Hi8 capture using Digital8 camcorder - Edge color issues

[memrah]

Hello everyone. I am new here and this is my first message. So.. nice to meet you all.

I have got lots of Video8 and Hi8 tapes I have been wanting to digitize and archive. Most of them are from the 90s and early 2000s. I decided the simplest and trouble-free way of doing the job would be using a Digital8 camcorder, capturing via firewire and storing the footage on my external harddrives as DV-AVI.

I have a Sony DCR-TRV460E and it plays back my old analog tapes fine, except for one weird issue. I have been doing research on it on the web for the past few days and the ONLY reference I could find that was similar to the issue I am having was posted to this very forum by another user a while ago.

My problem is, the right hand side of the videos have a thin vertical bar with a pink / green hue! It changes color depending on the scene. I have noticed that when there is something red behind it, those parts of the bar becomes green. When there is something blueish, like the sky, those parts of the vertical bar becomes a less noticeable pink! Again depending on what the image looks like, sometimes the bar becomes completely invisible. Very weird.

Now, the problem area is actually in the overscan area so, when viewed on a CRT screen, it is not even visible, but since newer display systems have very little to no overscan, and I am planning to share some of these videos to be viewed on computers, I would like to capture the image without the edge color distortion if possible. I tried using an old Hi8 (CCD-TR2000E) camcorder for playback and fed the signal from it to the Digital8 via S-Video, and the edge color issue was not there at all. So I know it is not on the tapes themselves. The problem with this setup is, I get a lot of color bleeding and noise because the Hi8 (TR2000E) camcorder is very old and worn so I'd really prefer using the TRV460E.

Here is the URL to the message that was posted on this forum. (https://forum.videohelp.com/threads/336485-56-hours-of-Hi8-to-digital-project-two-burning-questions) This person describes the exact same problem. The two images he attached also demonstrate the exact same problem I am having. People who responded all focused on his other questions though and other than one person suggesting "Edge color issues seem related to high luma." not much was said about it. I am attaching a couple of frame grabs from my tapes that show the problem even more dramatically. Notice how the red car looks completely GREEN over at the edge of the screen.

Could anyone give me some suggestions or ideas about what can be done to fix it? I have already tried turning the TBC and DNR settings on and off and that doesn't help.

Thanks a lot in advance and have a nice day.

[lordsmurf]

It's probably the camera's fault -- it was shot that way.
Mask and re-center.
It's in the overscan anyway.

[Brad]

Forgive lordsmurf for posting in a flurry and apparently skipping over that one part.

His suggestion is good anyway, if you don't want to play the camcorder lottery and try to find one that fulfills your requirements of no edge color problem and no color bleeding.

I tried using an old Hi8 (CCD-TR2000E) camcorder for playback and fed the signal from it to the Digital8 via S-Video, and the edge color issue was not there at all. So I know it is not on the tapes themselves. The problem with this setup is, I get a lot of color bleeding and noise because the Hi8 (TR2000E) camcorder is very old and worn so I'd really prefer using the TRV460E.

I'm intensely interested in any comparison frames you can offer, as I have an older Hi8 camcorder and have been wondering for years whether I should get a Digital8 and go the DV route or get a better Hi8 and go the external TBC+lossless AVI route.

I've got an eBay search setup for all of the Digital8 models that play Hi8 and have been trying to find ones where the camcorder functionality is broken while the tape and transfer functions work. No luck yet after a few weeks. Of course the mechanical tape transport is always more likely to fail than the electronics.

[memrah]

vaporeon, I can digitize a short section of the same Hi8 tape using my 460E and then the TR2000E via the 460E's passthru using S-video, and post sample frames here. If that would be useful for you for comparison purposes, let me know and I will do that. Keep in mind though that the 460E was purchased (second hand) in 2006 and never been used since, except for a few days of use as a webcam via USB. It looks like it is in great shape. It doesn't look over used or abused. The TR2000E on the other hand was purchased in 1995 (brand new) and has been heavily used for many many hundreds of hours. So it may not be a fair comparison. Let me know.

As for the green / purple band issue, do you think it is the 460E malfunctioning? First I thought the original camcorder that recorded the tapes could be faulty like lordsmurf suggested too, but that is why I did the test with the TR2000E and it turns out the green bar is generated by the 460E during playback. It is not there when the TR2000E is used for playback.

Is there an adjustment somewhere perhaps a hidden service menu in these D8 camcorders that can be used to fix this issue? Can it be bad video heads? I am really curious why or what is causing this. The camera used in that other thread I mentioned in my original post is a Hi8 and it has the same problem as my D8, so I am guessing this is not a Digital8-specific issue either.

