New capture setup, input on AVT-8710 captures?
 

Thanks to everyone on this forum, I've learned a ton, and have started over with an entirely new capture setup for my VHS home movies. (And, most importantly, I am no longer capturing to DV!)

The new setup uses a Windows XP capture system:
3.40 GHz Pentium 4 HT running XP Pro SP2
ATI AIW 7500 AGP
AVT-8710 TBC (the older, green one)
Turtle Beach Santa Cruz sound card (internally wired to AIW)
JVC SR-V101US (NORM, TBC/NR:on, Video Calibration:off, Digital R3:off, Video Stabilizer:off)
Mono-price 3ft S-Video cables
C-P7U VHS-C adapter (since almost all the movies are VHS-C tapes)

The setup seems to work well.

I am capturing via VirtualDub 1.10.4, using Lagarith YUY2. (I went Lagarith over HuffYUV, because the HuffYUV files do not seem to work in Premiere Elements.) I chose to go with a lossless encoding, because I have the disk space, and I'd like to keep the lossless files in case I ever do some more advanced restoration in the future. Not getting dropped frames, so seems all good.

I think I'm finally on the right path here. I do have a few questions related to the new setup:

1. I am definitely getting a less pleasant image through the AVT-8710. (Almost like there's a bit of a greyness to it.) I've seen postings here and elsewhere regarding the AVT-8710's default settings, and have seen just as many posts saying to keep the default settings on the TBC as there are to adjust the brightness and contrast... so wanted to get some input from you guys.

Without the AVT-8710:
Attachment 6180

With the AVT-8710:
Attachment 6181

The difference is easier to see when you can flip through the images on top of each other.

Is it expected that with the right settings, the AVT-8710's output should match (or close to match) the picture quality of a capture that didn't use it?

2. Almost all of the recordings I am going to capture are from the same camcorder, and seem to have little if any vertical jitter. There is some jitter on a few tapes, which is why I wanted to get a good external TBC. But, if the answer to #1 is that you always do have to sacrifice a little in picture quality to get the clean signal from the TBC, then do most of you guys selectively use the TBC where needed? (Instead of using it always.)

3. My current work-flow plan is:
Capture AVI in VirtualDub with Lagarith YUY2 encoding
Possibly apply some AviSynth filters (TBD)
Basic editing in Adobe Premiere Elements

If I want to make some DVDs for family, then I'll follow one of the DVD workflows in the guide here.

But, my primary playback is going to be streaming from a server onto an HD TV.

lordsmurf mentioned in another post that H.264 is the best MP4-wrapper streaming format and supports interlaced, but that devices rarely use it (and that I'd likely need to make a deinterlaced version for streaming [progressive, non-interlaced] use).

However, I will be streaming using an Oppo BDP-103D. This blu-ray player supports streaming MP4 H.264:
http://wiki.oppodigital.com/index.ph...s_supported.3F

Would this be the suggested way to encode them, then? Or would something else be suggested?

Thanks, again!

The values on the non-TBC image look hot, crushed, and possibly illegal. That can be a side effect of a TBC. It has to do with timing and AGC. Since the timing now corrected, AGC isn't allowed to do whatever it wants.

Also be sure to hold reset on each value. Perhaps one got changed?

Capture tapes archival, make streaming copy from archival master.
My suggestion =
- MPEG-2 15mbps to store on disk
- reduced bitrate MPEG-2 for DVD
- QTGMC-deinterlaced H.264 MP4 to use on internet/network

All external tbc's affect the image in some way -- e.g., the TBC-1000 tends to soften. My AVT raises gamma somewhat. This is usually corrected using VirtualDub capture's "Levels..." dialog, which hooks into the ATI 7500's proc amp. The AVT proc amp controls work, but they're a little clunky and difficult to adjust precisely. Use VDub capture's histogram to check your levels in "Preview" mode. Turn it off when you start capturing.

I used VirtualDub's simple brightness/contrast filter to adjust the "with AVT" image to have the same histogram as the "Without" image, as it works similarly to the 7500's proc amp brightness and contrast control. Brightness controls the black levels, contrast controls the brights. I increased contrast about 15% to 20% and lowered brightness about 10%.

