ðòïåëôù 


  áòèé÷ 


Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 

  óôáôøé 


  ðåòóïîáìøîïå 


  ðòïçòáííù 



ðéûéôå
ðéóøíá












     áòèé÷ :: Filmscanners
Filmscanners mailing list archive (filmscanners@halftone.co.uk)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[filmscanners] Blur or sharpen before downsampling? (was: PS sharpening)



On 8/23/02 2:28 AM, "Michael O'Connor" <omichael@optonline.net> wrote:

> Julian writes:
>
>> Are there specific circumstances where the multistepmethod shows its
>> advantage?
>
> I saw advantages to multistep when I was making thumbnails for the
> web, but they weren't my images at the time, and it involved images that
> needed to be scaled down dimensionally as well as downsampling
> resolution, and I think that's an important point. At simple massive
> downsampling I would think that very finely detailed images, where the
> detail must hold for the intent of the image to be conveyed, and (not
> that this is likely to be a concern for you), bitmaps of any kind, would
> probably benefit as well, though in these cases I really think you'd
> have to adjust your settings as you went along on an image specific
> basis.

I have not yet tried the multistep sharpening on such massively reduced
images.  It would be interesting to see if the multistep method works better
than simply reducing in one step and then intentionally oversharpening the
result.

Earlier in this thread (or elsewhere) someone offered an explanation for the
alleged advantage of multistep sharpening: if you apply a slight
oversharpening before each downsampling operation, you preserve details that
might be lost when pixels are tossed.  But then I read a note from Kennedy
McEwen on another list, where he explained that for a theoretically
optimized downsampling you should apply a blur before the downsampling, to
reduce the effects of aliasing.

How do we reconcile this?


n. b.:  The blur prescribed by Kennedy was small.  If gaussian blur is used,
the params are:

radius = (0.3 x S)/(S - E), where S is start res, and E is end res.

For example, r = 0.6 pixels before downsampling from 4000 to 2000ppi

--
Julian Vrieslander <julianv@mindspring.com>

----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe 
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or 
body



 




Copyright © Lexa Software, 1996-2009.