Test: Corel Dazzle vs. I-O Data GV-USB2 vs. Diamond VC500 vs. Dazzle DVC100

[Bwaak]

Testing four analog video capture USB dongles, three of them are currently available as new, the fourth is 18 years old.

Devices tested:

• Corel Dazzle DVD Recorder HD Video Capture Device
  • Barcode label name: DVCPTENAM
  • Name on the device: Dazzle Video Capture USB v1.0
  • Year on the box: 2019
  • Year in the operating instructions: 2013
  • Regular price: $50 (2022, early 2023), $70 (late 2023)
  • Sale price: $30 (Black Friday 2022), $33, $39
• I-O Data GV-USB2 Video/VHS DVD Dubbing PC Ingestion Video Capture
  • Name on the device: GV-USB2
  • Price: $50-$60
• Diamond Multimedia VC500 One Touch VHS to Digital File converter
  • Name on the device: One Touch Video Capture VC500
  • Year on the box: 2018
  • Price: $30-$38
  • Sale price: $26 (in 2022)
• Corel Dazzle DVD Recorder Video Capture Device
  • Name on the device: HW-SET DVC100 Rev.1.1 | Pinnacle a division of Avid
  • Price: $10-$25 on eBay (the same models with different packaging and software are DVC101, DVC103, DVC107)
  • Original release year: 2006

Used VirtualDub2 and AmarecTV for capturing on Windows 7 64-bit.

Watch on YouTube.

Attached frame grabs from VDub after applying IVTC filter. Captured off a VHS tape protected with Macrovision using an SVHS VCR, no TBC.

---

EDIT 1: Added some video samples.

GV-USB-Videoletter-SVideo-Shooting.avi
GV-USB-Videoletter-SVideo-Temple.avi

DVCPTENAM-Videoletter-SVideo-Shooting.avi
DVCPTENAM-Videoletter-SVideo-Temple.avi

VC500-Videoletter-SVideo-Shooting.avi
VC500-Videoletter-SVideo-Temple.avi

DVC100-Videoletter-SVideo-Shooting.avi
DVC100-Videoletter-SVideo-Temple.avi

[Sharc]

Watching on YT: All horrible quality IMO. Did YT butcher it like this? Extra long play tape recordings? Dupes? Overdenoised (wax) and oversharpened (halos, e.g. birthday scene). How did you deinterlace?

[lollo]

I was right then about Bwaak identity when he first arrived and started to post https://forum.videohelp.com/threads/408770-Do-digitizing-analog-video-tapes-companies-...=1#post2684572

I preferred you as ConsumerDV, actually.

[Bwaak]

Deinterlaced in VDub2 with MSU Deinterlacer. I know, I know, probably should have used QTGMC. Then again, QTGMC is known for noise reduction, so I suppose what you are seeing is a more raw look

The first tape is JVC's companion tape to the VHS-C adapter, I have it digitized here.
The highly unstable tape came from eBay, recorded in 1995 on SVHS-C cassette in SVHS mode.
"Doc Hollywood" is a normal VHS cassette with Macrovision.

[Sharc]

The comparison should be based on lossless captures (interlaced for VHS tapes) only. Otherwise we test and compare anything but the quality of the capture process and devices.

[Bwaak]

The comparison should be based on lossless captures (interlaced for VHS tapes) only. Otherwise we test and compare anything but the quality of the capture process and devices.

Interlaced cannot be watched on YT, I could do single field deinterlace of course, but then it would look even uglier. I will upload some interlaced clips a bit later, is Cineform ok with you or, unless it is truly lossless, it is unacceptable?

Some things can be noted even from the YT video, I did not put my notes in the video, hoping for the discussion in the comments .

[jagabo]

Nothing should be watched on youtube if you're trying to show anything about video quality. You should not be deinterlacing, resizing, denoising, or any other processing. Just provide the lossless ITU caps. And you should calibrate each capture devices so that the levels, saturation, and hue are close to spec. Turn off all "auto" proc amp features and set the sharpness and noise reduction controls to neutral (often hard to determine). Comparing at default settings is useless.

[Bwaak]

Added six clips, see "EDIT 1" section in the original post.

Nothing can be watched on youtube if you're trying to show anything about video quality.

I do not agree with this extreme point of view. Certainly, something can be gained even from an incorrectly-deinterlaced video with wrong aspect ratio and frame rate. In particular, depending on one's tolerance to certain artifacts and on a use case, one or two dongles from this comparison should be disqualified, so there would be no reason to tune them further.

You should not be deinterlacing, resizing, or any other processing. Just provide the lossless ITU caps.

I attached six clips to the first message, feel free to take a look. All in Cineform. I am planning to add more later.

And you should calibrate each capture devices so that the levels, saturation, and hue are close to spec. Turn off all "auto" proc amp features and set the sharpness control to neutral (often hard to determine). Comparing at default settings is useless.

