On Fri, 28 May 2010 02:10:48 +0200, nemesis <nam### [at] gmailcom> wrote:
>> Try one at 1280x720. The software in the TV may well be using this
>> resolution internally.
>
> just done.  It just displays smaller than the previous one, plenty more  
> space left. :P

It is possible that the software is using 1920x1080 as a baseline, with  
anything smaller getting scaled proportionally. Of course, since that  
particular model can only display 720p, even the baseline gets downscaled  
by a DSP eventually, but the software is probably near-identical on all  
models.



-- 
FE

Post a reply to this message

"Fredrik Eriksson" <fe79}--at--{yahoo}--dot--{com> wrote:
> On Fri, 28 May 2010 02:10:48 +0200, nemesis <nam### [at] gmailcom> wrote:
> >> Try one at 1280x720. The software in the TV may well be using this
> >> resolution internally.
> >
> > just done.  It just displays smaller than the previous one, plenty more
> > space left. :P
>
> It is possible that the software is using 1920x1080 as a baseline, with
> anything smaller getting scaled proportionally. Of course, since that
> particular model can only display 720p, even the baseline gets downscaled
> by a DSP eventually, but the software is probably near-identical on all
> models.

oh

so that would mean to display the images at full resolution, I should use "wide"
or "face zoom" mode?  I'm trying to think of an image that would lose obvious
detail when downscaled that would show up in those modes...

Post a reply to this message

nemesis wrote:
> so that would mean to display the images at full resolution, I should use "wide"
> or "face zoom" mode?  I'm trying to think of an image that would lose obvious
> detail when downscaled that would show up in those modes...

A grid.


-- 
Darren New, San Diego CA, USA (PST)
    Ada - the programming language trying to avoid
    you literally shooting yourself in the foot.

Post a reply to this message

Darren New <dne### [at] sanrrcom> wrote:
> nemesis wrote:
> > so that would mean to display the images at full resolution, I should use "wide"
> > or "face zoom" mode?  I'm trying to think of an image that would lose obvious
> > detail when downscaled that would show up in those modes...
>
> A grid.

that should be useful. :)

Post a reply to this message

"Fredrik Eriksson" <fe79}--at--{yahoo}--dot--{com> wrote:
> Make no mistake; the large pictures *are* getting downscaled. It could be
> that the software only supports scaling to certain predefined dimensions.
> It could be that the software looks at all the images and chooses a fixed
> scaling factor based on the dimensions of the largest one. It could be
> that a wizard did it. It could be any number of things, but nothing the
> software does can overcome the fact that the display panel only has
> 1366x768 pixels.

make no mistake:  I'm not suggesting the software is pulling a Jesus here. :)

What I'm suggesting is that perhaps they just advertise the display is 1366x768
because it can't, for some reason or another, display 1080p video content but
the physical screen is still 1080p.  That or the picture are being unreasonably
scaled down.  Yes, I realize that by Occam's razor the latter should win out,
but the facts don't seem to add up.

I just crafted away quite a few 1 pixel grids with 100x100px of spacing between
the lines, in several resolutions, including 1366x768 and 1920x1080.  I've not
rescaled of course, I generated a new 1 pixel grid for each one with Gimp _>
Render -> Pattern.  The 1080 one fits the whole screen just like the previous

patterns or anything like that.  The 768 one doesn't fit the whole screen as
expected.

It's very bizarre and amusing...

Post a reply to this message

On Fri, 28 May 2010 03:33:19 +0200, nemesis <nam### [at] gmailcom> wrote:
>
> I just crafted away quite a few 1 pixel grids with 100x100px of spacing  
> between the lines, in several resolutions, including 1366x768 and
> 1920x1080. I've not rescaled of course, I generated a new 1 pixel grid
> for each one with Gimp _> Render -> Pattern.  The 1080 one fits the
> whole screen just like the previous photos.  Each line is 1 pixel wide,

> The 768 one doesn't fit the whole screen as expected.

