Who uses a DVD recorder as a line TBC, and what do you use?

[danno78]

What is horizontal jitter anyway?

https://forum.videohelp.com/threads/306272-Computer-video-capture-vs-vcr-to-dvd-combo?p...=1#post1882662

Are there examples of it such as screen shots from a video (like the ones ive posted)?

Short part from a VHS tape with high level of time base errors on both dimensions and a sensitive card at high level of h-synch and v-synch errors.
Without ES15:
http://www.mediafire.com/?mfizanmozmw
With ES15:
http://www.mediafire.com/?mefo3djm3cz

[HDClown]

I have an ES10 which I am using as pass-through from my Toshiba W-808 S-VHS VCR. I am using S-Video. I'm not seeing any major difference in video quality if I pass-through the ES10 or go straight from the VCR. The ES10 is just plugged in and turned on, nothing else. Is this all I need to do with the ES10 to get whatever benefits it may offer?

[deuce8pro]

What is horizontal jitter anyway?

https://forum.videohelp.com/threads/306272-Computer-video-capture-vs-vcr-to-dvd-combo?p...=1#post1882662

Are there examples of it such as screen shots from a video (like the ones ive posted)?

Short part from a VHS tape with high level of time base errors on both dimensions and a sensitive card at high level of h-synch and v-synch errors.
Without ES15:
http://www.mediafire.com/?mfizanmozmw
With ES15:
http://www.mediafire.com/?mefo3djm3cz

Now I understand. Truthfully I didnt know the ES10 or ES15 corrected that kind of wavyness and I own an ES10.

[BrainStorm69]

What I also noticed on the ES10 was the compression of the color palette -- it gets posterized.

I haven't noticed any particular evidence of posterization so far with the ES15 tests. Maybe Panasonic has a better implementation of its chip's capabilities in the ES15 than the ES10. So how do we best test the ES15 for posterization? A capture of a smooth gradient from 0-255 of each of red, green and blue? Is posterization normally largely a function of color depth?

EDIT: Everything I have been able to find searching with Google (none of it official from Panasonic) seems to indicate that both the ES10 and the ES15 have 12-bit video ADCs and DACs.

[danno78]

All the tests done until show no major defect regarding ES15 pass-through performance. There is slight change in color very hard to see with naked eye even with high quality sources (DV source recorded with prosumer camcorder) advanced tests reveal this. I don`t have ES10 but ES15 add a firewire input, nothing more.
Anyway (regardless of device that can induce posterization) this side effect can`t be a issue for a format with 30 lines of color resolution like VHS, S-VHS, Hi8 or Video8.

[BrainStorm69]

I don`t have ES10 but ES15 add a firewire input, nothing more. Anyway (regardless of device that can induce posterization) this side effect can`t be a issue for a format with 30 lines of color resolution like VHS, S-VHS, Hi8 or Video8.

When you say that ES10 and ES15 are the same except for firewire, I thought that the Panasonic chip changed from the MV2 to the MV3. https://forum.videohelp.com/threads/299682-Wavy-video?p=1838490&viewfull=1#post1838490 Clearly added firewire, but do we know that's all it did? I can't find much info at all by Googling.

Also, can you explain the 30 lines of color resolution a little more? Does that mean that the best we could hope for is only 30 different gradients of a single color in a single frame?

[jagabo]

So how do we best test the ES15 for posterization? A capture of a smooth gradient from 0-255 of each of red, green and blue?

I don't think gradients like that are a good test but I ran some on static images of RGB and white gradients. I didn't see any significant posterization. I'll post some samples later.

Is posterization normally largely a function of color depth?

Yes. But it's sometimes indirect and hard to quantify.

can you explain the 30 lines of color resolution a little more?

He's talking about the horizontal color resolution of VHS. Look at the color resolution test charts I posted earlier. The low bandwidth of the chroma signals recorded on VHS tape limit the resolution of colors to somewhere around 30 lines, depending on just how you want to define it. But the resolution is a separate issue from the color depth.

As the frequency of a sine wave approaches the bandwidth limit of an analog circuit there isn't a sudden cutoff at a particular frequency. What happens is the amplitude of the output signal starts decreasing relative to the amplitude of the input signal. One way of defining the bandwidth is to say at what frequency the output amplitude is half the input amplitude. Lets look at the U graph from my VHS cap earlier:



The source video used a ~sine wave that increased in frequency from left to right. If you look at the same graph from the DVD (see earlier post) you'll see that the amplitude of the U signal was almost the same all the way across the graph. But the amplitude of the VHS signal decreases and more or less disappears at the far right.

