How does DVD-player resize?

[leonb]

Hi

The resolution of a Pal SVCD is 480 X 576 and that of a PAL VCD is 352 X 288. So I wondered how a DVD-player gets the picture to fill the whole area of the TV screen, with the correct aspect ratio? Does it resize on the fly or does it stretch the picture somehow? And does this resize/stretch have a negative impact on the video quality?

[mb1]

The player stretches svcd horizontally with factor 1.5 and vcd vertically and horizontally with factor 2.

Of course the more the image gets stretched the unsharper the picture is.

[avih]

this answer relates to TRUE interlaced material (such as live tv, home movies etc). however, the last paragraph deals shortly with NON interlaced material (mostly film). to understand more read other guides, read this post carefully and apply some logic.

about 576 and 288 and 25fps, i'm reducing this explenation to pal only (for ntsc, swap with 480 and 240 and 30fps respectively):

1. first regarding the horizontal resize: it doesn't matter what method is used, as the horizontal rezolution is stretched to full screen-wide (tv DON'T have a horizontal resolution, since it's analog, and the signal changes 'smoothly' throught a 'scan-line' (=a single row of the tv display). however it's 'common' to relate to a width of about 720 pixels as a horizontal resolution for digitized(=capture) image needs).

2. regarding the vertical resolution. tv display is interlaced. the tv displays 2 FIELDS one after the other every 1/25 of a sec (therefore each field is displayed for 1/50 of a sec, hence 50fps). each field is composed of 288 scan-lines (=vertical resolution of 288).

3. when we capture, or compose svcd, each 2 fields are captured into ONE FRAME. so we have 25 FRAMES a seconds, which actually holds 50 FIELDS a second since each frame has 2 fields interleaved (one of the fields is in the even lines, the other is in the odd lines). this interleaving is what's known as 'interlaced'. the interlace artifacts are due to the fact that each image (=captured frame) has 2 different fields that were originally captured in 2 DIFFERENT timestamps but are interleaved into the same image.

and now to the actual answer:

vcd has a vertical resolution of 288 at 25 fps. so when it's displayed by the player, each field is DUPLICATED and displayed twice on the tv set, and each of these duplications fills the whole vertical display. so eventhough the tv displays 50 fields/sec each 2 of them are identical, and the actual display is 25 images/sec.

svcd on the other hand (when used in 576 vertical resolution) actually contain these 2 DIFFERENT fields in each image (=frame). and the player displays each of these fields one after the other on the tv. again, each field is 288 lines high, and fills the entire vertical display. so when we're playing interlaced svcd, we get true 50 images/sec on the display, without any 'duplications'.

of course, if the source material is NOT interlaced (=originally NOT 50 images a sec), then 576 lines still display the same as explained before, but rather than giving us more TEMPORAL resolution (=more information on the timeline) it gives us more SPATIAL resolution (=more information per image), such that each 2 fields belong to the same image, but give different information, since one holds the odd lines of the 576 lines image, and the other holds the even lines of that image. so we get a 'richer' image.


hope this helps.
cheers
avi