I have JVC S9800U, Panasonic AG5710 and Panasonic AG1980 here. I am trying to figure out which one to use for finishing up my VHS capture project for the second time. The JVC has quite a bit of missing detail, no matter what settings I use on it, so I've ruled that one out. The three different types of VHS tapes I have all play better on the Panasonics. The VHS tapes I have are from an old 1980's VCR, a 90's compact VHS recorder, and one tape from a "professional".
The primary issue I don't understand is that my 5710 seems to capture stuff in between the frames captured from the S9800U and 1980. Can anyone give me some insight as to why this is, or how? I have spent days trying to figure out which VCR to use, comparing shots, and looking into reasons for behaviors such as the frame differences.
Both Panasonics seem to be performing similarly, other than the framing differences.
I am using an ATI USB 600 capture device, and these last few days, VirtualDub is set to record to YUV2 at an excessive 720x480 at 30fps. In the below shots the JVC was set to sharpen, the Panasonics set to the middle of the smooth-sharpen slider. The JVC was just a blurry smudge with any type of textures and sharpen was the best it could do to retain detail. Virtualdub isn't inserting or dropping frames according to its stats.
Middle image below is showing a different frame. This happens for every frame - it shows framing in between the other two VCRs. It happens on all tapes. The frames themselves are different, as easily seen in other video clips with more motion - the perspective/distortion between foreground and background objects is different.
I also did spot one frame in another video, only appearing on the 1980, that appeared a bit weird. The other VCRs did not show this frame, but rather the complete frame before and after it, each of which used half of this frame:
Any help would be appreciated. I want to get this figured out so I can start getting these things recorded, and remove all these VCRs off my desk. They have been there for over 6 months!
+ Reply to Thread
Results 1 to 27 of 27
Last edited by Masejoer; 2nd Sep 2014 at 23:06.
You should upload images at the capture frame size, not resized to full screen or some other size. Better yet, upload short, losslessly compressed, overlapping video samples.
You should never enable the sharpen filter on a VHS deck -- they all produce severe oversharpening halos. Also disable noise filters, they're very crude and usually lead to ghosting. Denoise and sharpen in software later.
200% blown-up was simply to try to compare detail. The video is too small on newer monitors for me to effectively compare - 13" screen at 3200x1800 needs all the help it can get.
I'll have to figure out some video editing before being able to upload short clips. I don't have anything but Virtualdub, and I haven't touched it for editing. I've only been trying to digitize these things - figured editing would happen later, if ever.
I'll have to figure out some video editing before being able to upload short clips.
huffyuv can't compress to YUV2. I take it that "YUV2" is a typo. The correct term is YUY2. Your capture frame rate should be 29.97, not "30fps". Capturing at the wrong frame rate might explain a few glitches you described. If your final output objective is DVD or standard definition BluRay/AVCHD, the per-spec frame size for those final delivery formats is 720x480. If you want PC-only or Web playback, capture at 640x480. You'll have to resize the latter if you also want DVD, etc. In both cases you'll have side borders to contend with.
VirtualDub to make frame captures directly from the video, or make direct frame captures by using a media player's capture or snapshot menu.
When it comes to individual frame glitches that you mention, remember that you're working with VHS. A ton of variables are involved during playback, as well as variables you can credit to the way media players render the input and upsample for your monitor.
The AG-1980 and AG-5710 are twins, pretty much. Both are good performers but both have capacitor aging problems and have to be periodically maintained by pro shops that specialize in that sort of stuff. One outfit that's often recommended is http://www.tgrantphoto.com/sales/index.php/categories/repair-services. Don't expect maintaining a pro VCR to be a cheap proposition. The 9800 requires pro maintenance as well, like any VCR.
There's nothing inherently "bad" about the Pannies or the JVC 9800. If less aggressive noise reduction is what you're after, then the JVC isn't what you want. The 9800 is a nice player but it's a poor choice for slow 6-hour or 4-hour tapes. Its dnr can make a really noisy tape look soft. The 9800 is best set to "EDIT" mode for capturing VHS. With the Panasonics, most people would set sharpness at the left-hand or center position. The "dnr/normal" switch that's supposed to turn off the Panny's feature doesn't really do anything. According to techs it's a switch added at the last minute but wasn't implemented in production. I prefer the Pannies, as they have fewer motion artifacts in their internal processing. But when set up correctly for capture, many prefer the 9800. It's up to you. And it depends on the tape.
Last edited by LMotlow; 3rd Sep 2014 at 10:00.- My sister Ann's brother
1) Video -> Direct Stream Copy (just copy compressed frames from the input to the output)
2) File -> Open Video File
3) Edit -> Set Selection Start (Mark-in) and Edit -> Set Selection End (Mark-out) to mark the portion you want to save
4) File -> Save as AVI.