I compared them with default settings in three use cases - good tape with built-in TBC, decent tape with Macrovision and with no TBC, bad tape with no TBC - played on the same equipment and deinterlaced with the same deinterlacer. They definitely can be compared. It took me a week to make this video, maybe I will create an addendum with the optimal settings, based on the comments in this thread. Thanks!

[Bwaak]

So, the VC500 abruptly changes brightness, a known defect that I confirmed, see two pictures attached with three frames difference between them. Also, the VC500 cannot handle unstable video and requires a TBC.

The current Dazzle "DVCPTENAM" is better with unstable video, but is confused with Macrovision (distortion on the top of the frame), so not fit for capturing Hollywood movies protected with Macrovision without a Macrovision-defeat device. Also, when a bright scene changes to a dark scene, it drops the brightness during several frames, not immediately. And, it brightens up dark scenes way too much.

The DVC 100 shows fine dithered noise, which sometimes looks like a cross-hatch pattern on still frame grabs, even via SVideo. Not visible when video plays. It is also brighter than the VC500 and the GV-USB2, but not as drastic as the DVCPTENAM. Otherwise, average in every regard.

The GV-USB2 is reasonably tolerable to unstable video, seems to have correct levels for bright and dark scenes without abrupt changes, shows no patterned noise, does not mind Macrovision. Hence, it is the winner.

[Sharc]

Nothing can be watched on youtube if you're trying to show anything about video quality.

I do not agree with this extreme point of view. Certainly, something can be gained even from an incorrectly-deinterlaced video with wrong aspect ratio and frame rate. In particular, depending on one's tolerance to certain artifacts and on a use case, ....

And that's where the debate starts and people may draw wrong conlusions on the dongles when there are too many variables in the equation. What if one says that the (initial) brightness fluctuation of the VC500 doesn't bother him?
Btw., apart from the initial "calibration step" (AGC adjustment) the VC500 (left picture) and the GV-USB2 (right picture) seem to behave very similar, see attachment.
(More annoying to me are the halos of the source. Have these been introduced by the editing of the original video?)

I imagine that the AGC of the VC500 may be advantageous when the dongle is connected to different VCRs/different cable lengths, preventing overload of the ADC without requiring manual adjustments or external attenuators - unless it acts too nervously. I don't have a VC500 so I can't really tell.


[Attachment 76431 - Click to enlarge]

[Bwaak]

I imagine that the AGC of the VC500 may be advantageous when the dongle is connected to different VCRs/different cable lengths, preventing overload of the ADC without requiring manual adjustments or external attenuators - unless it acts too nervously. I don't have a VC500 so I can't really tell.

You don't need to own the VC500 to access whether it acts too nervously - this is what I made the YT video for, and put the waveform along with the original videos. In the first test you can see unwarranted jumps of video levels at 7:10, 7:25, 7:42, 8:01. It is not attenuation, it is an increase. Sometimes you cannot see it without the meter, other times it is noticeable.

I agree that the VC500 behaves very similarly to the GV-USB2, but the brightness jumps disqualify it for me. It is also bad in handling unstable video, but I suppose for most forum members this is not a criteria, as a TBC is considered a must.

You may also take a notice how the Corel Dazzle "DVCPTENAM" behaves on scene changes - it takes several frames to adjust the levels. See the scene change at 7:45 - how the levels drop and almost bounce during several frames, and you can see the picture flashing. To me, this would be another disqualified dongle.

(More annoying to me are the halos of the source. Have these been introduced by the editing of the original video?)

I did nothing aside of deinterlacing. Did not adust procamp before capture, did not adjust levels in NLE. I have a machine that I know produces cleaner output with less objectionable halos, I can try using it. Then again, the point of the test was to compare the dongles, not to get the best picture out of them.

[Sharc]

In the first test you can see unwarranted jumps of video levels at 7:10, 7:25, 7:42, 8:01. It is not attenuation, it is an increase.

Yes, I noticed the step up of the luma at about 1s into my clip as well, see the waveform. I assumed it is an inital auto-calibration to exploit the luma range 16...235 (0....255 including peaks) and/or to set the levels for the AD converter (which would require an anlog pre amp or a controllable analog attenuator at the dongle's input though). If it does such adjustments repeatedly (nervously) throughout the video it is questionable of course.

[Bwaak]

I think these pictures show how levels differ depending on the brightness of the scene. On the bright scene with the airplane the levels are similar, and in fact DVC100 is brighter than DVCPTENAM. On the dark scene with the temple both Dazzle units have significantly higher black level. I actually like how the DVC100 lifts blacks just a bit, the question is, which one is correct.

On the Levels-Trinket.jpg image you can discern a white object in the lower right corner on the DVCPTENAM video.