AVS Forum banner

Status
Not open for further replies.
1 - 20 of 66 Posts

·
Registered
Joined
·
4,525 Posts
Discussion Starter · #1 ·
This topic came up in another thread. This is an attempt to show that although there is NO WAY to create source material that does not exist to begin with, it still is possible to improve upon the presentation of that source material to make it appear as though there is additional detail.


Since a picture is worth 1000 words...



Eric Clapton, at native resolution:

http://www.colorfacts.com/colorfacts...s/clapton1.jpg


The SAME EXACT source picture, on a higher resolution device.

http://www.colorfacts.com/colorfacts...s/clapton2.jpg

Note that this picture still has the same source material, upscaled for a higher resolution display. Although NO additional information is available from the source material, notice that additional detail that actually WAS in the source may be easier to make out at the higher resolution.


Specifically, note how the ring on the right hand, the watch on the left hand and even the eyeglasses themselves become more recognizable in the upscaled image, even though there was no more "information" in the image than the original picture had.


Of course, there is simply no substitute for having a higher quality source to begin with: :D
http://www.colorfacts.com/colorfacts...s/clapton3.jpg


(All pictures simulated)
 

·
Registered
Joined
·
4,525 Posts
Discussion Starter · #2 ·
PS - You may notice one more thing...


If you stand far enough away from your display, you will reach a point where the top two pictures appear identical, but the third picture (from the better source) may still appear sharper.


If you stand even further, the three pictures will become indistinguishable.


Summary:


FINDING #1: Resolution does matter...even past the resolution of the source material.

FINDING #2: However, there are practical real-world considerations like distance to screen, image size, source material quality (and possibly others like projector brightness and contrast) that will negate finding #1.
 

·
Registered
Joined
·
11,020 Posts
Agreed, thanks for a great post. It would be nice if you offered an explanation of what you are showing us here :). i.e. projector/s, source, resolution of each picture, etc.
 

·
Registered
Joined
·
4,075 Posts
Q, where in God's name did you get that phone recording, "Another happy customer"? One of yours maybe?? :D


Dan
 

·
Registered
Joined
·
1,587 Posts
Quote:
where in God's name did you get that phone recording
LOL

I was thinking the same deal.

Still I do however seem to remember being that emotional the last time my HDD crashed. It is not pretty when you loose info, and its getting hard to store because we keep so much of it nowdays.


DavidW
 

·
Registered
Joined
·
9,769 Posts
OK Mark, I totally concur. Hey maybe someday we will all have a JVC 4000 x 3000 projectors to show material.;)


I've never understood people saying higher resolution wont benefit!?!. I thought It was so obvious that it DOES make a difference that it didn't need to be explained. For another example look here

http://www.htguide.com/CFBoards/inde...estTimeout=500


BTW I'll take 1920 x 1080 for now. Won't upgrade till that happens. Manufacturers you listening?
 

·
Registered
Joined
·
8,441 Posts
Quote:
Originally posted by Tryg
...I've never understood people saying higher resolution wont benefit!?!. I thought It was so obvious that it DOES make a difference that it didn't need to be explained....
Not sure who the "people" are but they must not have all there smarts or they are trolling. Who would believe that a properly scaled higher resolution picture is inferior to SD picture displayed at native resolution? :confused:
 

·
Registered
Joined
·
4,644 Posts
Another way to look at this issue is that digital displays use squares to represent pixels, whereas an ideal display would theoretically use blurred circles that slightly overlap each other in order to more accurately reconstruct a continuous image. However, it is possible to use a higher number of smaller squares to approximate blurred circles. In other words, some displays use one square to represent each source pixel, but other displays use several smaller squares to represent each source pixel.


As your viewing distance increases, it becomes harder for your eyes to tell the difference between a square and several smaller squares that seem to form a blurred circle.
 

·
Registered
Joined
·
9,769 Posts
Quote:
Originally posted by Erik Garci
Another way to look at this issue is that digital displays use squares to represent pixels, whereas an ideal display would theoretically use blurred circles that slightly overlap each other in order to more accurately reconstruct a continuous image
What does more accurately mean? Sharpness?


CRT projectors use the "blurred circles" meathod don't they? Yet digital projectors can look much sharper.