Without AVT, orginal:
http://www.digitalfaq.com/forum/atta...1&d=1462896122

With AVT, adjusted:
http://www.digitalfaq.com/forum/atta...1&d=1462896135

Note that tte originals posted earlier are lossy jpg, which affects the results posted here. I agree with lordsmurf as the original images look a bit hot in the brights and jpg compression produces some posterization effects on gradients.

In the long run the difference is often deceptive, as you should always check levels before capture and will have to tweak later anyway. Most tapes will be slightly different, some will be very different.

I use my AVT with captures except those that are made with a non-tbc VCR and a Panasonic DMR-ES10 tbc pass thru which has decent frame sync capability. I always always use the AVT for macrovision, and use it when the ES10's frame sync isn't effective enough.

I don't know what level of "edit" you intend for Premiere Elements, but Adobe doesn't handle UY2 or RGB conversions very well. Premiere Elements works in RGB. I suggest you make that conversion with Avisynth.

Thank you for your responses!

I made sure all the settings on the TBC were reset, and re-captured. The result was the same.
I also captured an outdoor image from another tape for comparison (this time .png exports from VirtualDub):

Without the AVT-8710:
Attachment 6197

With the AVT-8710 (after making sure all settings are reset):
Attachment 6198

Without the AVT-8710:
Attachment 6199

With the AVT-8710 (after making sure all settings are reset):
Attachment 6200

After seeing both pairs, do you still think the "with" pictures show a truer picture?


Thanks for the pointer. Still learning how to properly adjust an image based on the histogram, but I will try using VirtualDub to do that when needed. (I'll spend some time learning how and when to adjust the levels properly before continuing with the project.)

But, I also just wanted to make sure that the TBC wasn't doing something harmful/unexpected here.


Looks much closer! But, from the sounds of it, I might be better off not trying to make it match... since the "with AVT" image was a truer image? Now that the attached images are lossless, do you think the "with AVT" image needs level adjustment?


Just cutting and splicing (and adding some effects). I didn't quite follow what you said here, though. Are you saying that Premiere Elements is converting YUY2 to RGB to use the file, and doesn't do it very well (and so I should convert to RGB in Avisynth first)? Elements seems to open and work the files fine, and didn't give any indication that it needed to convert anything... but I can't seem to find anything that indicates it is still working in the original encoding either.

Thanks for the help here!

Was thinking about this more. I can understand that when using the TBC, its own AGC is in effect (regulating the input signal to 1 V). And, since I'd assume the TBC corrects the timing beforehand... I'm guessing that helps reign in how much AGC regulation is needed to achieve the desired output?

But, for the non-TBC, there is no AGC present, right? Or, is there AGC happening in the VCR... which might be over-reacting since the timing is off? (And, therefore, the timing correction combined with the AGC in the TBC ends up countering the over-reaction of the VCR?)

The AVT-8710 has a side effect of raising blacks and lowering the contrast, even when the source is a DVD player. It isn't related to VCR timing.

This may be considered a good thing if your tapes play "hot" to begin with.

http://forum.videohelp.com/threads/3...screenshots%29
http://www.digitalfaq.com/forum/vide...n-avt8710.html

Capture cards and VCRs both use AGCs.

Thanks for that capture info.

BTW, I still use my original AIW 7500. I always adjust Levels in VDub before capturing. Adjustments are always required from tape, a little or a lot depending on the VCR and the tape. You should get into the habit of doing that, because your game image shows blown-out highlights, luma as well as chroma. By adjusting during capture, you can compensate for whatever anyone's AGC is doing, although I don't see that effect with my 7500. If the original camera has AGC pumping effects, the best you can do is find an adjustment range that will accept the worst case scenario of level changes. A minute or two of working with a tape that has obvious AGC effects is usually enough to find a suitable range.

Capture is usually YUY2 color, as you've been doing. The luma range for this type of capture should fall inside the range y=16-235. One factor here with the images is that they're converted from YUV to RGB which stretches dark and bright extremes to 0-255. So if highlights are outside the capture range of 16-235, then 0-255 is as far as you can go -- everything beyond that will be clipped. True, some tbc's try to limit that range but your "tbc" image still has brights and color extending beyond RGB 255. Still images often are sufficient, but in some cases you should submit a short clip of unaltered YUY2 capture, which you can do in VirtualDub with lossles AVI and "direct stream copy" mode.

