|
|
Posted by Smarty on 10/05/13 11:38
Specs,
It is hair splitting, but the definition of "resolution" has been used by
scientists for over a century and very precise meaning which was / is agreed
upon. "Resolution" (apparently defined initially by a physicist named
Focault) is simply "the number of line-space pairs per unit of linear
dimension in the plane of interest which are just separable as discrete
lines and spaces". Over the years some people have taken to simplifying the
"line-space pairs" as "lines", and thus resolution is expressed as lines per
inch, millimeter, or any other dimensional unit. There are a set of
derivative terms measuring sharpness, acutance, 'detail', etc. which have
their own meanings and correct usages.
Photographers, astronomers, vision specialists, and others who measure
resolution have an agreed-upon definition, and tools and techniques
supporting this method and nomenclature.
The world of video has (like the world of audio) invented a marketing
vocabulary which becomes the basis of my nit-picking. Since it is compelling
for equipment to offer 'higher resolution' performance to make it more
attractive, the fact that the ultimate delivered video or audio is not
resolving a higher number of line pairs per millimeter despite being touted
as 'higher resolution' seems to me like the worst type of obfuscation.
I agree entirely with your observation that time averaging and other
correlation methods do improve resolution, and indeed such is the very
premise upon which radio astrononomy and optical astronomy builds long
integration periods to resolve more detail. In the case of video, however,
the only opportunity for time integration is over a frame, and then only if
a stationary image is captured.
Smarty
"Specs" <No.Spam@Thanks.com> wrote in message
news:43dd0fdb$0$82633$ed2619ec@ptn-nntp-reader03.plus.net...
> One has to remember information is stored temporally and as such
> information
> can be rebuilt by using the data that is stored between frames/fields.
> When
> up-rezzing video for film its not simply a case of resizing and
> anti-aliasing the jaggies like it is in the cheap and chearful DVD
> players.
> I bristle too!
>
> If one looks at discrete frames then the information does not appear to be
> there but by analysing the differences between subsequent frames/fields
> additional information can be reconstrcted and then used in the resize
> algorithm to yield greater detail in the film blow up. There is a lot of
> information stored in a 60i or 50i video stream that is not simply
> disregarded when going to 24p and one can reasonably expect better than
> S16
> quality when blowing SD video up to 35mm. I'm not convinced its a
> doubling
> of resolution from 500 lines to 1000 though.
>
>
> "Smarty" <nobody@nobody.com> wrote in message
> news:IPKdncJTsudTU0HeRVn-qA@adelphia.com...
>> Exactly! I think the industry use of the word "resolution" is unfortunate
> in
>> the semantic sense however, because it inaccurately conveys the
>> impression
>> that the ability to resolve more information is achieved. To astronomers,
>> physicists, or optometrists, the resolving power of the eye to see more
>> detail is not benefited by such ("double the vertical resolution")
> methods.
>> When interpolation or other smoothing and filtering is done to create the
>> impression of an improved picture, the "apparent" resolution is, at best,
> a
>> visual deception, taking advantage of perceptual (as opposed to physical)
>> phenomena. Frank used the phrase "display/presentation frame size" in a
>> another recent HDV thread to refer to the specification which actually
>> ***
>> is**** being (in this case) doubled, and you really have an equal or
>> ***lesser*** resolution image being represented in a display/presentation
>> frame which has twice the number of vertical lines.
>>
>> I tend to bristle at this marketing confusion a bit since there are
> numerous
>> (successful) attempts to sell "up-converted" or "up-rezzed" DVD players,
>> projectors, etc. which make claims to improving resolution, turning SD
> into
>> HD, etc. None of them achieves an increase in resolution! They increase
> the
>> frame size, implying that their resolution is increasing, but it isn't.
>>
>> All of them ultimately cannot and do not increase the resolution, no more
>> than taking an mp3 audio signal and "up-converting it" to a higher bit
> rate.
>>
>> Smarty
>>
>>
>> "Martin Heffels" <mitch.mcNeilljn@sprint.ca> wrote in message
>> news:f04pt197hl3unin8m11jmvabtkkth8panm@4ax.com...
>> > On Sat, 28 Jan 2006 21:11:48 -0500, "Smarty" <nobody@nobody.com> wrote:
>> >
>> >>I fail to understand how uprezzing / clever mathematics can increase
>> >>the
>> >>resolution of a lower data rate / lower resolution signal. Doubling the
>> >>number of lines doesn't truly "resolve" any more detail, and the
> inherent
>> >>resolution of the originally sampled signal at the sampling rate it was
>> >>captured is the highest "resolution" achievable. Schemes developed by
> Yves
>> >>Faroudja and others (like DCD) can reduce jaggies or other artifacts,
> but
>> >>this is not in any way an increase in resolution.
>> >
>> > Companies like Du-Art in the US have developed proprietary software
> which
>> > they use to double the vertical resolution (I should have said that),
>> > to
>> > make a SD-based image look better when blown-up to 35mm. What it does
>> > is
>> > that it recreates the intermediate lines, based on the pixelsof the
> lines
>> > next to it. Of course this is not really going back to what it was
> (which
>> > is impossible after throwing away a lot of the information), but a
> pretty
>> > good approach, and makes a blow-up to 35mm look much better (from 500+
>> > lines to 1000+ lines)
>> >
>> > cheers
>> >
>> > -martin-
>> > --
>> > Never be afraid to try something new.
>> > Remember that a lone amateur built the Ark.
>> > A large group of professionals built the Titanic.
>>
>>
>
>
Navigation:
[Reply to this message]
|