What can it be??

[Brad]

Yes please.

I won't even pretend like I have any knowledge from which to base a guess on. I'll just say that the people engineering it probably didn't care about some garbage appearing in the overscan area at the time.

[nic2k4]

I tried using an old Hi8 (CCD-TR2000E) camcorder for playback and fed the signal from it to the Digital8 via S-Video, and the edge color issue was not there at all.

Have you considered the possibility that the D8 analog input might have "filtered out" part of the image (the overscan part)? To be able to say with 100% certainty this is not the case, you need to record a graduated static scene (I'd suggest a yard stick) with the Hi8, then do your comparison. Any masking done through the analog input will be easily spotted.

[memrah]

nic2k4, When I get home, I will post sample frames from both the TR2000E (via TRV460E's passthru) and the TRV460E of the exact same footage from the same tape, for vaporeon to compare. That should also tell us if the 460E is "filtering out" or cropping the image on the original tape. I kinda doubt it, but let's try and see.

[dvd3500]

I used a delogo filter in Virtualdub that used information from the edge to blend in and blur the green stripe.
To be honest, it was almost more distracting than if I had cut it and recentered.

[memrah]

dvd3500: So you have experienced the same problem too? What kind of camcorder/player were you using?

vaporeon: I have captured two copies of three different frames from the same tape for you. One with the 460E and one with the 2000E via the 460E passthru, using S-Video. TBC was turned off on the 460E. The 2000E doesn't have a TBC function. Let me know what you think. I will be capturing another tape in a couple of days, and while doing that I will capture a few frames from that tape too, using both camcorders, this time with the TBC ON on the 460E, so you get to do the comparison with TBC ON too. When I compare the frames side by side, I can't decide which ones look better but I know for sure, the green/pink line the 460E generates really bothers me.

Oh, one last thing: the headswitch noise at the bottom of the picture gets slightly worse when TBC is turned on on the 460E. That is why I turned it off, since the tapes I have don't really have drop outs.

Anyway, I am still desperately trying to figure out a way to get rid of that vertical green/pink bar at the edge of the image so suggestions from anyone would be appreciated.

thank you

[Richard_G]

I've noticed a similar effect with some 8mm and Hi8 tapes played back and transferred over Firewire from a D8 DC-TR8000E but have just accepted that they were put there by the camera that shot them. I've got a real mix of old tapes all shot on different camcorders and it isn't on them all so I've always put it down to the individual camcorder that shot the video. From your samples above, it looks to me like playing back on the TR2000E, feeding that into the TRV460E and using that as a passthrough device is about your best option. You've still got a line along the bottom of each frame but you could always crop he images slightly to get rid of that. Alternatively, leave the errors as they show the age of the footage. You wouldn't try to take the jerkiness out of a Charlie Chaplain film would you? That is part of the authenticity of it.

[jagabo]

The discolored edge probably has something to do with a chroma sharpening filter in the camera creating an over-sharpening halo at the high contrast edge. The V channel gets inverted:



The U channel looks ok:



I would look at extending the V channel to the right. Or maybe inverting and amplifying the chroma and overlaying it via a mask. An example of the former:

ImageSource("redcar460e.jpg") 
ConvertToYUY2(interlaced=true)

U = UtoY() # separate U channel
V = VtoY() # seprate V channel

V = Crop(V,0,0,-16,-0) #remove discolore portion and black border
extra=FlipHorizontal(V).Crop(0,0,12,-0) # the last 12 pixels of the V channel reflected
V=StackHorizontal(V,extra).AddBorders(0,0,4,0,$808080) # add the extra, pad to 720 with grey

YtoUV(U, V, last) # mix the separate U and V with the original Y

This will not work well with colored object that appear only at the edge very edge.

Another variation would be to stretch the last good pixels of chroma rather than reflecting. Change the "extra" line to:

 extra=Crop(V,342,0,2,-0).PointResize(12,V.height)

[memrah]

I've noticed a similar effect with some 8mm and Hi8 tapes played back and transferred over Firewire from a D8 DC-TR8000E but have just accepted that they were put there by the camera that shot them.

Richard_G: I wonder if it is a model specific thing. As far as I know the TR8000E is one of the earlier D8 models and you say it has the same ocasional edge discoloration habit. Again as far as I know the TRV460E is one of the late D8 camcorders Sony made. Looks like they didn't bother to do anything to fix the issue. (Perhaps because it is totally invisible on over-scanning CRT screens) Or perhaps not all TRV460E camcorders do this and mine is misadjusted internally? I was considering buying another D8 camcorder on Ebay (a TR7000E) but now I am wondering if that one may also have the same issue. By the way, similar to your experience, not all parts of the tapes have the colored bar. Look at the flower picture for example. There is no color issues there.