Repeat steps 3 and 4 to save other clips from the same video, repeat steps 2 through 4 for other source videos.
So if I understand this correctly, on the Panasonics, if I set the smoothing slider all the way to the left, it doesn't do anything but lower the unit's sharpening further? The JVC should be set to normal for less processing, correct?
I can re-capture this evening if this is indeed the case. I'm simply looking for the purist's approach in capturing the original video source. I've tried playing with contrast, but I can't get any more detail than what my USB stick's default settings provide - washout doesn't improve.
Virtual dub, or Virtualdub MPEG a decade ago to do some basic processing of video clips, but it has been ages. All I remembered was sticking some type of frame markers in a spot, removing beginning to that position, and doing the opposite to the end. I used to use Virtualdub to edit clips of recordings I had from SageTV.
VirtualDub for capture, use its capture menu's "Levels" filters to hook into your graphics card or capture device image controls. Don't add additional filters in other menus, as many of them will convert input to RGB -- not something you want during capture. VirtualDub's capture util also has a histogram feature to help keep incoming levels inside the RGB 16-235 video range. What you want is brightness and contrast settings that keep most of the video out of the "red" areas on each side of the histogram. Those levels will fluctuate during play, so adjust for the worst-case scenario for your tape. Further tweaks and stuff like saturation, denoising, etc., are much more flexible with software after capture.
The current version of VirtualDub is 1.10.4 . You can use earlier versions of you want, but VDubMPEG and VDubMod haven't been updated in a decade and some newer vDub filters require later versions.- My sister Ann's brother
3-1jpg: low contrast detail has been lost (over aggressive noise filter?). Gigantic oversharpening halos. Brights are the least crushed. Slightly crushed darks. Those may be fixable via the capture device's proc amp. More saturated colors (can be adjusted via proc amp or software).
3-2.jpg: low contrast details retained. Less oversharpening halos. Some crushed brights. Black level too high (should be fixable via capture device's proc amp, definitely fixable with software). Crushed brights (may be fixable in the capture device's proc amp).
3-3.jpg: low contrast details retained. It's a tossup whether those details are better than 3-2. More crushing of brights (maybe fixable in the capture device's proc amp).
At first glance I'd say it's a tossup between 3-2 and 3-3. But one would need to see video samples and new caps with the proc amp calibrated.
You seem to be implying there's some magical setting at maybe around 1/4 where both Sharpen and Smooth are zeroed. I don't believe that to be the case.
No I don't think there's a magic set point, but I've just left mine in the middle forever. I've seen varied testimony on this, but so far I haven't had a noise or edge enhancement problem. Thanks for the tip, though, I'll keep my eyes open for any effects. I'm just accustomed to noisy VHS, I guess, no matter what you do to it.- My sister Ann's brother
Hehe, noisy VHS. Yeah, it's a fact of life. I'd still take the noise from a Panasonic-produced capture over the mess made from a JVC capture. And the noise can be tamed in AviSynth. It's a shame that with the 1980, to use the TBC you have to also use the DNR. But with mine the two are independent, a much better solution.
Ah, the sharpening slider. As of this week I have been asking family and visitors what image they preferred (movement of the slider) as what I prefer is not always what other people prefer (when making them DVD copies of tape material). Almost unanimous was peoples liking for the sharpening to be at maximum! Whilst I saw halos and ringing they see a crisp image with 'definition'. I think its how we have trained our eyes - anything not sharp is 'faulty' and I was this way too once. There is a point on the slider where the image sharpens up a little but not enough to show halos - on my Panasonic PAL VCR's this is about 1/4 - 1/3 of the way between full left and centre detent. That is where I leave it for most captures.
Most people can't see worth a damn unless what they see is in-your-face effects. Seriously. Look at the way most TVs are set up in what is called Volcano Mode. If you want to sharpen stuff, get a cleaner capture without halos and ringing or other junk, then later use some good Avisynth sharpeners that avoid all that garbage. Sharpening during capture makes tape noise and other analog defects all but impossible to clean up.- My sister Ann's brother
Other capture devices allow brighter input levels to be captured, as they don't clip until they hit 255 and lowering their Contrast settings do recover brights.
That said, it may be the camera that produced bloomed whites. If you posted some short AVI samples it would be easy to check for hard, flat digital clipping.
Yeah, I am gathering video clips right now. I'm going to use wood grain texture as a comparison subject.