I marked a peak at which the amplitude is about half what the input amplitude was. The distance from that peak to the next one is about 39 pixels. But that represents two lines (the middle of a bright line, through a dark line, to the middle of the next bright line). So one line of the signal at that frequency covers about 20 pixels. The full image is 720 pixels wide. So that means you could fit 720/20 = 36 lines of that width over the entire frame. But video engineers state horizontal resolution as a function of a portion of the width of the image equal to the height. The full frame is a 4:3 image so the horizontal resoolution is about 36 * 3 / 4, or "27 lines".

[jagabo]

I don't think these full range RGB ramps are useful for looking for posterization but I thought I'd post them anyway. Full RGB 0-255 ramps were converted to YV12 with the usual rec.601 matrix. The final output images are also converted the rec.601 matrix so the range should be RGB 0-255 once again. They were captured on the same Hauppauge PVR-250, calibrated as in earlier posts.

VHS tape:

VHS Y:

VHS U:

VHS V:


ES15 Passthrough:

ES15 Y:

ES15 U:

ES 15:V:


The ES15 reduced noise a bit even though its noise reduction filter was disabled. There is also the same reduction of amplitude in the graphs that we saw with the levels testing. This has caused a slight darkening and loss of purity in the RGB ramps.

The ramps are too steep and pure to see much posterization. I'll be looking for some other tests...

By the way, if anyone wants to repeat these tests with another TBC or DVD recorder I can make the MPG videos available.

[BrainStorm69]

jagabo, thanks for your efforts. I haven't got time to do much at the moment, but please post the MPGs as I just scored an ES20 tonight for $20.

[jagabo]

Here are the m2v files. 23.976 fps progressive with 3:2 pulldown flags.

[BrainStorm69]

Well, I'm glad I only spent $20 on this POS. As you can see from the attached images, the ES20 doesn't have much, if any, ability to straighten wavy lines caused by scanline misalignment. BTW, this was through the back inputs, for those who have seen posts saying the ES20 TBC functionality is only through the inputs on the back.



From JVC 9900 with TBC off



From JVC 9900 TBC off, passed through ES20

Not anything like what the JVC TBC can do, or what I believe I have seen evidenced from the ES10 and ES15. A large disappointment.

Also, it didn't matter what combination of input and out black level setting I used, none were close to transparent. See below comparison of capture of the color bars generated by my TBC-3000 using a variation of jagabo's Avisynth comparison script. Top image is straight to capture card. Next lowest image is through the ES20. Next lowest image is the delta. Lowest image is delta 8x.



input set to "light"; output set to "light" (I assume "light" = 7.5 IRE)



input set to "light"; output set to "dark" (I assume "dark" = 0 IRE)



input set to "dark"; output set to "light"



input set to "dark"; output set to "dark"

Like I said, not too impressive compared to what has come before in this thread about the ES15. I'd say Panasonic's implementation of the LSI chipset is not so good.

jagabo, sorry, but I can't even bring myself to waste time doing all your tests at this point.

[bizzybody]

I have a Panasonic AG-1980 S-VHS deck. It has a line TBC. Some tapes turning that on makes it better, some it makes look worse.

What it doesn't do is prevent over sensitive copy protection sensors from claiming nearly every tape is "copy protected", even ones I recorded myself years ago and have been attempting to digitize to save the video.

Wishing I had my old GeForce 4 AGP card with VIVO and the old drivers that ignored MV. It did a decent capture from composite in, would do better on S-Video.

[lordsmurf]

Can I say "I told you so" yet?

[jagabo]

Here's what I get with the ES15. Similar bars, output from Philips 5990 DVD, with/without ES15 passthrough (input high, output low, noise filter off), to Hauppauge PVR-250:



After adjusting the ES15 passthough with ColorYUV(gain_y=-7, cont_u=12, cont_v=12):



As we've seen before, the ES15 is changing colors and levels slightly and it can be corrected pretty easily. Of course, your typical VHS recording has colors and levels much further off so this is pretty much inconsequential.

[lordsmurf]

With a good proc amp after the DVD recorder, yes, you can sometimes tweak it back better. It really changes tape to tape.

[BrainStorm69]

Can I say "I told you so" yet?

With respect to the ES20, probably so. With respect to the ES15, not so much.

Just a couple of more data points before I go to bed.


Here is the Histogram of the TBC-3000 color bars directly to the capture card.