The discolored edge probably has something to do with a chroma sharpening filter in the camera creating an over-sharpening halo at the high contrast edge. The V channel gets inverted

jagabo: Thanks a lot for examining the frames and the AviSynth script. I wish there was an easy way to readjust that chroma sharpening filter via a hidden service menu or something. Extremely unlikely, I know. How would "inverting and amplifying the chroma and overlaying it via a mask" work on scenes where there is no edge color problems? (Like the frame with the flowers) Would it then unnecessarily invert the edge, thus introducing the exact same problem that it is attempting to solve?

Also, could you tell me what you think of the TR2000E captures in terms of overall quality? They look slightly darker and the chroma level seems a tiny bit different, but like I said before the TR20000E is so old and has been used so heavily that I keep thinking the signal from it can't be very good anymore. If it were you would you go the AviSynth route, or would you just use the TR2000E instead like Richard_G suggested? (or.. attempt to buy another camcorder on Ebay?)

[jagabo]

How would "inverting and amplifying the chroma and overlaying it via a mask" work on scenes where there is no edge color problems?

From what I've seen it wouldn't be a problem. In shots that don't have the problem there is no color at the edge. You would be inverting and amplifying zero (literally 128). So you still have zero, no change in color.

Also, could you tell me what you think of the TR2000E captures in terms of overall quality? They look slightly darker and the chroma level seems a tiny bit different

Overall I think the TR2000E is doing better. It doesn't have oversharpening halos. Look at the vertical posts in the sunset scene. That problem at the right edge also appears other places in the picture. Again, in the sunset scene you can see discolorations around the vertical posts. The black levels are a little darker in the TR2000E but not to the point where it's a problem (no crushed blacks). You can always adjust them with filtering. Both have blown out brights. The only issue I see with the TR2000E is it may have a little more line time base jitter. I'd go with the TF2000E.

Inverting V and overlaying with a mask leaves a miscolored strip at the joint:

mask=ImageSource("mask.png").Crop(0,0,-2,-0).AddBorders(2,0,0,0).ConvertToYUY2(matrix="PC.601").Blur(1.0).Blur(1.0).Blur(1.0).Blur(1.0).Blur(1.0).Blur(1.0).Blur(1.0).Blur(1.0).Blur(1.0).Blur(1.0).Blur(1.0).Blur(1.0)
ImageSource("redcar460e.jpg").ConvertToYUY2(interlaced=true)

U = UtoY()
V = VtoY().Invert().ColorYUV(cont_y=300)
inv=YtoUV(U, V, last)

Overlay(last,inv,0,0,mask)

VideoScope("bottom", true, "V", "V")

You might be able to play around with the settings but I don't think you'll get a perfect result.

[memrah]

Jagabo, thanks a lot for the feedback and the information you shared. I really appreciate it.

[Brad]

Thanks for the comparison images! I don't think I have anything to add to the expertise offered already in response.

My desire for a Digital8 passthru model is swiftly falling. I have enough problems with overblown levels on my VHS captures where I actually have modest proc amp controls; I don't want to be stuck with one baked-in setting on my Hi8 tapes.

[memrah]

vaporeon, here is another one, this time with the TBC ON, on the TRV460E.

[Brad]

The little one's belly looks noisier on the 460, to me.

At this point short video samples would be more useful than stills though.

[memrah]

I would upload samples, but DV files are huge. Even a minute of it would probably be too big to attach to messages here.

I noticed the TR2000E seems to have a little bit more noise in red and green colored areas of the picture. Can someone tell me what the correct technical term is for the color crawl that is visible in very bright reds and greens? Does it count as color bleed too?

[jagabo]

I noticed the TR2000E seems to have a little bit more noise in red and green colored areas of the picture. Can someone tell me what the correct technical term is for the color crawl that is visible in very bright reds and greens? Does it count as color bleed too?

I think you're talking about color bleed (or blur) but it looks to me it's more a matter of the TR2000E capturing with more color saturation -- so the colors bleed is more obvious.

[memrah]

Color bleed is when the color of an object like a red rose looks like it is smeared to the outside of the actual boundaries of that object right?

Is it also called color bleed when the color stays perfectly within the boundaries of the object, but there is this faint, almost transparent chroma noise inside that area? (Most visible with red)

[Brad]

I would upload samples, but DV files are huge. Even a minute of it would probably be too big to attach to messages here.

The limit per AVI file is 100MB, which is 27 seconds for PAL DV.