VirtualDub filters work in RGB. Huffyuv can save VirtualDub output as either RGB or YUY2. The colorspace for DVD/BluRay is YV12 (which huffyuv can't use), so a DVD encoder will downsample RGB or YUY2 to YV12. Some software makes that conversion cleanly, others don't do it so well. I do that in Avisynth, or else I save VDub output to YV12 with Lagarith or UT Video codec.

The most frequently used frame sizes for DVD are 720x480 or 704x480 NTSC (720x576/704x576 PAL). If you capture at 640x480, that frame size is invalid for DVD and the encoder will resize the video. That's another area where some software doesn't do so well, especially with interlaced media.

I don't use all-in-one NLE's or conversion apps, instead I use separate apps for encoding, authoring, burning. An all-in-one app that seems to work well is AVS2DVD, but it's been years since I used it. AVS2DVD uses Avisynth for input (its own version is furnished with the package and doesn't affect any version of Avisynth if you already installed one), uses two or three different encoders of your choice including HCenc, uss another built-in for creating DVD folders and menus, and uses its own version of ImgBurn to burn discs.

A brief explanation of how to view and use the VDub capture histogram and how to use it is in this post: http://www.digitalfaq.com/forum/vide...html#post43646. Scroll down in that post to find the image of the capture histogram and nore info about it. Whether you use the tbc or not, invalid "y" levels are often impossible to repair after capture. If values are clipped (destroyed) during capture your video won't have the dynamic range or "snap" that it could have. Captures are tweaked for color and other elements after capture anyway.

Streaming: I use an Oppo player as well, which works with either MPEG or h.264 encodes. But most of my DVD's are on disc.

Okay, now I think I understand. Since the non-TBC capture is not correcting the timing, the AGC on the capture card may negatively affect the picture more than it would have if the TBC was present?


Thanks! This is good to know. As LS points out in there, though, each AVT unit has the potential to be a little different (not to mention how it interacts with my specific VCR). I guess I may end up finding my own TBC adjustments that end up being the sweet spot on most (I hope!) tapes.

Hopefully, I'll get a good technique down for making compensating adjustments after I get a better feel on what to do after viewing the histogram (or rather several histograms) for a given tape.

sanlyn, you absolutely rock! Thank you for spending so much time with the detailed responses. I am digesting and reading up on everything you pointed me to. In the mean time, I am attaching some short clips (done with "direct stream copy" mode and then "Save as AVI".)

Just to clarify, I don't anticipate making many DVDs. I will make them if family want copies of some of the videos and don't want to stream off their own devices... but, as far as I'm concerned, almost all playback will be off my Oppo using whatever encoding is suited best for streaming to a HD TV.

Thanks, everyone!

Thanks for those avi samples. They give more detailed information to work with.

Three of the samples have out of bounds luma levels. The images below have Avisynth simple YUV histograms attached. Thye luma portion of the histogram is the white badn at the top of the graph. The luma band resembles VDub's capture histogram. The unshaded portion of the graph on the left and right sides indicates data that exceeds the 16-235 range. The central portion is the safe zone. The black borders of the original frames are cropped off to avoid affecting the histogram.

frame 61 from the no-tbc birthday avi shows highlights in the unsafe zone at the right of the YUV graph. This effect doesn't last long, because the camera's AGC severely darkens the image near the end when the bright gift card appears in the frame. Sudden AGC luminance changes are known as "pumping".
http://www.digitalfaq.com/forum/atta...1&d=1463005828

Frame 61 from the tbc-birthday version. It's a little more greenish but at least your AVT's usual behavior kept brights in the safe range -- a nice little accident, if nothing else.
http://www.digitalfaq.com/forum/atta...1&d=1463006012

Frame 47 from the no-tbc game avi. Darks and brights both overrun into the unsafe zone. The video is oversharpened and looks somewhat "etched". It looks dim despite the y-levels overrun, and has a heavy purplish color cast. Red is oversaturated, skin tones are weird...and so forth.
http://www.digitalfaq.com/forum/atta...1&d=1463006211

Frame 47 from the tbc version of the game avi. Y-levels are little better but still out of range. Red is somewhat more subdued, but otherwise the video has the same problems its no-tbc mate.
http://www.digitalfaq.com/forum/atta...1&d=1463006411

The idea with color work and other tweaks later is not to make two videos such as these look identical, but to make each look "correct" according to your preference, with convincing color and valid luma and chroma values.

