Aspect ratio 3:2 versus 4:3 with ATI capture card

[Darryl In Canada]

I have always been confused by display aspect ratio, pixel aspect ratio (input and output). I would like to know definitively how to handle this and what settings to use. Specifically...

I capture 8mm tape with an ATI AIW-9600XT. According to MediaInfo, my AVI files (HuffYuv) have a display aspect ratio of 3:2 and a size of 720x480. When I open in Hybrid, in Crop/Resize--Base, the Input Par is set to 1x1.

By comparison, when I capture a MiniDV video via firewire, according to MediaInfo it has a display aspect ratio of 4:3 with a size of 720x480. When I open in Hybrid, in Crop/Resize--Base, the Input Par is 8x9 by default.

Bottom line: what should I set as the Input Par and Output Par for the 8mm captures, and the MiniDV transfers? Thank you!!



[Attachment 78682 - Click to enlarge]

[jagabo]

Virtually all SD devices capture following the ITU standard. They capture SD analog NTSC with of frame size of 720x480 where the inner 704x480 comprises the 4:3 picture (the extra little bits to the two sides are there in case the source or cap is slightly off center). The SAR (Sampling Aspect Ratio, often called PAR or pixel aspect ratio) is 10:11.

The general equation is:

DAR = SAR * FAR

where:

DAR is the Display Aspect Ratio, the final shape of the displayed picture
SAR is the sampling aspect ratio, the relative "distance" between samples across the width and height (often called PAR, pixel aspect ratio)
FAR is the frame aspect ratio, ie frame width : frame height

So:

DAR = SAR * FAR
DAR = 10:11 * 704:480 
DAR = 10 /11 * 704/480
DAR = 0.9090909... * 1.466666....
DAR = 1.3333... 
DAR = 4/3
DAR = 4:3

By the way, MediaInfo is just reporting the DAR as the frame aspect ratio (3:2, for a 720x480 frame) in the absence of a flagged DAR or SAR.

[Darryl In Canada]

Thank you for this detailed explanation.

For 8mm video captured via ATI capture card, what should I enter in Hybrid for Input Par and Output Par?

I should also mention that the ATI capture version looks more stretched out. Is that related to the 3:2 versus 4:3 thing?


[Attachment 78687 - Click to enlarge]

[DB83]

And to add to that, only mpeg2 and DV actually have a DAR flag as part of its capture. Which is why you do see a DAR of 4:3 for your MiniDV capture even though the SAR is still 720*480 (3:2)

[jagabo]

The "correct" SAR (PAR) is 10:11. But it may depend on exactly what you are doing. Commercial DVDs, for example, just ignore the difference between the ITU spec and the MPEG2 spec and flag the full frame as 4:3 DAR (implied 8:9 SAR).

[dellsam34]

Crop to about 704x480 to get rid of the black sides and set the SAR to 10:11.

[Darryl In Canada]

Thank you. When you say set the SAR to 10:11, do you mean the Input PAR or Output PAR (in Hybrid)?

[dellsam34]

I don't use Hybrid, so I'm not familiar with its setting, Try both and see what difference does it make.

[jagabo]

Your source is SAR 10:11. What the output SAR should be depends on whether you resize during your conversion. If you retain the original frame size (or crop to 704x480) set the SAR to 10:11. If you resize to a 4:3 frame size, say 640x480, then set the SAR to 1:1.

[Sharc]

@Darryl In Canada: BTW your ATI screenshot looks interlaced. DON'T resize vertically without prior deinterlacing.
Set the input SAR (aka PAR in Hybrid) as 10:11. When you resize by defining the frame width and hight, the output SAR (aka PAR in Hybrid) is greyed out because the output SAR (aka PAR in Hybrid) is then determined by the resized frame dimensions. Either you resize by means of the input/output SAR (aka PAR in Hybrid) or by setting the resized frame dimensions. In an equation a=b*c one can only select 2 variables independently, not all 3. The 3rd quantity is calculated (or greyed out in Hybrid). You can play with the settings and see what changes in Hybrid.

[Darryl In Canada]

I did five tests with different combinations of input PAR and output PAR. (Click to enlarge table.)


[Attachment 79583 - Click to enlarge]

Things I am still confused about:
- Which Input and Output PAR I should be using in Hybrid from my SD capture card.
- Whether I should force the resize of 704x480, 640x480, or something else.


[Attachment 79584 - Click to enlarge]

[Sharc]

For square pixel output with your cropping and no vertical resizing because your capture is interlaced:

[Sharc]

If you want to leave it anamorphic:

[Sharc]

If you deinterlace you can apply resizing, for example to 1440x1080

[Darryl In Canada]

Sharc, thank you for this. A few more questions.


Assuming the source material are all home video capture as 720x480 with an SD capture device (ATI 600)...
Assuming I deinterlace (QTGMC + BOB), and upscale to 1080p...
And assuming I intend to show my videos on:
(1) HD television screen
(2) computer screen (VLC/MPC)
(3) YouTube

What output settings do people-who-know-what-they-are-doing use? Which of your options should I use?

[Sharc]

See my last example in post#14. It produces a 1440x1080 4:3 frame. Small black borders will be added according to your cropping to make it overall 1440x1080, undistorted picture in a 4:3 canvas, compliant with your (1), (2), (3) above.
(Maybe select Auto adjust 'height to mod 4' to be on the very safe side.)