Try a small-scale checker pattern, i.e. with each square just one or two  
pixels pixels wide. You could also try a resolution test chart:
  http://www.graphics.cornell.edu/~westin/misc/res-chart.html


Because the scaling is so slight (1080 -> 720 is only a reduction by one  
third), any loss of detail will be very subtle. Furthermore, a pixel-wide  
line is likely to stay pixel-wide and just get a slight reduction in  

unlikely to be a problem unless the scaler is really crappy.



-- 
FE

Post a reply to this message

"nemesis" <nam### [at] gmailcom> wrote:
> "Fredrik Eriksson" <fe79}--at--{yahoo}--dot--{com> wrote:
> > Make no mistake; the large pictures *are* getting downscaled. It could be
> > that the software only supports scaling to certain predefined dimensions.
> > It could be that the software looks at all the images and chooses a fixed
> > scaling factor based on the dimensions of the largest one. It could be
> > that a wizard did it. It could be any number of things, but nothing the
> > software does can overcome the fact that the display panel only has
> > 1366x768 pixels.
>
> make no mistake:  I'm not suggesting the software is pulling a Jesus here. :)
>
> What I'm suggesting is that perhaps they just advertise the display is 1366x768
> because it can't, for some reason or another, display 1080p video content but
> the physical screen is still 1080p.  That or the picture are being unreasonably
> scaled down.  Yes, I realize that by Occam's razor the latter should win out,
> but the facts don't seem to add up.
>
> I just crafted away quite a few 1 pixel grids with 100x100px of spacing between
> the lines, in several resolutions, including 1366x768 and 1920x1080.  I've not
> rescaled of course, I generated a new 1 pixel grid for each one with Gimp _>
> Render -> Pattern.  The 1080 one fits the whole screen just like the previous

> patterns or anything like that.  The 768 one doesn't fit the whole screen as
> expected.
>
> It's very bizarre and amusing...

well, after several tests... we have a winner!  And 768 it is indeed!  Made a
grid with 1-pixel lines separated by exactly 4 pixels and I could clearly count
each pixel inbetween in several different spots by using the "face zoom"
visualization.  No such luck with the other resolutions, that displayed

what really made it click with me was adding a small "foobar" writing in small
9-pixel font and reading it clearly and identifying exact pixels. :)

Damn, why can't they simply use 1:1 pixel instead of re-scaling smaller
pictures?  bizarre...

Post a reply to this message

On Fri, 28 May 2010 04:59:19 +0200, nemesis <nam### [at] gmailcom> wrote:
>
> Damn, why can't they simply use 1:1 pixel instead of re-scaling smaller
> pictures?  bizarre...

I think the software simply creates a 1920x1080 image (by downscaling  
larger images, and adding borders to smaller ones) regardless of display  
panel. That image then gets sent to a final scaler that fits the image to  
the panel resolution. I am guessing they do it that way because it lets  
them use the exact same software on all models.

For "best" results you should pre-scale any images to 1920x1080 before  
putting them on the USB storage.



-- 
FE

Post a reply to this message

"Fredrik Eriksson" <fe79}--at--{yahoo}--dot--{com> wrote:
> On Fri, 28 May 2010 04:59:19 +0200, nemesis <nam### [at] gmailcom> wrote:
> >
> > Damn, why can't they simply use 1:1 pixel instead of re-scaling smaller
> > pictures?  bizarre...
>
> I think the software simply creates a 1920x1080 image (by downscaling
> larger images, and adding borders to smaller ones) regardless of display
> panel. That image then gets sent to a final scaler that fits the image to
> the panel resolution. I am guessing they do it that way because it lets
> them use the exact same software on all models.
>
> For "best" results you should pre-scale any images to 1920x1080 before
> putting them on the USB storage.

thanks for the advice, dude.

Post a reply to this message

Why not try one of the test-cards on this page (far down):

http://en.wikipedia.org/wiki/Test_card

Post a reply to this message