And here is the histogram of the TBC-3000 color bars through the ES20 with input black level set to darker, and output black level set to lighter (as suggested by an old and respected forum member (GShelley61) some time back as the correct way to set the levels for the ES20)

It appears that the ES20 is messing with the luma at the higher levels, shifting them higher. Luma at the lower levels appears less affected. So the black level setup appears close to correct, but perhaps the contrast has been increased.

[deuce8pro]

And here is the histogram of the TBC-3000 color bars through the ES20 with input black level set to darker, and output black level set to lighter (as suggested by an old and respected forum member (GShelley61) some time back as the correct way to set the levels for the ES20)

Have you tried setting both the input and output black levels to darker? Id like to know what you think if you do. Thats how I did it with the captures I posted.

[BrainStorm69]

Have you tried setting both the input and output black levels to darker? Id like to know what you think if you do. Thats how I did it with the captures I posted.

deuce8pro, as you can see below, that is definitely not the correct way to do it with the ES20. the line for the black bar is now well below 0 IRE as shown by the yellow line in the left brown area. Note that even now the contrast is still too high in this as shown by the yellow line in the brown area on the right.


Color Bars passed through ES20 with input black level set to darker and output black level set to darker.



Color Bars passed through ES20 with input black level set to lighter and output black level set to darker.

I also decided to show this combination as although it isn't the correct setting if you are actually burning to DVD with the ES20, it doesn't really matter since we are just using it for passthrough. As you can see, however, it doesn't quite get the black level right, and there is still too much contrast.


Here is a composite of all three for easier comparison.

[jagabo]

I made a tape with severe time base errors by tapping on the drum as it was recording. Here's a sample field without and with the ES15 passthrough:



On a frame with lesser problems (no tapping, just a crappy VHS deck):



And a GIF animation of several consecutive frames:



Not perfect but certainly an improvement.

The ES15 does appear to be crushing the darkest grays in this test (recorded on a different VCR than the one I was using in the earlier tests).

[BrainStorm69]

OK, I ran a few of jagabo's tests. I created a DVD with his m2v files. Played the DVD with my Sony DVD Player (DVP-NS575P).


first, played with my Sony DVD player straight to the card. I seemed to have the opposite problem from jagabo; Blacks seem to be getting crushed a bit and whites are being clipped.



Then thru the ES230. Notice I was able to capture some of what looks like the too hot white after passing through the ES20. But maybe this is because a large range of the white levels at the top end are being somewhat crushed.


Straight to the card


Thru the ES20. Note again a little contrast boost and the amplitude of the sinusoidal wave is a bit larger.


Straight to card


Thru ES20


Straight to card


Thru ES20

@jagabo - any analysis you can provide would be appreciated. BTW, your post above shows some pretty impressive time base error correction by the ES15.

[jagabo]

Like the ES15, the ES20 appears to be doing a little automatic adjustments of the contrast and saturation. Not too terrible, in fact, probably good for the average consumer who has VHS tapes that are out of whack. But it also seems to blur colors both horizontally and vertically. And your earlier post shows no line time base correction. Of the two, it looks like the ES15 is the one to get.

If you still have the caps, could you post one of the chroma graphs for the low res chroma test pattern (the one with the green/magenta vertical bars)? Use:

VideoScope("BOTH", true, "Y", "U") # or "V" in the last argument.

Thanks!

By the way, I haven't found any evidence of posterization with the ES15 passthrough yet. At least not with the noise reduction disabled. Still looking... With noise reduction enabled I have seen a bit of ghosting.

[deuce8pro]

Have you tried setting both the input and output black levels to darker? Id like to know what you think if you do. Thats how I did it with the captures I posted.

deuce8pro, as you can see below, that is definitely not the correct way to do it.

I think that is the way to go though when you are using a JVC dvd recorder to compensate for the Jvc's brightness.

[BrainStorm69]

If you still have the caps, could you post one of the chroma graphs for the low res chroma test pattern (the one with the green/magenta vertical bars)? Use:

VideoScope("BOTH", true, "Y", "U") # or "V" in the last argument.

Thanks!

By the way, I haven't found any evidence of posterization with the ES15 passthrough yet. At least not with the noise reduction disabled. Still looking... With noise reduction enabled I have seen a bit of ghosting.

Here is the U channel straight to the card.



Here is the U channel thru the ES20.



Here is the V channel straight to the card.



Here is the V channel thru the ES20.