Blending leads to softness. The crisp accurate image that a high res digital with high res source material can do is nothing short of astonishing.
 

·
Registered
Joined
·
4,644 Posts
Quote:
Originally posted by Tryg

What does more accurately mean? Sharpness?
I mean that it more closely matches the smooth, continuous nature of images that represent real world scenes.


Sharpness has more to do with the MTF (modulation transfer function) of the display device.
 

·
Registered
Joined
·
4,649 Posts
Originally posted by Tryg:


"BTW I'll take 1920 x 1080 for now. Won't upgrade till that happens. Manufacturers you listening?"


I strongly agree with 'ya...and would also add that I would be perfectly happy with upgrading if Sony ever gets off their butts and begin selling consumer front HT projection systems using the GLV (gradient light valve) engine...for ultra high contrast ratios...1920x1080p displaying...perfect colimetry...superb reproduction of blacks (better than CRT)...perfect gray scale (or so the theory goes)...long life...on the fly selection of aspect ratios without the need for anomorphic lens attachments...at affordable prices (under $10K anyway, the only way Sony could beat so-called "high-end" DLP front projector manufacturers). Ah! Come 'on, Sony -hurry up!...


-THTS
 

·
Registered
Joined
·
232 Posts
william, i'm happy to say i'm one of "those people" who don't understand why the native resolution of a dvd disk is NOT the limiting factor. i resent you implication that i'm not smart. i'll have you know i've just been promoted to assistant manager at Burger King. so that pretty much blows your argument out of the water.


further, there's a Marquee 9500 owner currently slumming in the sub 5k projector forum that says switching back and fourth between resolutions makes little difference. he's also the proud owner of a new vga dlp.


i bet yer feelin' pretty small about now...
 

·
Registered
Joined
·
8,772 Posts
Another good example of the virtues of resolution scaling, Mark!


For those that missed the thread that started this, see:

http://www.avsforum.com/avs-vb/showt...1&pagenumber=2


The main points I outlined there were that spatial resolution discussions are meaningless unless the distance to an image and the image size are specified. Human eyes that are rated at 20/20 vision have a resolution limit of about one arc minute (1/60 of a degree).


At 2 image/object widths distance from an image or object, a human eye with 20/20 vision can only resolve about 1720 pixels horizontally.


Attached is a screen cap from a quick spreadsheet, which tabulates the number of pixels a human eye can resolve for viewing distances from 0.25 through 4.0 image widths. All pixels rounded to the nearest whole pixel.


You can check the math from the formulas in the thread above and the academic link in that thread. The numbers for 1.0 and 2.0 image widths seem OK, so I'm assuming the rest of the values for the sheet are correct.


If you have better than 20/20 vision, your upper resolution limits would be higher, and lower for worse-than-20/20 vision.


If you genetically engineer new eyes with a higher rod/cone density, then maybe a 4000 x 3000 pixel projector makes sense. But for the rest of us viewing from 1.5 screen widths or more, 1920x1080p should do just fine
.
 

·
Registered
Joined
·
1,226 Posts
Mark,

Totally convincing, as usual! Thanks.



Totally off topic,


QQQ,

What did that guy think CDRW discs were for, or tape backup systems, or ZIP disks, or even the good old floppy drives? I heard once of a tax accountant who lost ALL of her clients' tax data for YEARS in a crash - and had never made backups. :eek: She went bankrupt, from what I remember. My husband has a great saying: Only back up work that you want to keep after a crash. Well, that's just about everything, isn't it?
 

·
Registered
Joined
·
4,644 Posts
Quote:
Originally posted by Tryg

CRT projectors use the "blurred circles" meathod don't they? Yet digital projectors can look much sharper.
CRT projectors don't actually use blurred circles, although in some cases they may appear to use blurred circles, especially if adjacent scanlines overlap each other slightly, and if edge transitions between horizontally adjacent pixels look smooth.
Quote:
Blending leads to softness.
Are you using the term "blending" to refer to downconverting, i.e., converting a signal to a lower resolution? If so, I would agree that it leads to softness, but I don't think that has anything to do with the advantage of displays that have higher pixel counts than the source material.
Quote:
The crisp accurate image that a high res digital with high res source material can do is nothing short of astonishing.
Yes, and a higher res digital with the same high res source material would be even more astonishing.
 