There are a couple good comparison frames (for detail) in the following clips - and most of these VHS tapes are shaky, amateur shots. I set the JVC to "edit", panasonics to full "smooth" and "normal". All I've ever found has been mixed opinions on auto/edit and normal/edit modes for VCR's, so I am still as confused as I was a year ago on the topic of which settings provide the purest/unprocessed recording of what are on the cassettes. It seems the JVC is quite "sharp" when in edit mode.
When comparing different VCR's, it seems randomly the three different units will display a frame that is different. At the very beginning of these three clips, the JVC is showing the different framing. In this same area, the 5710 matches the 1980's frames - different than what I was seeing in the other ~1990 clip.
TBC enabled on all three. This tape has some waviness at the top without it enabled.
Panasonic 1980 Clip
Panasonic 5710 Clip
JVC S9800 Clip
JVC Settings #1
JVC Settings #2
The following clip has the camcorder placed on a solid object, so it is steady. There is good detail and some color in-frame.
Panasonic 1980 Clip #2
Panasonic 5710 Clip #2
JVC S9800 Clip #2
Last edited by Masejoer; 7th Sep 2014 at 16:54.
I take it these were captured with the Contrast set lower than default? The samples are hard-clipped to a luma range of 16-198 according to ColorYUV(analyze=true).
I have no idea, sorry.
Thanks for the samples. These are my comments based on clip #1. Early in the clip are the most obvious differences IMO; the JVC capture doesn't bloom as much so portions of the house that are washed away in the other two can actually be seen.
I took a screenshot showing the other things that irk me. I aligned the JVC to the others by adding 4 pixels to the left of it, and trimming one field from the start of both Panasonic captures. Odd-numbered fields still aren't aligned vertically; I just chose to show an even field rather than align those as well.
You can see some side effects of the temporal DNR: as the camera pans left, 2-3 ghost images of the wall shadow are visible and objects lack color until it catches up because in the previous frames those pixel positions were white. The AG-5710 chroma DNR looks different than the AG-1980's.
Meanwhile, the JVC has some black halos and noisy edges.
I think you'll get better results from the JVC. Although the time base jitter is a little worse it's not too bad. And the halos are minimal. It flickers less that the others too.
Last edited by jagabo; 8th Sep 2014 at 00:15.
The need for desk space is making me try to finish this up. I can't tell what is oversharpening, versus more detail, or what else to look for. I just want to get a final recording of all these tapes and be done with this project.
I have 5 clips taken below, each with different settings for normal versus detail processing. All are set to "edit" mode on the VCRs.
What are peoples' opinions on the best to keep? The last four have obvious chroma noise still appearing.
Last edited by Masejoer; 26th Oct 2014 at 17:46.
My impression is clips 2-5 were all played in the same VCR whereas the first clip was played in another. Is that correct?
The first clip wins in terms of, well...everything: detail, noise, contrast range, saturation. In other words, it looks most natural and most faithful to the source in my opinion.
For me it's no question which one I'd use – the first one.
I tried to look for detail, but not sure I see the first one providing any more of it? I swear in some places it looks like there is less detail, but noise can also lie.
Clips 4 and 5 are the worst. The brights are horribly squashed in the analog domain (in the first few frames, her right shoulder and face). In clip 3, they are digitally clipped, and to a lesser extent in clip 2. Clip 1 is still missing the very tops of the peaks for the shoulder highlights, due to the ATI 600's limitations. (One reason I continue to recommend the Diamond VC500.)
Clip 3 preserves the very darkest parts of the video more than the rest, but there doesn't seem to be anything of value lost anyway. The slight black crush in clip 1 aids it in looking more contrasty.
The chroma noise is worst on 5, almost as bad on 4, reasonably suppressed on 2 & 3, and virtually eliminated on 1 with the help of blurring (the red fabric on the floor is visibly smudged, but to a minor degree).
Clip 1 loses some detail in the wood grain, present in all the others. Her eyes look brown in clip 1 and possibly green in the rest...?
Overall I would go with clip 1, since clipped brights are far more noticeable in motion than some lost detail I noticed by comparing single frames like a maniac.
Thanks for the info! I have no idea how you got the earlier images, but those told me the most.
The eyes are green, not brown. Obviously, some differences there.
Clip 1 matched up with what you said earlier, and what the "chroma" view showed - the JVC seems to generally do better.
The last four are 1980 with sharpness turned up, 1980 with middle of the slider, 5710 with sharpness turned up, 5710 with middle of the slider. The Panasonics seem to blur a lot when the sharpness slider is moves more to the left.
I'll have to look into the VC500 real quick. I only purchased the ATi 600 due to the overwhelming recommendations on the forums. Not sure what the VC500 may lose over the 600, but if it'll help, I'll grab a unit before I start the mass recordings.