You can attach multiple files, for whatever reason. Don't know if there is any upper bound. I did 500MB in one post.

[jagabo]

Color bleed is when the color of an object like a red rose looks like it is smeared to the outside of the actual boundaries of that object right?

Yes. It's caused by passing a signal through a low pass filter. That causes it to spread out and shift temporally, to the right in the case of VHS.

Is it also called color bleed when the color stays perfectly within the boundaries of the object, but there is this faint, almost transparent chroma noise inside that area? (Most visible with red)

No, that's chroma noise.

[memrah]

No, that's chroma noise.

OK, then, that's what I am seeing more of with the 2000E (not "color bleed") and you are probably right about that being because the 2000E is outputting the video with more saturation.

Thanks Jagabo.

[memrah]

Jagabo, I just thought about this; Can the DNR also be playing a role in reducing the chroma noise of the 460E output? I am not sure if the TBC and DNR functions on these D8 camcorders work only for analog tape playback or for the analog->digital passthrough function as well. I kinda doubt that they work in passthrough mode. So that may be another reason why I am seeing more chroma noise with the TR2000E.

Do worn video heads cause increased chroma noise? I just can not believe that this 15 year old, heavily used TR2000E is producing better playback results than the 460E I bought in 2006, when it couldn't be more than 2 years old. I haven't used it at all since 2006 either. I guess the TR2000E was one of the better models of its time.

Vaporeon, I didn't think it was possible to upload 100MB attachments. I will be capturing more tapes in a few days, so I will upload some 25 second samples then.

PS. Prior to buying the TR2000E, I had a CCD-V600E, and I really liked its picture quality. I had so many mechanical/tape transport issues with it though, I ended up selling after a couple of years. I was excited about the TR2000E but when I watched the first few hours of footage I shot with it, I was kinda disappointed because I thought its picture quality didn't match the V600E at all.

[jagabo]

Can the DNR also be playing a role in reducing the chroma noise of the 460E output?

Yes. Noise is pretty easily fixed with software filtering. Things like over sharpening halos and ghosting are harder to fix.

Do worn video heads cause increased chroma noise?

I don't know.

I didn't think it was possible to upload 100MB attachments. I will be capturing more tapes in a few days, so I will upload some 25 second samples then.

You can easily cut/paste segments together with VirtualDub with no reencoding (Video -> Direct Stream Copy). Usually a few seconds per shot are enough.

[Brad]

I just can not believe that this 15 year old, heavily used TR2000E is producing better playback results than the 460E I bought in 2006

My 2003 VHS VCR is worse than decks made in the 80s, so it isn't really so surprising.

[memrah]

OK, here is a couple of clips from the same tape, captured using the TR2000E (via 460E's passthru) and the TRV460E

Let me know what you think.


PS. I am going with Jagabo's suggestion and using the TR2000E for digitizing my tapes, by the way. I may end up buying another Digital8 camcorder or player that doesn't have the edge color issue, if I can find one locally. I don't want to buy one on Ebay just to find out that it has the same issue(s). I need to be able to test it and make sure that it doesn't do the edge color thing, and that it performs at least as good as my good old TR2000E.

PPS. This person used a GV-D200 to digitize analog Video8 and I don't see any of the analog playback problems I encountered with my TRV460E in this video: http://www.youtube.com/watch?v=6aJJVd32qnM&feature=player_embedded

PPPS. Vaporeon, I have a feeling that 80s VCR that outperforms your 2003 VCR has a lot more use on it than the 2003 one too.

[Brad]

I think you may be focusing too heavily on "wear and tear". There are mechanical devices inside VCRs, but they are also electronic equipment. The parts that are too far gone could be swapped out from another unit. Don't be scared off by the salesmanship fear tactic of "get your VHS tapes converted before your VCR dies and everyone else's does too!".

Many thanks for the samples.

[Brad]

I definitely prefer the 2000E image.

[jagabo]

I think the 2000e capture is better overall. It has better black/bright levels. It may have a little more chroma noise or maybe it just shows up more because the the better levels. It's a little less sharp but you can easily adjust that with a horizontal sharpening filter. And of course, it doesn't have that big color inversion problem at the right edge.

Compare the sample video with this:

v1=AviSource("compare460e.avi") 
v2=AviSource("compare2000e.avi") 

SeparateFields(v1)
Trim(1,0)
v1=StackVertical(SelectEven(),SelectOdd()).Subtitle("460e")

SeparateFields(v2)
v2=StackVertical(SelectEven(),SelectOdd()).Trim(6,0).Subtitle("2000e")

StackHorizontal(v1,v2)