Whats the actual size of my video?: miniDV NTSC widescreen

[vid83]

My source video is standard definition miniDV NTSC 720 x 480 shot in widescreen mode.

When I open the video in mediainfo it says that my video is 720 x 480, which is 1.5:1 aspect ratio, not 16:9.
When I open the video in QuickTime player and use the movie inspector, it says that my video is 853 x 480, which is close to 16:9.

How can I figure out the actual size of my widescreen video?

[edDV]

My source video is standard definition miniDV NTSC 720 x 480 shot in widescreen mode.

When I open the video in mediainfo it says that my video is 720 x 480, which is 1.5:1 aspect ratio, not 16:9.
When I open the video in QuickTime player and use the movie inspector, it says that my video is 853 x 480, which is close to 16:9.

How can I figure out the actual size of my widescreen video?

Your camera records 720x480 for 4:3 or 16:9 (both non-square pixel). For display, 4:3 is resized to 640x480 (square pixel) or ~853x480 (square pixel) for 16:9. HDTV sets then upscale to display resolution.

Recording 16:9


DVD, ATSC, DVB, etc. record SD 16:9 the same way.

As recorded


As displayed

[Nelson37]

Ed, I've got a question on this and I'm not even sure how to put it.

Let's say you have the same broadcast available in both 4:3 and 16:9, assuming both with correct AR. Both are captured, in correct AR. The first one displays at 720x480, the second at ~853x480.

Does the 16:9 contain more video data? Was it captured with more video data?

Is the apparent increase in resolution actually there, or is it a simple resize?

This makes my head hurt, and I used to do differential calculus without notes.

Is the answer that both captures have the same amount of data, but the 16:9 is compressed at capture and de-compressed on display, threfore resulting in more data actually displayed? But, if compressed, is it lossy or lossless?

Where does the "extra" data come from, if there is any? I need aspirin.

[minidv2dvd]

less actual video data(in most DV cams) in WS as the sensor uses fewer pixels. the used pixels captured image is compressed horizontally for recording and expanded for playback. basically the same thing is done in HDV cams that record 1440x1080 and display the video as 1920x1080.

[manono]

The first one displays at 720x480, the second at ~853x480.

If the first one is 4:3 then it displays at 640x480. A Hi-Def cap is either 1280x720 or 1920x1080. It's not 16:9. If you're capping Hi-Def at Standard-Def 16:9 720x480, then that's something else.

If the first one is widescreen 4:3, then of course the 16:9 one has higher resolution and displays more data, as much of the 4:3 one will be taken up with black bars and will be effectively displaying 640x360. If the first one is pan-and-scan or "full screen", then why would you want it at all, if you have the 16:9 one available?

[jagabo]

Let's say you have the same broadcast available in both 4:3 and 16:9

The properties (bandwidth, snyc pulses, etc.) of the broadcast analog video signal are always the same. There is no 16:9 or 4:3 in the signal. The display device determines the displayed aspect ratio. Of course, until ~10 years ago there were only 4:3 display devices so all broadcasts were designed for 4:3 display. Any widescreen source was cropped or letterboxed for 4:3 display. To this day, analog broadcasts, at least in the USA, are designed for 4:3 display.

Both are captured, in correct AR. The first one displays at 720x480, the second at ~853x480.

This doesn't really make sense without a context. But assuming standard definition analog broadcast -- again, the signal is always the same. Whether you display it 4:3 or 16:9 it can only resolve as much detail (horizontally) as the bandwidth of the signal provides. So, no, displaying the signal at 16:9 will not show any more resolution than displaying it at 4:3 (details of a particular display device aside). Given that the horizontal resolution of broadcast standard definition analog NTSC video is equivalent to about 450 vertical lines (across the full width of the picture), capturing at 720 pixels is sufficient. Higher bandwidth sources may go as high as 600 to 700 lines across the full frame.

Also, keep in mind that these numbers are not absolute. Saying the analog resolution is 450 lines doesn't mean that 449 lines displays perfectly and 451 doesn't display at all. What happens is the amplitude of a sinusoidal signal decreases as it approaches the bandwidth/resolution limit. A low frequency signal will swing from full black to full white. As the frequency increases the swings will range from not-so-dark-black to not-so-bright-white. At even higher frequencies the swings will eventually blur into a solid gray. The quoted bandwidth/resolution is often the point at which the amplitude of the signal has degraded to 50 percent of the original source.

