USB600 better quality than AIW9600?

[lv2race]

Hi,

I've been comparing captures made at the same time with my AIW9600 card and USB600 stick and noticed the quality (sharper image) is coming from the USB600.

Workflow for AIW9600:
Panasonic AG1980>DataVideo TBC1000>AIW9600 (mounted in a Dell Dimension 4600 PC)

Workflow for USB600:
Panasonic AG1980>DataVideo TBC1000>USB600 (plugged into Windows 7 PC)

The captures happen at the same time with the VCR feeding the TBC1000 via S-video and then sending the signal to the AIW card and USB stick with two sets of S-video cables.

Capture is with VirtualDub(1.9.11) with Huffyuv v2.1.1. The XP machine with the AIW installed is a Pentium 4, 2.40GHz with 2GB RAM running XP SP2. The PC with the USB600 stick is an i7, 3.20GHz, with 24GB RAM (one can never have enough RAM!).

Observations from several captures:

1. The USB capture is crisper (see attached image produced from avisynth script. Image on the left is from AIW9600, image on right is from USB600. The lettering on the car is clearer with the USB device. Capture levels were at default settings for both devices.
2. The AIW seems to "stretch" the image slightly compared to the USB, leaving less black edges around the image
3. From the Vdub capture data:
a. more frames are captured with the AIW for the same length capture (about 10-15 more frames over 3-minutes)
b. Average capture rate is closer to the standard 26.9697 with the USB600 (29.9699 versus 29.9649 on AIW)
c. Data rate during capture is always better with the USB600 by about 600-800 KB/s (8604 with USB, 7789 with AIW)
d. Average frame size is always better with the USB600 (293K pixels versus 266K).
e. Under the Sync category, the USB600 always has 0 for VT adjust, Rel latency, and Current Error. The AIW9600 typically has 4 for Rel Latency and 1-2 for Current Error.

Any thoughts? I thought for sure the AIW 9600 captures would look better but with a side by side image comparison they are noticeably not as sharp.

I'm going to to a test capture and swap the S-video cables in the workflow to see if the cables are causing the difference (just want to eliminate that possibility). Also, I'm going to upgrade the CPU and RAM on the XP machine, but not sure that will help image quality.

Thanks for any input.

[msgohan]

I think your comparison image was crunched down by the forum software. The forum has a maximum width limit. The JPEG we're seeing is 50KB and only 890x297.

By default, the ATI 600 oversharpens, creating halos. You can get neutral sharpness by manually setting Sharpness to 0 in VDub's Levels menu or the Proc Amp tab of the Capture Filter.

Data rate / average frame size being larger isn't necessarily better. Could just be noisier or otherwise less compressible.

[lv2race]

Interesting comments, thanks. I was a little worried the photo would be resized.

I noticed the range of the levels are different between the USB600 and AIW9600. The default sharpness for the USB is 2, with a range of 0-15. The AIW default is 128, with a range of 0-255.

I’ll try a capture with the USB sharpness set to 0.

If sharpness isn’t adequate in the capture, can it be improved through one of the avisynth scripts? I hadn’t noticed any haloing in the capture and have always left sharpness at default thinking if it’s not captured it can’t be recreated. I have the sharpen setting on the VCR set to “soft”.

Thanks again for your input, it’s very helpful.

Rob

[msgohan]

Quote:

Originally Posted by lv2race

I have the sharpen setting on the VCR set to “soft”.

Panasonic AG-1980 sharpness slider all the way to the left? I don't think anyone would recommend that, barring some special circumstance. The closest thing to a neutral setting on the slider is a bit to the left of the center position.

Note that the slider has no effect if Edit mode is turned on.

[lv2race]

Edit mode is on.

[traal]

What does moving the AG-1980's sharpness slider all the way to the left do? Does it actually make the video blurrier than neutral?

I'm experimenting with Topaz Video Enhance AI and one issue is that it sharpens halos, so I'm looking for a way to avoid that. An ounce of prevention is worth a pound of cure, right?