Go Back    Forum > Digital Video > Video Project Help > Project Planning, Workflows

Reply
 
LinkBack Thread Tools
  #1  
02-26-2019, 04:15 AM
AnimeGolem AnimeGolem is offline
Free Member
 
Join Date: Feb 2018
Posts: 4
Thanked 1 Time in 1 Post
Howdy Folks,

I've been reading this site off an on for a couple months now as (I think...) figured out the absolute basics of VHS capture. All of the footage I am working to capture is of animation that was never released outside of the VHS format.

At this point I was hoping to bounce my setup and workflow off of you kind folks to see if I'm on the right track of if there is more I should be trying to tune before I start working through all my tapes.

Currently my chain is as follows;

JVC SR-MV40US (TBC OFF) ---> Panasonic DMR-ES10---> IODATA GV USB2 ----> Losless HuffYUV via VirtualDub

The JVC was a lucky find, it's not in perfect condition mostly likely but it was only 20 dollars at a goodwill. I felt like a kid in a candy store as soon as I saw it! The DMR was ordered off Ebay and has thus far not given me any issues that my eyes are trained enough to see.

I imagine the IODATA is the first thing that's going to raise questions so I'll run through the capture cards I've used.

I started out using a terrible EasyCap device and a non-SVHS player over composite. It was terrible with detail, clipped blacks and only passed audio in the right channel. The less said about that one the better.
The second card I tried was a Hauppauge usb-live2 which captured reasonable detail and passed both audio channels but I could not for the life of me get to not clip blacks. I still have and can test this card.

The 3rd card I tried was a Diamond VC500. This was a pretty darn good card and didn't clip blacks. I suspect it's more sensitive and picking up issues the other cards might be ignoring because occasionally it would pass a hard green bar that extended across the entire image where the other cards would often just show a moving tracking line. I didn't see this when testing today and I have reinstalled windows on this machine since I was primarily using that card, it's possible this was some driver issue. I still have and can test this card.

An example histogram from the VC500:



The final (for now...) card I've purchased is the IODATA GV USB2. I've found this card does not clip blacks or to my very untrained eyes introduce any visual distortion. I've also not had the odd almost rejected frames I was getting with the VC500.

An example histogram from the IODATA GV USB2:



As I've had great luck with this card so far I am happy to do any testing if folks like Lord Smurf or Sanlyn think it would be useful, just let me know what data you'd want.

In order to get started I'm going to attach four short clips of uncompressed footage of the same scene from both the VC500 and the IODATA. I'll make another post shortly about my limited attempts with Avisynth.


Attached Images
File Type: jpg IODATA.jpg (4.1 KB, 167 downloads)
File Type: jpg VS500.jpg (5.8 KB, 161 downloads)
Attached Files
File Type: avi IODATA +ES10.avi (89.42 MB, 22 downloads)
File Type: avi IODATA +JVCTBC..avi (79.38 MB, 32 downloads)
File Type: avi VS500 +ES10.avi (81.14 MB, 23 downloads)
File Type: avi VS500 +JVCTBC.avi (75.84 MB, 20 downloads)
Reply With Quote
The following users thank AnimeGolem for this useful post: archivarious (09-08-2020)
Someday, 12:01 PM
admin's Avatar
Ads / Sponsors
 
Join Date: ∞
Posts: 42
Thanks: ∞
Thanked 42 Times in 42 Posts
  #2  
02-26-2019, 04:58 AM
AnimeGolem AnimeGolem is offline
Free Member
 
Join Date: Feb 2018
Posts: 4
Thanked 1 Time in 1 Post
So far I haven't done anything complicated restoration wise with avisynth, mostly just figured out how to get the thing running and how to do some basic scripts.

Using the above clip 'VS500 +ES10' I've been using something like this;

Quote:
AVISource("G:\VS500 +ES10.avi")
tfm
tdecimate(mode=1)
#----- RGB32 for VirtualDub CamcorderColorDenosie and other RGB work -----*
ConvertToRGB32(interlaced=true,matrix="rec601")
Crop(16,0,-4,-10).AddBorders(10,5,10,5)
LanczosResize(640,480)
return last
I then used Camcorder Color Denoise in VirtualDub at 10 and slightly brought in the levels. If I was making a master I wouldn't use TFM or the final resize. The "web" encode was exported via RGB lagrith prior to being encoded, is this ideal to limit color space changes and to perform them via avisynth?