Note one point. When I originally "calibrated" my card for brightness and contrast, I also calibrated it for saturation. I have a vectorscope Directshow filter that I can use with prefiltering in VirtualVCR to actually set saturation in realtime. Normally, when I run my colorbars from the TBC-3000 through it, I have to set my saturation up from "64" to "68" on my card to hit the boxes correctly. With your RGB sweep, I had to set my saturation way down at "51" to get the red, green and blue into their respective boxes (is that because your RGB sweeps go all the way to 100% saturation rather than 75%?).

So when I originally captured for the "Y" screencaps above, the saturation was set at "51" (where the saturation was set shouldn't make a difference for the "Y" channel graph, correct?). I had also made screencaps of the U and V channels at those settings, but I think the low saturation setting caused a problem with them (the amplitude of the waves was too small). So the screencaps above were made with the saturation setting at 64, which is the card's default saturation setting. The brightness and contrast settings were the same as for the "Y" screencaps above. So hopefully this isn't all screwed up to a point where I should do it all over. I did notice that the amplitude of the waves still isn't quite as high as yours are. Maybe I should have set my saturation to "68" (my calibrated setting with my TBC-3000 colorbars) rather than "64" for these caps.

[jagabo]

BrainStorm, thanks for the chroma graphs. It does show that the ES20 is blurring colors more than the ES15. The amplitude of the chroma sine waves is more attenuated at the right of the graph.

Regarding the saturation settings: when I made the chart I wanted the chroma signal to have enough amplitude to measure easily but not to exceed the 16-240 range allowable for U and V. The channels range from about 38 to 217. When you view the graph the sine wave touches the solid lines because the solid green and magenta bars match those peak amplitudes. The U and V channels are both in phase (ie, the peaks and valleys are aligned). The chart is constructed from a grayscale sine wave image, completely in YV12, in AviSynth.

What I'm not sure about is if this results in illegal YUV colors. Ie, there may be more restrictions than just not exceeding the 16-240 range. If you convert the video to RGB and back to YUV the amplitude is much attenuated and the graph is no longer a nice sine wave. U from the original, after ConvertToRGB().ConvertToYUY2():



Not all YUV colors result in valid computer RGB colors:

http://software.intel.com/sites/products/documentation/hpc/ipp/ippi/ippi_ch6/ch6_color...ls.html#fig6-4

So I'm not sure if this is a case where the YUV colors are simply outside the RGB range, or if the colors are inherently illegal as YUV. Maybe I should make a new chart where the colors are within both the RGB and YUV ranges.

[BrainStorm69]

Of the two, it looks like the ES15 is the one to get.

Yeah, I'm about to ask the mods to delete this thread. I've been watching ES15s on ebay since I started this thread, and I think the going price has risen by at least 25%.

Have you tried setting both the input and output black levels to darker? Id like to know what you think if you do. Thats how I did it with the captures I posted.

deuce8pro, as you can see below, that is definitely not the correct way to do it.

I think that is the way to go though when you are using a JVC dvd recorder to compensate for the Jvc's brightness.

It could be if the JVC has no way to lower the incoming signal (normally expected to be 7.5 IRE) to 0 IRE.

[jagabo]

Of the two, it looks like the ES15 is the one to get.

Yeah, I'm about to ask the mods to delete this thread. I've been watching ES15s on ebay since I started this thread, and I think the going price has risen by at least 25%.

Yes, I've noticed ES15 prices going up over the last few years too. Maybe I'll stop recommending it too!

[BrainStorm69]

Of the two, it looks like the ES15 is the one to get.

Yeah, I'm about to ask the mods to delete this thread. I've been watching ES15s on ebay since I started this thread, and I think the going price has risen by at least 25%.

Yes, I've noticed ES15 prices going up over the last few years too. Maybe I'll stop recommending it too!

Well, I won one on ebay today for $71 shipped. About the going rate. Hopefully it's in as good a shape as the seller says it is.

[BrainStorm69]

Well, I got my ES15. Here are a couple of screen shots comparing its ability to correct the waviness in a picture caused by scanlines with time base errors. I would say that its not quite as good as the JVC internal TBC, but pretty darn close.


JVC 9900 internal TBC off


JVC 9900 internal TBC on


JVC 9900 internal TBC off pass-though ES15


JVC 9900 internal TBC off


JVC 9900 internal TBC on


JVC 9900 internal TBC off pass-though ES15

[BrainStorm69]

Just FYI, I "accidentally" won an ebay auction (I didn't expect to get it for the price I bid) for an ES10 that should arrive tomorrow. So if I can find some time this weekend, I may do some comparisons between the ES10 and ES15.

[jagabo]

Cool!