DSR-450WS still life

[Peter J DeCrescenzo]

I hope the following images are helpful.

 

All of these images are from my NTSC DSR-450WS 2/3" 16:9 DVCAM camcorder, grabbed via Firewire @ DV25 onto my PDX10 camcorder's memory stick which saves them as 640 x 480 JPEG files. Other than resizing them to 640 x 360 and saving them as JPEG files @ 98% quality, I haven't changed them.

 

For all the images the DSR-450WS was set to 0db gain, 5600K electronic preset color balance, 1/4ND, with the Canon YJ19x9B4-KRS lens focal length at approx. 15mm (which is approx. equiv. to a 50mm focal length for a 35mm photo cam).

 

In all the images I exposed the MacBeth "skin" color chip (the chip in the top row, 2nd from the left) at 70 IRE using the cam's zebras. As a result, the white color chip was slightly underexposed; in other words, it didn't trigger the cam's 100 IRE zebras unless I opened the iris very slightly. Instead I opted for the skin chip to be at 70 IRE.

 

The first image is the result of the DSR-450WS being set to its factory default settings, including its detail functions turned on and set at 0 (zero), etc. The cam is in 60i interlaced mode, 1/60th sec. iris at F5.6.

 

The next image also uses the cam's default settings, except I switched it into 24p progressive mode, with a 1/48th sec. shutter. As a result I had to open the iris to F4 to maintain the skin color chip at 70 IRE.

 

Next I enabled some of the cam's various film-look features, including one of its film gamma settings, lowered detail level, etc. The values I used are the ones listed at the end of the article on Simon W.'s website. (Neither Simon or I propose that these settings are ideal, but are only an example of what's possible.)

http://www.simonwyndham.co.uk/camerasetup.htm

 

The final image uses the same settings as the previous one, except I've enabled the MTX LINEAR feature and set all the user values to -55 which results in B&W video.

 

Those were carrots sticks, a small tomato and a nectarine in the foreground. I'm nibbling on the carrot sticks now. :)

 

All the best,

 

- Peter DeCrescenzo

[Alvin Pingol]

Interesting.

 

The interlaced image almost appears sharper than the 24P version, which upon close inspection appears to have slightly stronger sharpening. I know you kept the other settings identical, so this is a little odd. Were your pulldown settings correct for the 24P frame? I'm seeing heavy aliasing on near-horizontal lines, such as on the edges of the wooden table and plate, which is much less visible on the interlace frame.

 

The low edge enhancement on the film look frame helps a lot, but I'm not seeing a substantial increase in dynamic range. In fact, it almost seems as if highlights are being sacrificed faster; Highlight detail on the label of the Gerolsteiner bottle is harder to see, compared to the default settings frames.

[Phil Rhodes]

Hi,

 

> I'm seeing heavy aliasing on near-horizontal lines

 

Hmm, me too - what gives?

 

Phil

[Peter J DeCrescenzo]

> I'm seeing heavy aliasing on near-horizontal lines

 

Hmm, me too - what gives?

<{POST_SNAPBACK}>

Hi Alvin & Phil,

 

Sorry I did reply sooner; I've been very distracted -- or should I say appalled -- by the news coming out of the US GUlf Coast this week. Heaven help all the folks suffering there. But I'll try to keep this thread about video cameras.

 

Yes, I see the aliasing you're referring to in the images, too. I just checked the original frame grabs to confirm I presented them here in the correct order. I did, and they exhibit the same issue.

 

I wonder if the way I'm grabbing these images is causing the aliasing?

 

I'll do another test soon (this weekend I hope) and I'll pay more attention to what I see "live" on the video monitor compared to what I get in the frame grabs.

 

Thanks for the heads-up.

 

- Peter DeCrescenzo

[Phil Rhodes]

Hi,

 

How're you resampling these things for display? Try posting an unadulterated frame grab (yes, we understand it'll be all tall and thin, but for diagnostic purposes...)

 

Phil

[Peter J DeCrescenzo]

The original frame grabs have the same artifact. Here's the original "default" 60i image:

 

... and here's the original "default" 24p image:

 

The PDX10 I'm using to grab the frames might (?) be introducing the aliasing in the progressive frames, but I'll investigate this further.

 

- Peter

[Phil Rhodes]

Hi,

 

Well clearly something's going on as they're only 640 pixels wide. There's resampling happening somewhere, probably in your memory stick image capture process.

 

Seems like we really need to be looking at tape grabs, and preferably TIFF files thereof or a single-frame AVI or Quicktime movie.

 

Phil

[Alvin Pingol]

I agree that there is resampling happening somewhere in the chain, but am still puzzled about the aliasing. While the frame width is smaller than it's supposed to be (which is most likely the capture device rescaling for square PAR), the frame height is correct, and that's what really matters - What's even more confusing to me is that the 60i frame does not suffer from the aliasing, so it can't be the most obvious (and most common) problem of capture software automatially deinterlacing anything it can get its hands on. Perhaps it's the PDX10 not properly interpreting 3:2 pulldown?