Video format

[zaphead]

Hi, I have captured some analog video (NTSC) and now have an AVI file on my computer. If I open the file with Windows media player, it displays as 640x480. Other players (e.g. CreateMedia Source) display the video as 720x480.

1) How do I know what the native format of the video is?
2) What are the players doing with the pixels when they open the AVI in a different format? Does the player interpolate and fill in pixels?

Thanks!

Ben

[minidv2dvd]

wow not a good first post, but welcome to the forum anyway. not enough detailed info.

what were the hardware and software used to capture with? you choose the settings, what were they?

never heard of createmedia source, what is it?

you can use mediainfo to see what the specs of the captured file are if you don't remember. use the text output setting and post a screenshot of the results here.

[Cornucopia]

My guess is it's DV-avi data and some players will show it as its true SAR of 720x480, while other players (QT is notorious for this) will display it with square pixels AND will modify its reported resolution to being 640x480.

Use Mediainfo to get a better idea about what the true specifics are for this source, and then if a player is changing that, you'll know from now on at least which apps are transparent replayers and which ones aren't.

Scott

[zaphead]

Thank you all for your responses.

I used a 20B44 camera from Videology. The CMOS sensor on the camera has 640x480 picture elements. It has analog (NTSC) output. Here is the datasheet:

http://www.videologyinc.com/cameras/cmos-board-camera-20B44.htm

The video capture device is the Sony GV-HD700E. Here is the product page
http://www.adorama.com/SOGVHD700PAL.html

I downloaded the video from the Sony device via firewire interface to my PC. I just used the stock Windows application that comes with Vista to do that. The only option using that application was the format I wanted to use to save the video file. I chose AVI.

I used MediaInfo and discovered that the AVI file format is 720x480.

I know the physical output of the CMOS is 640x480, so where do those extra pixels come from for the file? Can anyone venture a guess as to where the format changes in the process? Is the Sony device interpolating and recording in 720x480 format, or could it be Windows?

I would really like to be able to look at the video in its native 640x480 format.

Ben

[jagabo]

It would have been helpful if you posted the full report from MediaInfo. But my guess is you have a 720x480 NTSC DV AVI file.

The camera you recorded from may have only had 640x480 resolution but the DV capture device resampled the analog signal to produce a 720x480 (non-square pixels) frame to match the DV spec. As you have seen, some players will respect the DV pixel aspect ratio flag and display the frame with the proper display aspect ratio, some do not.

If you are planning to make a DVD from this video leave it at 720x480.

[zaphead]

It's important that I maintain the native 640x480 format (it's a machine vision application) so it sounds like I need to configure the GV-HD700E capture device to not resample.

[jagabo]

It's important that I maintain the native 640x480 format (it's a machine vision application) so it sounds like I need to configure the GV-HD700E capture device to not resample.

If it's a DV device you have no choice. That is the only size supported by DV.

[edDV]

If you use the Sony GV-HD700E or any other ITU-Rec601 based device (e.g. DVD recorder or pro SDI recorder), it will resample 640x480 (or 768x576 PAL) to 720x480 (720x576 PAL). 854x480 (1024 x 576 PAL) 16:9 wide gets resampled to the same 720x480 (720x576 PAL). A flag distinguishes wide from 4:3 for display.

Most current consumer capture devices will also capture to ITU-Rec601 specs. If you want 640x480 without resample, best to use industrial capture devices.

[2Bdecided]

In the ITU standard, only the centre 704 pixels are within the image area - the extra 8 pixels either side (if present - 704 pixels wide is a perfectly legal resolution) are outside the picture area, and may or may not be black.

Lots of modern equipment isn't ITU standard, and treats all 720 pixels as being within the image area.


To get the raw pixels from your sensor, you need to access the CMOS directly. There are people doing this for 8mm film transfers - check out the threads on doom9 and elsewhere on videohelp, e.g.

http://www.super-8.be/s8_Eindex.htm

Cheers,
David.

[jagabo]

Even if you capture at 640x480 the 640 pixels of the capture would not likely align with the 640 pixels of the CMOS sensor. You would need a digital transfer directly from the camera to the computer for that.

[edDV]

Lots of modern equipment isn't ITU standard, and treats all 720 pixels as being within the image area.

Yes but still approximately follow Rec601 pixel aspect ratios and explain the user's issues with the DV recorder.

The CMOS camera module he is using outputs 640x480 interlaced analog at either 25 or 30 fps. There is no digital access to pixels. Also has AGC, AWB and only 34dB SNR so it is of limited use as an instrument sensor. It is similar to a low end security camera.

[zaphead]

Thank you all for your responses. I now understand the limitations of analog video.

[edDV]

Thank you all for your responses. I now understand the limitations of analog video.

I wouldn't blame analog video, just low end security cams. Low end digital imagers can be equally bad or worse. Depends on the job.

[Cornucopia]

Maybe you didn't understand our responses.

Really doesn't have ANYTHING to do with ANALOG video, per se. (Analog video can be HiDef) Has to do with the low quality camera you've got and the unusual choice of capture.

The capture device you chose was meant for DV (which is inherently 720x480 in NTSCland). When your camera, which has the stated resolution of 640x480, gets output via analog. Then your cap device is REDIGITIZING it (well, something has to if you wanted it on a computer), but at a different resolution than was originally. How would it know?

Think of it like this (audio analogy)...

You have a sampling synthesizer with 32kHz sample rate and play your music with it. But you want to get an EXACT multiple of a sample for use in some complex echo DSP calculation. However, when it comes time to record it, you aren't recording a digital output, you're playing it through a D/A and amp and speakers (so you can hear it), and THEN you're recording that with a soundcard set to 44.1kHz. Of course it's not going to be exact; you've gone though 2 conversions already!!

My suggestions:
1st - get a MUCH better camera, and one that outputs digitally (even old-hat SD DV does that)
2nd - capture direct digitally (prob. via Firewire) and you won't even need that Sony device.

Scott