What more can or should be done with samples like this?

-- merged --

looking at the samples again I think I may have set the IODATA a bit hot in all the switching. I also forgot to mention the JVC has been set to EDIT for all captures.


Attached Files
File Type: mp4 VS500 +ES10 +VDUB.mp4 (6.01 MB, 20 downloads)
File Type: avi VS500 +ES10 +VDUB.avi (60.46 MB, 8 downloads)
Reply With Quote
  #3  
02-26-2019, 09:08 PM
jwillis84 jwillis84 is offline
Free Member
 
Join Date: Aug 2017
Location: College Station, TX
Posts: 800
Thanked 217 Times in 174 Posts
Its up to you, but your kind of doing things "not the way" most people suggest.. if I may be so bold..

The TBC in the JVC-MVS is a line-TBC and corrects for stretching and recoil in the longitudinal distortion of the video tape in the cassette as it travels across the playback drum. This occurs because the winding motor is a feedback servo which pulls and slacks depending on the needs to get it moving and keep it moving from one reel to the next. It can't help but be in error from time to time.. and this produces longer, and shorter horizontal lines across the screen.

The "tearing" or "flagging" common to horizontal problems at the top of each frame (if they are noticeable) is because the leading edge of each frame is dragged across the video heads from the same edge at the bottom of the tape which is scribbled diagonally across the tape from bottom to top. This edge meets the whirling video head and is most likely to "warp" or "actually flail like a flag in the wind" as the video head sweeps up from no where on the tape to the edge of the tape and then diagonally across it. Its this meeting place.. the intersection of a real "edge" and the lack of support of the edge which gives it a chance to be the "most variable" and flip and flap back and forth. Feathering.. or permanently stretching the edge does occur in misaligned transports, sometimes.. and permanently "bakes" this problem into tapes.

A line-TBC grabs a snapshot of each horizontal line and digitizes it, and store it in a line buffer memory for two or three lines, or however much room it has for multiple integers of lines. It "scales" the lines by automatically "stretching" them in digital memory to "fix" the period of time to sweep across the video monitor screen in the same amount of time.. so that there are no "outliers" or "variable line lengths". This prevents "zigzag patterns" from top to bottom in the picture.. sometimes called "squiggles". The more memory for more lines the better.. up to a full field, or a full frame.. but usually a full field is enough.

The ES10 is used in this pipeline as a "frame-TBC" or "synchronizer".. it can try to stablize or correct line length problems.. but by the time the video signal travels through video cables.. its "smeared" just a bit.. so that the information of the true edge of a line is lost and it becomes much more difficult to detect the edge.. so the "strength" of the line-TBC is effectively "lost" and cannot be recovered. Instead the "sychronizer" reconstructs "vertical sync". It basically digitizes each field and strips the poor quality sync signal and slaps on a new copy of a perfect sync signal. It also detects the start and end of each field and frame set and makes sure they are evenly spaced.. so they appear in the output "consistently". At one time this was important for "Genlock" or "splicing in frames from different sources" so that they appears perfectly on screen without jumping around, or "rolling" when switching from one scene source to another.. today its just good for making sure there is no "jitter" or "disturbanced in the [up-and-down] vertical frame sync".

The gyrating output of the VHS player will jump forward and backwards, that's just the nature of a servo-controlled feedback system.. its like a ball on a spring.. but it does "average" around a point.. this is called "hunting" and the frame synchronizer will help smooth this out by acting as a dampener.. sooner or later the signal will arrive tardy, or early all squished or strung out.. and the frame sychronizer (the ES10 in this case) will even all that out.

The ES10 does its job by performing analog to digital conversion, as part of that it has to filter its input. It also has chroma and luma "noise" filters.. and if you are using the composite input of the ES10 it also has a comb filter to minimize dot-crawl. After its converted it to digital, it then converts it back to analog and sends the signal out on its merry way.. hopefully for the better.

The capture device IODATA GV USB2 isn't on the recommended list, so no one will likely have an opinion about it, except that it will be suspect if there are problems. It might be good it might be bad.. but no one knows.

