Video clips from 2 capture devices, 1 better than other?

[stevevid]

The attached video clips are from USB-based VC500 and PCIE-based HVR-2250 capture devices. The workflow was with S-video from a Mitsubishi HD-2000U vcr --> capture device --> VirtualDub. The vcr had its 3D-Y/C and DNR+TBC functions turned on, and no filters were used in VD.

Before capture, I initially cropped the images to get rid of the black sidebars and head switching noise so I could set the brightness and contrast levels to keep the video levels within range and make each clip look about the same. I REMOVED the cropping (unlike I did for another thread) and captured roughly the same video with each device.

Is there an appreciable difference between the two captures? I don't see any.

Test VC500 capture - Scene 8 short clip.avi

Test 2250 capture - Scene 8 short clip.avi

I notice a couple of issues in both clips--ghosting and chroma bleeding/smearing. This shows up on items like the red ladder. Question, what is the difference between chroma bleeding and smearing? To see if the cabling was causing the ghosting, I changed the cables and did not see any change. What other issues are in the clips?

I captured these clips directly using VirtualDub's stop condition. I had it stop at 95MB. The actual size was a bit more--98+MB for 10 to 13 seconds of video. Initially, I tried VD's editing capability to extract 10 seconds of video. These clips were much larger--3 to 4 times as large as the direct capture. Why were the extracted clips larger for the same video length?

I'm studying several different and excellent threads on this site about filtering to get the best image. Unfortunately, a few threads have links to sites that no longer exist or have been taken over by malware. For example, the neuron2.net site now links to malware. I'm glad that at least part of the material was saved here as a text file. The rest appears to be lost to history. Also, an older recommended tutorial on color management has been taken over (I don't remember the site at this time). I sent another bad link to this site's admin. Should I continue to do this or just ignore the bad links?

Thanks,
Steve

Update: sharpening was set to 0 for each capture.

[lordsmurf]

You know what, now that I'm looking for it, I do see some odd AGC type brightening and dimming. It's especially noticeable when the girl is in the swing. When her bright self gets in the image, there's the smallest twinge of AGC at work. I'd have never noticed that without comparative samples, and someone point-blank asking about it. But I do see it, you're not nuts.

Now that 2250 card is extremely touchy to timing errors, obvious jitter (up/down laymen jitter),

"Bleed" refers to too many things. Specifically, what you see is chroma offset, easily fixed in Avisynth. After offset is fixed, then you'll see bleed.

You don't have smearing, that refers to something else.

Larger clips probably due to not settings the compression. Are you referring to the Huffyuv that were attached?

I see no ghosting.

neuron2 aka Donald Graft (now user "videoh" at doom9) had some sort of public kerfuffle some years ago. At the time, I was still mostly bedridden and in daily physical therapy, so it was at least a year later that I learn about it. He has his own little forum now, and let the old site and domain get taken over by squatters. For a time, I halfway remember his new domain having been used for some weird spiritual quest sort of thing, but looking at it now I see no trace of that. The Archive.org Wayback still has much of the old neuron2 site.

The 2250 has oversharpening mosquito noise.

[stevevid]

Quote:

Originally Posted by lordsmurf

You know what, now that I'm looking for it, I do see some odd AGC type brightening and dimming. It's especially noticeable when the girl is in the swing. When her bright self gets in the image, there's the smallest twinge of AGC at work. I'd have never noticed that without comparative samples, and someone point-blank asking about it. But I do see it, you're not nuts.

Well at least someone thinks I'm not nuts. My wife would disagree.

Quote:

Originally Posted by lordsmurf

Now that 2250 card is extremely touchy to timing errors, obvious jitter (up/down laymen jitter),

Please tell me where you see the jitter so I know what to look for.

Quote:

Originally Posted by lordsmurf

"Bleed" refers to too many things. Specifically, what you see is chroma offset, easily fixed in Avisynth. After offset is fixed, then you'll see bleed.

You don't have smearing, that refers to something else.

Larger clips probably due to not settings the compression. Are you referring to the Huffyuv that were attached?

Thanks for the info. I did multiple tests with the VC500 and the 2250 to see if one produced more of what I thought was bleed (about burned a hole in the tape). So I assume the offset is caused by the camera or VCR--true?

I found what caused the large file size. I did not set VD to direct screen copy. I had full processing on.

Quote:

Originally Posted by lordsmurf

I see no ghosting.

neuron2 aka Donald Graft (now user "videoh" at doom9) had some sort of public kerfuffle some years ago. At the time, I was still mostly bedridden and in daily physical therapy, so it was at least a year later that I learn about it. He has his own little forum now, and let the old site and domain get taken over by squatters. For a time, I halfway remember his new domain having been used for some weird spiritual quest sort of thing, but looking at it now I see no trace of that. The Archive.org Wayback still has much of the old neuron2 site.

Thanks for the info on Donald. I'll check out his new site.

(BTW, I see ghosts all the time after watching Ghost Busters. Maybe that's why my wife thinks I'm nuts.)

Quote:

Originally Posted by lordsmurf

The 2250 has oversharpening mosquito noise.

My fault. I set sharpening to full (level 15). At the time, I thought the image looked better. What are the signs of oversharpening? Does oversharpening cause or just make mosquito noise worse?

Thank you very much,
Steve

[lordsmurf]

I though the jitter, aka rapid vertical/up-and-down bouncing, was obvious.

You mean Ghost Busters with the ape, or Ghostbusters that warned you to never cross the stream? Because they're not the same! Very different shows, and technically different show names (with space and without). The "with ape" had a good cartoon run in the 80s, and the live-action in the 70s had Larry Storch. You can catch a few eps on Youtube. Gilligan's Island type humor.

[stevevid]

I only saw the original movie, none of the other recreations. I guess I'm a purist when it comes to ghosts.

[sanlyn]

In asking which capture is "better' than another, your sample unfortunately compares two different sets of frames from the same tape. So it isn't a frame by frame comparison. The VC500 seems to have less interlace combing (relatively) and cleaner colors. They look very similar bun they don't look "alike". The 2250 has a little more blue displacement and bleed along the house's rooftop. The 2250 looks a little "darker" because the Vc500 clip has slightly higher black levels, visibly brighter gamma (midtones) and has slight bright overrun beyond y=235.

The VC500 huffyuv YUY2 clip has 329 frames, is 10.98 seconds playing time, has 720x480 NTSC frames, and plays at the correct NTSC frame rate of 29.97 fps.

The 2250 huffyuv YUY2 clip has 345 frames, is 13.8 seconds playing time, has 720x480 NTSC frames, and plays at the incorrect PAL frame rate of 25.00 fps.

[stevevid]

PAL??? I checked several 2250 short clips and yep they are 25fps. Not sure why because the full two-hour recording is 29.97. I guess I fell asleep when doing the short captures. It's additionaly odd since I didn't have trouble setting the proper fps for the 2250 as opposed to getting the VC500 to work at 29.97. With the VC500, I had to follow a different, convoluted setup order than what the DigitalFAQ guide shows to do a capture at 29.97 fps .

I think he clips don't fully cover the same same frames because at the time I posted the clips I hadn't progressed to capturing longer clips then picking just the sections I wanted. I captured each clip by setting up the stop condition in VirtualDub before starting the capture and the vcr. Very inefficient and my tape positioning wasn't very accurate. I now know how to get matching clips.

Thanks Sanlyn for the feedback on what you saw in the clips.
Steve

[lordsmurf]

You have to verify VirtualDub settings in detail every time you capture. Then verify the capture file once done.

I didn't even notice the fps was wrong (but I also wasn't looking for temporal errors).