I made some corrections using the tbc-reset version of the birthday clip, attached as tbc_reset_bday.mp4. Most of the work was in VirtualDub in this case. Not easy color problems to correct, as somehow the autowhite in the camera or else the local lighting made a mess of the histogram and skin tones look weird, with glassy highlights in facial features. Took some patient fiddling to make things look natural. At least the mp4 demonstrates the possibilities with post-capture tweaks.

I made corrections on both of the ballgame avi's, more or less to show that in some ways they look identical, in others they don't, but they still play better than the captures. Both had to take a ride with Avisynth and VirtualDub to look better, but they're still too sharp. Attached as no_tbc_game.mp4 and tbc_game.mp4. Things are easier with better levels control during capture. As for precise color correction, don't even think about it with analog tape during capture or you'll be a basket case in no time.

Thank you for taking the time to do this! Color certainly seems to look more real. I will look at these in more detail when I get home.

I agree that I don't want to become a basket case trying to do precise color correction.

The real question for me is, what (if anything) I should do during capture? It sounds like your overall suggestion is to leave the AVT's settings at default, and then use VirtualDub's level controls to have the capture card adjust as it captures... and to base those controls off of what I'm seeing in the histograms for the video (as a whole).

And, based on what I'm seeing in other posts, I may find myself doing some similar adjustments over and over, due to the AVT-8710's nature.

To be quite frank, I was hoping to get to where I could make some adjustments to the levels for each tape... capture it... and feel that it is good enough for now. Maybe someday down the line, when I feel more comfortable with various restoration techniques, I could go back and work on the captures. But, for now, my family could at least enjoy the videos... and know that I have as good of a capture as I am going to get from these tapes. (Again, having adjusted the levels when capturing.)

But, I was surprised at how the AVT seemed to be making things look a little more grey/greenish, and wasn't sure if:
1. that's just how the camera originally recorded them
2. the AVT was making things worse
3. the AVT was making things better (and that it was the absence of the TBC was causing artificial effects due to timing/AGC issues as LS mentioned back at the top of the thread)

It does look like the AVT is helping to keep luma in the safe range, but is it making the color better or worse? (I'm still not sure.) Either way, it sounds like keeping the AVT in play is the way to go, but that I am going to need to make some levels corrections while capturing every tape for sure.

You should be making adjustments for at least nominally valid levels with every tape during capture whether you use the AVT, another tbc, or no tbc. Most tapes will need similar adjustments, many will need something more, or something less. If it were routinely possible to make perfectly color balanced and noiseless, glitch-free captures from analog tape, there would be no need for restoration forums.

Color differences with or without the AVT aren't major, IMO. In one sample it made changes in one direction (greenish, but the original was already greenish anyway), in the other it helped a bit with hot reds. I see people talk often about the AVT's AGC action. I haven't seen it. Someone will probably throw a link at us showing all kinds of test patch results, but it doesn't change the fact that a tbc is necessary and each tbc will have one problem or another. Short of outright malfunction, everyone who uses an external tbc, including the big boys at Disney, has to make some kind allowance for the device.

Was away from this project for a bit (we had a baby!), but I finally am able to get back to this.

Your information has truly been invaluable. Thank you!

We last left off that I now have a good working setup, and am trying to get the best captures I can so that I do not have to deal with the tapes anymore. I accept that I will be spending a lot more time in the future trying to do color correction (and other editing), but for now I just want to adjust levels in VirtualDub before capture so that I can have as high quality of a capture as possible.

You have given great instruction on this, and thank you for the recent VirtualDub settings posts as well!

