What exactly is scaling?

This is the action of scaling a bitmap of a given resolution to another resolution.


Stricto sensus, upscaling means going from a given resolution to an higher one while downscaling is the opposite. But most of the time, when using scaling, we're dealing with upscaling.


Why do we do that ? Because each image of a movie is encoded as a 720x480 (or 576 depending on the region) bitmap and some projectors can go much higher (no scan lines, a more detailed image ...) than that.


That's why every dvd playing application out there, and Dscaler as well, are closely linked to the notion of scaling.


Scaling is done on the fly in hardware by any recent vga card. Using this mechanism, the application has to provide the initial-resolution bitmap, and the vga card will display it at an higher resolution on its own. No cpu activity involved.


Hope this clarifies things a bit.

I understand your question.


Let me answer in two points.


First, with high resolution crt projectors, they are so precise that if you don't provide it an high resolution (aka scaled) image, you'll see scanlines (black/empty lines between colors/useful lines) that will distract you.

Also, by getting higher in resolution, you can sliggly modify the picture to prevent aliasing for instance.

Finally, with a matrix-type (lcd or dlp) projector, you need to scale the source to the resolution of the matrix. This can be done by projector itself, but sometimes it's better to leave the pc generates an image of the required resolution than letting the projector do so.


Second, and this is why the "mystery" subject .. in theory, 720x480 pixels are encoded, so scaling shouldn't improve the image. However, intelligent scaling algorithms associated with high-resolution ready projector makes sometimes people say "hey, watching Gladiator in 1600x1200 on my 9" crt was just incredible, I've discovered details I couldn't see before".


Sounds like the issue "bits are only bits" when talking about sp/dif output. However, even if this works in theory, it doesn't stand the test of reality


If someone want to go on the philosophical side of the issue of scaling "Why can I see more when scaled ?", go ahead

Scaling can often "creates" information from the information that it has.


For instance, if I were to give you 3 lines of information:


1

3

5


and your projector could display 5 lines of information. Where would you put the extra lines? You would evenly insert them:


1

?

3

?

5


What would you put in place of a question marks?


This is kinda what a scaler does. A good scaler can interpolate and guess on what the information would be. While the "2 and the 4" weren't part of the information, their presence helps to fill out the picture.


-Steve

I personally dont put a lot of stock in the subjective recollections of seeing 'more detail' in an upscaled image.


For a digital projector the display device has a particular fixed resolution (XGA, SVGA, etc). If you feed it a signal with a different resolution, the projector has to scale it to its fixed resolution. Usually the scaling done in the projector is poor. If you scale it yourself with an HTPC or external scaler, the results will be better.


For an analog FPTV, although you typically dont *have* to scale the image to view it, there are two reasons why you might:


1)to eliminate the gaps between scan lines by creating new scan lines


2)Often the electronics of an analog display device will have a 'sweet-spot', some particular Pixel Clock where the bandwidth frequency response of the electronics is at its maximum. If you scale the image close to this particular resolution you will get better performance.


For an HD RPTV, you could in theory scale NTSC to 1080i using an HTPC, I dont know if this would provide significantly superior results to letting the RPTV scale it itself though.


Andy K.