Beyond that you get into a discussion of nyquist frequencies, digital scaling algorithms, and other complexities.

[minidv2dvd]

did you guys read the same op? he's taking about miniDV cam. the cam always records in 720x480 whether it's 4:3 or 16/9. the difference is in the shape of the recorded pixels. 4:3 has square pixels, 16x9 has horizontally compressed/expanded pixels. 16x9 miniDV should be encoded to 852x480 if transferred to a different type video.

[jagabo]

I was answering Nelson37's questions about analog video. edDV answered the OP's question in the very first reply.

[minidv2dvd]

gotcha. not sure he was using "broadcast" as in tv, but what do i know.... it seemed the numbers corresponded more to DVavi than ota.

[Nelson37]

I guess I made a mistake in using the word "broadcast".

Forget the source. Assume both are in correct AR, and full-screen at that AR. No black bars.

It is the apparent "compress/de-compress", that I am interested in. It's like there is a freebie here.

To use an extreme example, if I "resize" a 352x240 video to 704x480, I have "cheated" and do not really have a video that is truly 704x480, though technically it is. While the 720 to 853 WS resize is far less dramatic, it is still a resize. Though done to correct the AR of a video that has already been horizontally compressed.

If both 4:3 and 16:9 vids are full-screen, same res on disk, no black bars, when displayed in correct AR there is apparently "more data" in the widescreen version.

It just seems like something for nothing, and TANSTAAFL.

[gadgetguy]

Think of it at the pixel level. There is only one pixel, whether it's displayed as 4:3 or 16:9 and it only contains one color. You haven't actually resized to a new resolution, only the shape of the pixel. Only when it gets converted to a square pixel format is the resolution actually changed and part of the function of the resizer is to interpret how the colors get reallocated to "create" the extra information.

[edDV]

The following is for 480i/480p only. 576i/576p "PAL" is somewhat different in details.

For DV, DVD, ATSC, DVB, the picture is always horizontally rescaled for square pixel display. 4:3 is H "compressed" from 720 to 640. 16:9 is H expanded to 852x480. How this gets done gets a bit more complicated.

DV camcorder (records 720x480i):
- Digital transfer over IEEE-1394 is done at 720x480 (no change). H rescale is done in the player. If you display 720x480 as square pixel, 4:3 will show people a bit fat, 16:9 will show them thin.
- Analog composite, S-Video or component outputs are scaled prior to output. A 4:3AR flag results in 4:3 analog. A 16:9AR flag is usually scaled for letterbox 16:9.

DVD Player: (records 720x480, 704x480, 352x480 ...)
- HDMI (interlace mode) - SD passes pixel for pixel data for 720 and 704. 352x480 and below are scaled to 704x480. In upscale mode all get processed to 1280x720p/59.94 or 1920x1080i or 1920x1080p at 29.97* fps.
- HDMI (progressive mode) - 720 or 704 @23.976 gets frame repeated 2:3 to 59.94 fps. In 1280x720p or 1920x1080p upscale mode, frames are rescaled. In 1920x1080i upscale mode, source frames are telecined to 29.97 fps and fields are upscaled to 1920x1080i.
- Analog (interlace mode) - same as DV except component wide options are letterbox, pan/scan or 16:9.
- Analog (progressive mode, 480i out) - composite, S-Video and component get telecined to 29.97i.
- Analog (progressive mode, 480p out) - composite, and S-Video get telecined to 29.97i. Analog component is frame repeated 2:3 to 59.94p.
- Analog (upscaled analog component) - similar to HDMI above.

ATSC/DVB SD (typical transmission is 704x480i/29.97 but 704x480p/23.976 or 704x480p/59.94 are allowed)
- Output modes are similar to DVD player above.

ATSC/DVB HD (typical transmission is 1280x720p/59.94 or 1920x1080i/29.97, cable sometimes uses 23.976 fps)
- HDTV is transmitted with square pixels.
- 1080i Film source needs IVTC to 1920x1080p/59.94, then is scaled to native display resolution.
- 1080i video source is deinterlaced to 1920x1080p/59.94, then is scaled to native display resolution.