Just one last request for advice before I finally start capturing from my tapes. Is adjusting the brightness and contrast as simple as:

1. using brightness to control the black levels so that they're not in the red zone
2. using contrast to control the brights so that they're not in the red zone

And, should I be doing the minimal adjustment to maintain having the levels in the safe zone as much as possible for the tape (maybe by jumping around and testing some various scenes to find a good average)?

Finally, should I expect that sometimes I won't be 100% in the safe zone... but try and get adjustments that result in almost everything being in the safe zone except any weird outliers that occasionally happen?

I'm just wondering if I have a graph where 99% is in the safe zone on average... but there is a tiny outlier that is far into the red zone, should I bother adjusting the brightness/contrast affecting the other 99%? (Or is this subjective?)

Thanks, again!

Congrats on the kiddo. :congrats:

Software-based proc amps on capture aren't ideal. The better choice is hardware-based, even ones inside of TBCs (AVT-8710, TBC-3000, etc). The reason? The software ones rarely work correctly. And it adds overhead to the CPU, and can cause dropped frames. You can try, but it may not work as desired.

Yeah, this is great: http://www.digitalfaq.com/forum/vide...-settings.html
I'll be making a guide for VirtualDub, and asked a few members for their input on it.
sanlyn did a great job there.

I always forget if "brightness" or "contrast" is the IRE (aka black level) adjust. They use dummy terms, and mask the actual video jargon.

Yes, minimal adjustments on capture to keep everything legal values.

Homeshot footage is rarely 100% (or even 75%), but the same can apply to other sources. TV isn't perfect, nor were the tapes and VCRs used to record.

Even pro Hollywood movies aren't 99%, and that's often due to artistic license. 90% is a good target. I'm referring to everything from gamma to IRE to white balance (color cast/tint issue). That last one is a major issue for home-shot video. Of all the formats, Hi8 always looks best, even more than DV (digital tape, but still consumer and tape).

Don't have low expectations, but don't have too-high ones either! :wink2:

FYI: "safe zone" is generally jargon to refer to overscan, and safe content placement so as not to show borders. You're using that term wrong, so I mention it so as not to confuse others. You really mean "values" (or "legal values"), referring to IRE/contrast/etc here.

Thanks, lordsmurf! =P

Whoops. Got it!


This makes sense. So, I will preview in VirtualDub (cropping out the edge noise during preview only), and make adjustments to brightness/contrast directly with the AVT-8710 based on what I see in the histogram.

Yeah, I can't echo enough how valuable that is for newbies like me. I had actually gathered a lot of the same settings from elsewhere in the forums (and from the How to Capture AVI Video in VirtualDub guide), but to have all of those detailed settings in one place is amazing. Thank you, sanlyn, and to everyone else who will contribute!

Interesting. So to make sure I understand... you're suggesting that keeping about 90% in legal values is a good target? Or did I misunderstand? I was going to try and get a lot closer to 99% where possible... but not to the point that I unnecessarily affect most values that would already be legal just to get a tiny outlier in range. For example... in sanlyn's "Frame 47 from the tbc version" up above, the brights are still fairly out of range, and I'd expect to make adjustments there. Probably on the darks, too, a little. But, it looks like well over 90% of the values are already legal there to begin with.

Also, this brings to question whether I should be doing adjustments to tint and color on the AVT-8710, too.

Is there a good method for measuring how much (and in what ways) to adjust the tint/color? (I was planning on using VirtualDub's capture histogram to see how I should adjust the brightness/contrast for capture. Is there something similar to use?) Or would tint/color be okay to adjust post-capture at a later date?


Hah. I'm definitely trying to find the right balance here of getting the best quality out of the captures without losing too much sanity. =P

Thanks!

The closer you can get to 100%, sure, go for it. But don't chase the impossible. All things considered, 90% is often a fair target for homemade VHS sources. Simple histogram values for IRE/brightness/contrast probably can be about 95%+.

The thing to remember is that everything is connected. Adjusting one value offsets the other. So it can be impossible to tweak each individual value to 100% perfection (or 99%). Like I always say, restoration is about making something better (hopefully much better!), not perfect.

