DV Pix - Image Defects due to Tape Problems
Pix: Sampling
Pix: Artifacts
>Pix: Defects<
  Pix: Generations
Pix: Codecs
 
Text: The DV Format
Sony dropout 
Single-frame Dropout demo image
(c) 1998 Origin8 Video
A single-frame dropout of 32 horizontal pixels by 8 vertical pixels. It occurred on a Sony miniDV tape from October 1995, shot in a VX1000. This dropout appears every time the tape is played in a Sony DCR-VX1000 or DHR-1000, but oddly enough it does not appear when the tape is played in a JVC GR-DV1u -- the only case I know of in which the GR-DV1u outputs a superior picture! Perhaps this particular defect is due to an obscure bug in the Sony codec, causing a "dropout" to appear where one isn't actually present? In close-up this patch is non-uniform; it consists of differently colored pixels, but does not appear to replicate any other set of pixels in the image. 
Panasonic dropout
Single-frame Dropout demo image #2
(c) 1998 Origin8 Video
Another single-frame dropout. This one, measuring 32 horizontal by 20 vertical pixels minus a 16 x 4 chunk, is on a Panasonic miniDV tape shot in a VX1000; it's the worst "real" single-frame dropout I've seen in three years of shooting. I've had other similar dropouts occur in playback due to debris on the tape, but replaying the same scene again worked without any visible dropout. (This image is 2/3 of the way through a dissolve, explaining the odd, ghostlike superimposed pictures.)
525/59.94 banding
10-stripe image banding demo
(c) 1998 Origin8 Video
banding during recording
10-stripe image banding demo
(c) 1998, 2000 MeetsTheEye, Inc.
used by permission
Banding: this defect occurs when one of the two heads on the scanner clogs or fails. In this example, the VTR is starting to play through a section of damaged tape (the tape became jammed in a camcorder and a portion was creased and crumpled). As the tape plays, bands "freeze" as one of the heads ceases to reproduce clean data, so whatever was last placed in the frame buffer remains onscreen while the other head continues to play. In this example, allowing the playback to continue usually causes the remaining "live" bands to freeze as the most heavily damaged section of tape is reached; as the damaged section passes, one set of bands starts to show "live" data again, followed by the entire picture going live. Other times, head cleaning is necessary to restore a fully-live picture. This sample is from a 525/59.94 ("NTSC") tape, so 10 bands are present. In 625/50 ("PAL"), 12 bands would be seen. 

I have only had this sort of noticeable banding occur once when recording. A head clogged momentarily, and I had five frozen bands for about three seconds in the middle of a take. Apparently a speck of debris (such as residual ME tape coating) stuck to the head, then wore away.

A single frame of this sort of banding occurs fairly often, perhaps as often as once every tape or two, but is hardly noticeable with most picture content in which there is little motion from one frame to the next to call attention to it. It's less of a problem with higher-quality tapes, and I've never seen it on DVCAM or D-7.

Multiple Dropout
Sample of multiple dropouts
(c) 1998, 2000 MeetsTheEye, Inc.
used by permission
Other dropout: following the head clog described above, the frozen bands "digitally dissolved" out, resulting in the mosaic of dropouts as shown here during the "dissolve". At this point, the head clog was mild enough that some correct data were getting to tape; only intermittent failures were occurring as the speck of debris was being polished away by the passage of the tape.

 

Technical Details: images were shot on a Sony DCR-VX1000 3-chip DV camcorder. The tapes were played in a DHR-1000 DV VTR and captured digitally using the DPS Spark (Adaptec AHA-8940) IEEE-1394 I/O card on a P133 Wintel machine. The resulting AVI video clips were opened in Adobe Premiere 4.2 and the desired frames were exported as 24-bit Windows BMP files using the Adaptec DVSoft codec; the BMPs were then read into Adobe Photoshop LE. Images were downsized to 180x120 using bicubic interpolation and saved as high-quality JPEGs to preserve visual quality; these JPEGs are very close in appearance to the original images. Images are shown with square pixels, although for a proper 4:3 aspect ratio these 720x480 images should have pixels that are 12.5% taller than they are wide.

Copyright (c) 1998, 2000 by Adam J. Wilt.
You are granted a nonexclusive right to duplicate, print, link to, frame, or otherwise repurpose this material,
as long as all authorship, ownership and copyright information is preserved and a link to this site is displayed.


Contact me via email

 
 
 
 

Last updated 20 August 2000.