DV Quality - Check This Out

[J. Baker]

Ok, so I was fiddling with my Optura10 and noticed a increase in quality that I hadn't seen before. Basically I recorded two videos from my cam directly into WinDV via firewire. One with the switch in normal "Tape Mode" and the other switched to "Card Mode". The "Card Mode" had a signification increase in quality. As you might know, "Card Mode" when recorded onto the cam itself is only 320x240 or 160x120, on my cam anyway. But if you hook it up to the pc via firewire and have it in "Card Mode" you get the full 720x480.

So here are the png images below, taken from VirtualDub, with no other processing. The first one is taken from "Tape Mode" which is normal when you are recording a DV. And the second is from "Card Mode".

The first one as you can see has more noise, faded color, and of course there were scan lines which you don't see here. The second has less noise (focus on jacket), better color, wider view for some reason, and it had no interlacing in the video. Viewing the uncompressed videos on my pc showed just how much of a difference there was but you should be able to get a since from the two images.

[edDV]

I'm not following. Did you record to tape or flash card?

Did you record "wide" or 4:3 ?

Both appear to be recorded 720x480 4:3 with conversion to progressive and 0-255 luminance scaling. Image 1 is zoomed.

A more objective screen cap would be from VLC (deinterlace = disable) from the DV-AVI file.

[J. Baker]

I'm not following. Did you record to tape or flash card?

Did you record "wide" or 4:3 ?

Both appear to be recorded 720x480 4:3 with conversion to progressive and 0-255 luminance scaling. Image 1 is zoomed.

A more objective screen cap would be from VLC (deinterlace = disable) from the DV-AVI file.

1.) Neither were recorded to tape or card. I hooked up my cam via firewire and recorded straight to WinDV. I just adjusted the switch on my cam to either select "Tape Mode" or "Card Mode". As you can see, they have different effects.

2.) 4:3

3.) The first one had interlacing but the frame I saved as a png didn't show any interlacing. The second one had no interlacing at all.

4.) I didn't adjust the luminance at all on either one. No processing was done at all. Images save straight from VirtualDub.

5.) As for the zooming. I didn't use any zooming. For some reason when I switch my cam to "Card Mode" it gets a wider shot.

[edDV]

4) Virtual dub converts YCbCr DV to RGB and rescales luma from 16-235 to 0-255 and inn the process clips off any white over 235.


I don't have that model camcorder so can't explain why the DV 4:3 mode is zoomed. My guess is in photo 4:3 mode, more of the sensor is being sampled. That doesn't mean you see higher resolution since it gets scaled back to 720x480 for transfer over IEEE-1394. Photo mode will be deinterlaced but that should show more choppy motion.

[J. Baker]

4) Virtual dub converts YCbCr DV to RGB and rescales luma from 16-235 to 0-255 and inn the process clips off any white over 235.


I don't have that model camcorder so can't explain why the DV 4:3 mode is zoomed. My guess is in photo 4:3 mode, more of the sensor is being sampled. That doesn't mean you see higher resolution since it gets scaled back to 720x480 for transfer over IEEE-1394. Photo mode will be deinterlaced but that should show more choppy motion.

I had thought about photo mode using more of the censor. But just didn't understand why DV mode wouldn't as well?

As for photo mode being choppy. It is when you record a 320x240 video directly to the card itself. But when having it photo mode and hooked up directly to WinDV via firewire like I did, it wasn't choppy at all.

Now, as for the photo mode being less noisy then DV mode, I wonder if that has to do with it using more of the censor like you said?

And now for photo mode being more colorful then DV mode, I'm guessing it uses the full 0-255.

Now I just wish I could record to tape in photo mode as there are some advantages. But it doesn't allow for that, card only. And my cam only does 320x240 15 fps max for 10 seconds at a time.

[JohnnyMalaria]

My guess would be electronic image stabilization. In video mode, this works by using a portion of the total image sensed. As the image vibrates etc, the electronics track the motion and select that particular region from the full image. It works quite well but introduces noise when active (at least based on my Sony's that have such stabilization rather than optical). In photo mode, there's no need for stabilization so the whole sensor is used. Also, the shutter speed may be slower compared to the video case thereby further reducing noise.

One of my early Sony DV camcorders has the electronic stabilization and, since the sensor was oversized, it could also record true 16:9 but with the restriction that the stabilization had to be turned off. The PDX10 also has a large sensor that permits significantly higher resolution photo images than for the video mode.

[dun4cheap]

If I remember correctly the optura 10 and 20 did not have true 16:9 however when it 16:9 it uses more of the actual coming from the lens then the 4:3. I had ran some tests long ago with my old 20 which I still have but never use and noticed this long ago. The optura series cameras had some decent optics.