better done in one step or should we do a number of reductions.

would we be better off with less pixels and upsampling ??

and then a customer wants a print at A1 size.......

 

[Photodiva]

Dunno about the math but upsampling is never a good idea. Web display is never consistent with different monitors and diff resolutions. Just downsize, sharpen and post pn the web.

 

[swanseamale47]

I doubt you'll get a better result using steps to downsample, personally I'd not worry about it so much and resize to the final output size using bicubic sharper (in photoshop) then sharpen for output as needed.
I don't quite follow the upsampling bit is that to get big prints or are you suggesting starting with a small image and making it bigger to say 1000 pixels?

 

[AJQS]

Simple: Bicubic and global sharpen for resizing down, bilinear for increasing for printing (followed by selective HPS and softening where needed)

 

[Chrisbie]

thanks guys
that serves me right for looking at it from a mathematicians point of view