GV-USB2 vs. ATI TV Wonder 600 USB

[Darryl In Canada]

Sharing a clip from the bad tape portion of my ongoing GV-USB2 vs. ATI TV Wonder HD 600 USB comparison.

This tape was recorded in EP (SLP) mode in 1985 off broadcast television. It's my worst tape in the collection – heavy waviness and flagging at the top of the frame – which is exactly why I use it as a stress test. If a capture device handles this tape well, it handles most things well.

Setup details:

• Player: Panasonic PV-VS4821-K (S-VHS) – no line TBC on this model
• Connection: S-Video out to each capture device
• OS: Windows 7
• Capture software: AmarecTV
• Codec: HuffYUV

This clip is part of a full 5-minute test. AmarecTV reported zero duplicate or dropped frames across the entire capture on both devices.
Post-processing was minimal by design. I deinterlaced using QTGMC at its least destructive settings via VCG Deinterlacer and output to ProRes 422. No denoising, no colour correction, nothing else. The two MOV files were placed side by side in a Premiere Pro timeline and exported as H.264 (MP4).

Based on what you can see in these two clips, the GV-USB2 handles this tape more resiliently than the ATI. This likely comes down to what's inside the device. A hardware teardown by Taruru (https://ikidomarinosaki.fc2.net/blog-entry-61.html) found that the GV-USB2 contains a Renesas (formerly Techwell) TW9910 10-bit video decoder, a Cirrus Logic CS5340 audio decoder, and a Syntek STK1150 USB bridge. The TW9910 datasheet specifies "advanced synchronization processing and sync detection for handling non-standard and weak signals."

I'm not an expert, but based on things I've read, I believe the following breakdown is accurate. If I've got anything wrong, please correct me.

There are three tiers of signal correction you can apply to an unstable tape:

PLL (phase-locked loop) – Built into most decent decoder chips, including the TW9910. Re-clocks the sync pulses on each line but doesn't touch picture content. This is what the GV-USB2 has. It's better than nothing – and better than what's in most cheap dongles – but it has a ceiling.

Line TBC / DVEC – Goes further by correcting the timing of the actual video lines, not just the sync pulses. Handles flagging, velocity errors, and mild jitter. Found in better S-VHS decks and some standalone TBC units. Sony's equivalent is DVEC (Digital Velocity Error Correction), built into MiniDV camcorders and the DVMC-DA2 – it does the same line-level velocity correction on the analog input path before DV encoding. The JVC line TBC has an edge over DVEC on badly degraded tapes because it also integrates DNR, but for typical consumer VHS they're in the same tier.

Full-frame TBC – Buffers an entire frame and reconstructs it from scratch. The heavy artillery. Handles severe instability, vertical roll, and the worst-case tapes. Standalone units like the DataVideo TBC-1000 operate at this level.

A few observations and questions:

1. Overall the GVUSB2 performs better, but there is at least one instance where the ATI600 performs better.
2. Often, a scene changes causes the waviness top go away. What is the reason for this?
3. The final commercial is very static with no camera movement and so both devices perform well. Why does movement trigger waviness?
4. What else do you notice about the two videos that points to differences in the way each digitizes?

[Attachment 92080 - Click to enlarge]

[Attachment 92081 - Click to enlarge]

[Attachment 92082 - Click to enlarge]

[Sharc]

Well, I think we agree that both are useless for such pathological cases, and any "conclusions" are doubtful.
Look for example your video where one is readily inclined to "conclude" that the GV-USB2 excels the ATI600. Now look at the pylon in the background: The GV-USB2 shows it stably but badly tilted (like after a thunderstorm), while the ATI 600 shows it straight upright. So which one is better? Comparing bullshit with horseshit