The lossless VirtualDub step is a good one and tells us you are planning to capture without compression, and that's good.. but most capture devices were designed with some type of compression in the output. They often came with custom compression codecs in their hardware or software.. which might compensate for all kinds of unpredictable things (a green "tint" for example, or a tendency to clip blacks or blowout highlights). When capturing losslessly, the "king is without clothes" and all flaws are revealed.. which you then have to be aware of to cover them back up.

So capturing lossless with an Unknown capture device has its own set of unknowns.. so most people will just not comment and wait to see the output.

Its best to nail down what you don't know first, before changing and adding things.

But in general most would recommend gaining some control over your "filters".. and that usually means either by having an on/off switch in the VHS player, or some sort of actual filter level controls. After that point the "frame synchronizer" you use might have a proc-amp.. or some kind of "sharpener" or other type of filter controls.. a proc-amp is just a fancy name for a type of filter.. other kinds are noise filters and color corrector.

Most people would also recommend "do as little damaged as possible" to the signal.. and some follow the idea that "least adjustment in the signal path is best" clean it all up in software after the capture. So collecting lots of devices to insert in the signal path.. is probably a bad idea.. there are diminishing returns to trying too hard.

Mindful .. collecting of gadgets to insert into the signal path, once you've determined a "need" however is not a symptom of hoarding.. and pretty much keeps you from cluttering your mind up with questions (now why did I buy that thing?) or buyers remorse.

Last edited by jwillis84; 02-26-2019 at 09:34 PM.
Reply With Quote
  #4  
02-27-2019, 04:14 AM
AnimeGolem AnimeGolem is offline
Free Member
 
Join Date: Feb 2018
Posts: 4
Thanked 1 Time in 1 Post
Thank you for the detailed reply, it's been one of the more clear explanations of the mechanical functions of the two kinds of TBC that I've read so far.

That said other than potentially adding some kind of sharpener or detailer I am a bit unclear on which part in specific I am doing that is "not the way". Is the recommendation that I capture using both the JVC's line level TBC as well as the ES10? I had avoided that due to reading it caused issued but I attached a quick capture with both on via the VS500. The NR function you mentioned on the ES10 is disabled.

I've not thus far planned to add a proc amp or a detailer of any kind. I'm not opposed to investing more but as you said it's often better to "do as little damage as possible". What are the specific benifits to these vs doing the clean up to the lossless capture after the fact?

You mentioned in regard to the IODATA that I had talked about capturing lossless but most people will wait to see the output. I will just point out that I attached two examples from that card with both the ES10 in the chain and straight out of the JVC in my first post.


Attached Files
File Type: avi VS500 +ES10 +JVCTBC.avi (82.27 MB, 12 downloads)
Reply With Quote
  #5  
02-27-2019, 02:43 PM
jwillis84 jwillis84 is offline
Free Member
 
Join Date: Aug 2017
Location: College Station, TX
Posts: 800
Thanked 217 Times in 174 Posts
Its not meant as a hypercritical review of your pipeline. Its more of an observation that your using a pass-thru as a "frame-TBC". Using an ES10 is good form when (a) its all that is available, (b) the specific tape warrants it and you have to settle for less in order to get any kind of stable viewable picture.

As you noted line-TBCs can have conflicts with ES10, but as a general rule you want "both" if they don't conflict.. because we're talking a line-TBC and pass-thru its hard to give advice.. its more (whatever works).

Many a heated discussion has been had over using an ES10 and accepting it is less than optimal. Its a special tool for specific cases. I have one too.. but don't use it unless it has to be used. Trading off by turning off the line-TBC is a worst case scenario.. I would probably start with the line-TBC "on" and not use the ES10 and see how it goes and keep an eye out for a frame-TBC. I noted for example Lordsmurf just posted he had some more available.. what kind I do not know.. but if its within your budget.. its something to check into.

There's no decision that having an ES10 in the arsenal is a bad choice either.. its quite elegant.. and rare.. its just not something you would use as a first choice.

Another thing to consider is a true "frame-TBC" will defeat true/false Copy protection because that lives in the "sync" area of the frame. Replacing the sync removes even the possibility of a false detection which becomes more common with older tapes as the sync becomes more unstable. I don't recall if the ES10 obeys or disregards Copy Protect.. but it was suppose to.. and since it wasn't designed to avoid it.. it could be vulnerable to passing any existing Copy Protect on through. (I think the ES10 did not pass-thru CP, but did obey CP when recording only to its own HDD).