·
Registered
Joined
·
8,441 Posts
Quote:
Originally posted by bgl90042
william, i'm happy to say i'm one of "those people" who don't understand...
Congratulations!
Quote:


i'll have you know i've just been promoted to assistant manager at Burger King. so that pretty much blows your argument out of the water....
No offense, but being an assistant manager of a BK hardly qualifies you as an expert on computer scaling and display technologies in my book. Mark Hunter on the other hand does qualify.

Quote:


i bet yer feelin' pretty small about now...
Well let me think about that. :confused: Not at all!:D
 

·
Registered
Joined
·
2,091 Posts
Quote:
Originally posted by William
Congratulations!



No offense, but being an assistant manager of a BK hardly qualifies you as an expert on computer scaling and display technologies in my book. Mark Hunter on the other hand does qualify.



Well let me think about that. :confused: Not at all!:D
You do realize he has to be joking ;)


Or maybe you do realize and I missed your joke:confused:
 

·
Registered
Joined
·
6,227 Posts
Quote:
CRT projectors don't actually use blurred circles
Dont they ??? I am no CRT guy at all but I thought that was what CRT's do when they display a pixel... The shoot a photon beam of light from each gun to (hopefully ;) ) all hit the same spot and create a circular spot of coloured light.... As the light is strongest at the centre of the circle the edges are by definition blurred.... I even thought that was what the term gaussian (spelling ?) blurr related to...


I also believed that was one of the touted factors of why CRT's have a creamy / more fluid appearance... That the gaussian blurr allowed the pixels to allow for a slight edge blurring between them...
 

·
Registered
Joined
·
1,265 Posts
Quote:
Originally posted by DanHouck
Q, where in God's name did you get that phone recording, "Another happy customer"? One of yours maybe?? :D


Dan
that phone recording has been around for year. I must admit I have not heard it since 1999 and it still crack me up.


Where I work we have a huge technical service phone support, every month we all get together and pick "the bone head phone call of the month"


We got some classic one's that will blow your mind.
 

·
Registered
Joined
·
4,720 Posts
Quote:
Originally posted by Phat Phreddy
Dont they ??? I am no CRT guy at all but I thought that was what CRT's do when they display a pixel... The shoot a photon beam of light from each gun to (hopefully ;) ) all hit the same spot and create a circular spot of coloured light.... As the light is strongest at the centre of the circle the edges are by definition blurred.... I even thought that was what the term gaussian (spelling ?) blurr related to...


I also believed that was one of the touted factors of why CRT's have a creamy / more fluid appearance... That the gaussian blurr allowed the pixels to allow for a slight edge blurring between them...
Phat Phreddy,


Yes - a CRT does "paint" its pixels in the manner you describe - with Gaussian falloff of the electron beam

intensity.


However, Erik is incorrect in stating that an "ideal display" would use blurred circles. That's not ideal

at all. The ideal method would be to preserve the information captured in the original signal. The blurring

is a side effect of the fact that you can't perfectly collimate an electron beam. However, it does have the

pleasing effect of "blending" the pixels together.


The main point is that the original image contains more information than a straight representation of the raw data

would indicate.


Case in point, take a look at the boundary between Clapton's left hand and his shirt, just above the watch. That boundary

is a jagged line in the original. Now intuitively you know that edge of Clapton's hand is not a jagged edge, but is

a smooth line. A good scaling algorithm uses the information from the surrounding pixels to deduce the location of the

actual edge of Clapton's hand, and then create new pixels at a higher resolution that represents this.


By quantizing the image into pixels, you've added some high frequency noise to the Fourier transform of the image. But

the fact that you know that noise shouldn't be there, means you can filter it out - and smooth the image. That's what

a good scaler does. A CRT is in effect doing some filtering because it has a limited bandwidth.


However, that's due to a limitation on the CRT - not because it is "ideally" reproducing the image. A better CRT with

higher bandwidth would do a better job of reproducing the quantization noise - and the picture would look worse with

the better CRT.


The ideal solution is to use a good scaler with good scaling algorithms to do the best job possible of filtering out the

noise - and then to display the filtered image on a display device with the highest resolution available.


Dr. Gregory Greenman

Physicist
 
1 - 20 of 66 Posts
Status
Not open for further replies.
Top