* HDMI 1.3 allows for the followig progressive modes. Only high end HDTV sets receive 23.976fps for film source.
- 720x480p/59.94,
- 704x480p/59.94
- 720x480p/23.976
- 704x480p/23.976
- 1280x720p/59.94
- 1280x720p/23.976
- 1920x1080p/59.94
- 1920x1080p/23.976

[jagabo]

I guess I made a mistake in using the word "broadcast".

Forget the source. Assume both are in correct AR, and full-screen at that AR. No black bars.

It is the apparent "compress/de-compress", that I am interested in. It's like there is a freebie here...

Nothing is different except the shape of the pixels as they are displayed. When displaying a 720x480 frame as 4:3 the individual pixels are thinner than they are tall. When displayed at 16:9 they are wider than they are tall.



On the top is an array of 4x4 square pixels in a black and white checkerboard pattern enlarged so you can see each pixle clearly. In the middle is the the same 4x4 pixel array as displayed to deliver a 4:3 image from a 720x480 source. At the bottom is the same 4x4 array of pixels as displayed to deliver a 16:9 image from a 720x480 source. The 16:9 image has no more detail than the 4:3 image.

[minidv2dvd]

jagabo you have your 4s and 3s mixed up. the first number is always the x axis or horizontal. 4:3 is wider than it is tall. DVavi SD is square pixel, but there are more square pixels along the horizontal axis (720 pixels long x 480 pixels high).

[manono]

No, he has it right. He's only showing a 4x4 part of the overall 720x480. When resized for 4:3, that 720 gets "squeezed" or "compressed" from the sides, so that each pixel becomes higher than it is wide. He's sort of turning around the way we usually think of what's going on. Rather than thinking of square pixel 720x480 being resized to square pixel 640x480, think of it remaining 720x480 but the shape of the pixels changing.

That way of thinking of what's going on makes it easier to understand how 4:3 and 16:9 sources can have the same amount of detail after being resized.

[jagabo]

jagabo you have your 4s and 3s mixed up. the first number is always the x axis or horizontal. 4:3 is wider than it is tall. DVavi SD is square pixel, but there are more square pixels along the horizontal axis (720 pixels long x 480 pixels high).

I don't have anything confused. To display a 720x480 frame as 4:3 the individual pixels have to be drawn narrower than they are tall. If the pixels were displayed square the display would be 4.5:3 (ie, 3:2) not 4:3.

[minidv2dvd]

DVavi isn't resized ever. it is only 720x480 at all times if it is DVavi.

[jagabo]

DVavi isn't resized ever. it is only 720x480 at all times if it is DVavi.

Either the pixels have to be drawn the right shape or the frame has to be resized in order to display at the correct aspect ratio.

[JohnnyMalaria]

Jagabo is indeed correct - checkmate (sorry, couldn't resist).

It's all down to the final display unit. With CRTs, it's trivial. The 720 pixels simply fills one scan line. It's the CRT's aspect ratio that determines the appearance. (Excepting displays that letterbox/pillarbox).

For PAL, it is the same except that there are 576 scan lines instead of 480. This means that the DV protocol's D1 sampling of 720 samples per line results in a digital frame that is narrower than the 768 x 576 square pixel image (the opposite of NTSC).

[deadrats]

did you guys read the same op? he's taking about miniDV cam. the cam always records in 720x480 whether it's 4:3 or 16/9. the difference is in the shape of the recorded pixels. 4:3 has square pixels, 16x9 has horizontally compressed/expanded pixels. 16x9 miniDV should be encoded to 852x480 if transferred to a different type video.

actually 4:3 has pixels that are 4 units wide and 3 units high, and 16:9 likewise has pixels that are 16 units wide by 9 units high.

[jagabo]

actually 4:3 has pixels that are 4 units wide and 3 units high, and 16:9 likewise has pixels that are 16 units wide by 9 units high.

No, 4:3 and 16:9 refer to the final shape of the picture. 4:3 NTSC DV pixels are about 91 percent as wide as they are tall. 16:9 NTSC DV pixels are about 21 percent wider than they are tall.