On the plus side ES10 has great filters... if you need them.. they will be there.. but I don't think you have much control over them. Great is a matter of degree.. in some cases.. they grate more than great. Its a mixed bag of tricks.

Cleaning up in analog space before digital is certainly an advantage to the software "challenged" its more intuitive and baked in.. but it is baked in.. clipped blacks and blown whites can't be corrected in software.. if the signal is out of bounds on the input to the digitizer.. it won't be captured and there will be loss before the lossless capture. So at least for brightness and contrast.. that minimal level of proc-amp care is warranted.. the good thing is most capture devices have a mini software proc-amp that can help.. if you remember to use it.. and you have some confidence in the levels without a meter or scope.. but there again.. the software "challenged" will be at a disadvantage.

I'm sure someone will take a look at your output samples, they are usually appreciated.. it just takes time for the more experienced software inclined of the forum to get around to looking at them.. but... the ES10 kind of doesn't make them happy (shall we say?).. its the static filters you have no control over.. they sort of taint or limit what can be done after the fact. Like I was saying.. the ES10 is (the nuclear option).
Reply With Quote
  #6  
02-27-2019, 07:30 PM
AnimeGolem AnimeGolem is offline
Free Member
 
Join Date: Feb 2018
Posts: 4
Thanked 1 Time in 1 Post
I appreciate the response and I didn't read you as being overly critical --No point in posting all this if I'm not open and looking for what I am doing wrong! I just wasn't clear what I was being recommended to change.

Thanks for the clarification and I do understand now what you mean. I do see the limitations of the ES10 but considering the cost and lack of reliability I've read about in the full TBC's these days it would be pretty hard for me to justify spending 300-500 on one so I may just have to make do.

Thus far on these JP tapes I haven't seen a hint of macrovision so I don't think it really caught on over there during the big boom period. (*Knocks on wood*)

In terms of the capture filter I can say for sure both devices have a mini-proc amp accessible via virtual dub that will let me pull up the histogram and adjust brightness and contrast to keep in withing legal levels. As mentioned in a prior post I was rushing though and ended up setting the IODATA a bit hot.

If I were to look into additional filters is there a direction you would recommend that would address an issue in the samples I've posted? In general if the main advantage is skipping the avisyth learning curve I'm fairly happy to just hit the books for a bit.

Again thanks for all the advice.
Reply With Quote
  #7  
09-17-2020, 04:23 AM
archivarious archivarious is offline
Free Member
 
Join Date: Sep 2020
Posts: 23
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by AnimeGolem View Post

An example histogram from the VC500:

The final (for now...) card I've purchased is the IODATA GV USB2. I've found this card does not clip blacks or to my very untrained eyes introduce any visual distortion. I've also not had the odd almost rejected frames I was getting with the VC500.

An example histogram from the IODATA GV USB2:

As I've had great luck with this card so far I am happy to do any testing if folks like Lord Smurf or Sanlyn think it would be useful, just let me know what data you'd want.

In order to get started I'm going to attach four short clips of uncompressed footage of the same scene from both the VC500 and the IODATA. I'll make another post shortly about my limited attempts with Avisynth.
Thanks for this histogram. What does the difference in smoothness/edginess in the two different histograms mean? I'm learning as I go. I currently have a GV-USB 2, thinking of switching. Before doing so, would be good to be able to know how to read the histograms.
Reply With Quote
  #8  
09-23-2020, 11:00 PM
Hushpower Hushpower is offline
Free Member
 
Join Date: Apr 2020
Posts: 703
Thanked 131 Times in 124 Posts
I have just installed a new GV-USB2 and it works well. I too will be interested in the interpretation of the histograms.
Reply With Quote
Reply




Similar Threads
Thread Thread Starter Forum Replies Last Post
New Guides: Video Workflows for MPEG, AVI and H.264 lordsmurf General Discussion 10 11-21-2013 09:43 PM
Best capture choice-USB device or AGP Card (EZcap vs ATI All In Wonder) Eagleaye Capture, Record, Transfer 33 12-16-2010 07:11 PM
Help me decide on best CompactFlash cards admin Photo Cameras: Buying & Shooting 1 04-13-2009 02:15 PM

Thread Tools



 
All times are GMT -5. The time now is 07:28 AM