DAR = SAR * PAR

DAR = final Displayed Aspect Ratio
SAR = Storage Aspect Ratio (frame size)
PAR = Pixel Aspect Ratio (shape of individual pixels)

For example, 16:9 DV should have the 16:9 picture within the inner 704x480 portion of the frame with 8 pixels of padding at both sides. So the 16:9 frame size is really 704x480 (actually the 4:3 image is contained in a 710.85x486 frame but since 6 lines cropped on the vertical dimension you have to cut the width down to ~702 which is rounded up to 704 to be a mod 16 size):

16/9 = 704/480 * PAR
16/9 * 480/704 = PAR
1.21... = PAR

http://lipas.uwasa.fi/~f76998/video/conversion/

[gazgaz]

so can someone explain in simple terms

why I can watch a blu-ray on my 58inch panasonic plasma in all its glorious size

and when I rip it to a TS (TSmuxeR,ToMNT) it becomes a letter box shape

and I am forced to resize by zooming my screen and sacrificing the ends

I confess as TS plays on my NMT
I was wondering why people go on to create alternatives (AVI, MKV)

and yes I confess I have little idea but do have a headache from reading so many pages to try to find answers

whenever I use a program like TMPGenc, AVS, 321 video converter, Divx, xilisoft, etc I indicate 16:9 pal thinking this was all I needed to do

I will continue to read in the hope for insights

thanks in anticipation

Gaz

[minidv2dvd]

DVavi isn't resized ever. it is only 720x480 at all times if it is DVavi.

Either the pixels have to be drawn the right shape or the frame has to be resized in order to display at the correct aspect ratio.

i'm taking about the video itself. not how it's displayed. have you ever tried to write back to miniDV tape anything that's not 720x480? ain't gonna happen.

[jagabo]

so can someone explain in simple terms

why I can watch a blu-ray on my 58inch panasonic plasma in all its glorious size

and when I rip it to a TS (TSmuxeR,ToMNT) it becomes a letter box shape

and I am forced to resize by zooming my screen and sacrificing the ends

ToNMT? To NMT? Networked media tank? I don't know much about that device. Are you watching the ripped file on the same HDTV? Did you only rip the video? Did you convert, crop, or resize the video?

Note that a lot of widescreen movies are wider than 16:9 and will be letterboxed in a 16:9 frame on Blu-ray discs. For example, a 2.35:1 movie should have a 1920x816 portion of the 1920x1080 frame filled with picture, the rest with black bars.

whenever I use a program like TMPGenc, AVS, 321 video converter, Divx, xilisoft, etc I indicate 16:9 pal thinking this was all I needed to do

When you convert a video you may need to make sure the conversion program knows the display aspect ratio of the source as well as specifying the display aspect ratio of the output.

[vid83]

Back to the original post, what is the maximum size I can make my video without losing quality (1:1 pixel aspect ratio widescreen)? Is it 853 x 480?

[JohnnyMalaria]

If you mean losing quality once it is compressed to DV then it is 720 x 480. Anything higher than that will be reduced in size prior to compression.

Depending on how you intend to view the final video, it probably won't matter anyway. Using the S-video output of a consumer DV camcorder or stand-alone converter will result in an analog signal with less horizontal resolution and the quality of the display unit will also be a factor.

If you are using DV as a source and creating a different format for output then the argument is different. For regular MPEG2 (for DVD), keep it as 720 x 480 since the very same factors apply as for DV. For a format that stores the video as 1:1, it will have to be stretched anyway. Irrespective, if the video was recorded through the lens, the level of detail won't be anywhere near the resolution of the DV format.

[vid83]

The video is for YouTube. YouTube recommends 640 x 360, 16:9 aspect ratio, 1:1 pixel aspect ratio. But how much larger than 640 x 360 can I go?

[jagabo]

1920x1080

[gazgaz]

sorry to ask my question in your post vid83

thought I might get away with a quick "size" question

thanks for offering something to think further on jagabo

I will continue to read the forums

gaz

[minidv2dvd]

The video is for YouTube. YouTube recommends 640 x 360, 16:9 aspect ratio, 1:1 pixel aspect ratio. But how much larger than 640 x 360 can I go?

the smallest size that will get you the watch in HD link is 720p - upsize your DVavi into h264 mp4 1280x720 progressive with aac audio. any bigger won't make any difference. 720i to 480p will get a watch in HQ link, 480i and down just the normal .flv video.