Proc amps can rarely fix white balance issues. That's pure software work. In the analog days, it had to be shot right, period. There was no way to fix mistakes like that.

Got it! So stick to adjusting brightness and contrast using the AVT-8710, but leave the tint and color settings at default... and then start tackling other adjustments like white balance issues post-capture? Assuming I have that right, I'll give this all a go tomorrow.

Thanks, and have a good night!

Assuming that your TV or computer monitor is calibrated, let your eyes be the judge.

And don't forget most all modern TVs have the ability to quickly alter many of these values on playback, if they not to your liking. I do it all the time on footage, and I refer to commercial sources (DVD, Blu-ray, streaming, etc).

Histograms are tools, not bibles. Histograms can be wrong (visually), even if not scientifically.

Thanks! So, I was about to start capturing, but noticed that I am sometimes seeing what I can only describe as a scrolling pattern. Attached is a capture from VirtualDub to demonstrate. (No tape is playing... this is just the "stopped" screen from the VCR.) In this particular capture, the scrolling is downward, and moving relatively quickly.

Sometimes the scrolling is slow... sometimes it is fast. Sometimes the scrolling moves upward... and sometimes it moves downward. Sometimes the scrolling stops all together, which is what I would expect.

I've tried to make sure that there are no electronics next to the VCR or the TBC. I am using Mono-price 3ft S-Video cables, to try and minimize issues. I've also tried to make sure that any power cables for the devices are separated from the video cables.

Is this a sign of some sort of external interference?

It's strange, because there doesn't seem to be any consistency. And, like I said, sometimes the scrolling does seem to stop... which makes me think I need to eliminate it before capturing video.

-- merged --

And sure enough, right after positing that last message, the scrolling stopped. Here's a new capture that shows none of that scrolling pattern.

I'm wondering if it only happens after starting up the equipment. I have noticed the scrolling for a little while after turning everything on... and then it seems to disappear a few minutes after. (Though, I don't know yet if it eventually comes back.)

-- merged --

Besides the scrolling issue (which, btw, I haven't seen much of since posting), I have been trying to figure out the best way to adjust input audio levels for the capture.

It looks like I can control the recording audio level via the Santa Cruz control panel. It is currently at about 60% (or whatever the default is). When I turn on the Volume Meter in VirtualDub, I see the sound levels bouncing around on most tapes getting to about -20db. Since it doesn't seem to bounce all the way to the right (0db), I don't appear to be getting any clipping.

However, I do have some tapes with low volume (often from a PA system which didn't get picked up very well by the camcorder's microphone). I think that I'd like the volume to be louder (even though I'm likely to pick up more camera/background noise), but I'm not sure when to make those adjustments.

Do most people adjust the recording audio level per tape before capturing by adjusting the input audio level?

Or, do you tend to just always leave it in at some default 50-75% (which is what most of the guides seemed to recommend), and then adjust post-capture afterwards (maybe using something like Soundforge to also remove any camera hiss)?

If you do it before capturing, how do you typically find a good level to record at? (Just by whatever sounds good, or do you do rely on some other tools to determine how to adjust the level?)

Thanks!

You have a power interference pattern. It could be anywhere. So let's start here:
- Is everything on the battery of a UPS?

No UPS here. Is that the usual recommendation to get a clean, consistent power source?

The fact that it looks like power interference might also make sense with the fact that I mainly saw it on the startup of all of my equipment, when there would be the largest power draws.

-- merged --

Just ordered a Cyberpower cp1500pfclcd, which I had been meaning to get to replace a bad APC for my main computer anyhow. I haven't seen that interference in the last day, but it will still be good to get all of the equipment on a UPS with AVR.

Still have the questions above on audio levels if anyone is able to give their thoughts.

Thanks as always!!

A UPS will fix power-related problems at least half the time. Not all the time. Let's see what happens.

If not the wall power, then perhaps its being generated elsewhere -- VCR, computer, etc. I've dealt with my share of power issues in the past 15 years. It's never fun to figure out, but at least you have somebody that can help guide you.

Excellent UPS, I have that exact one for my Skylake build.

I actually use a separate non-PFC 1500 for my video hardware. I try to keep computer power isolated from video power.

It is suggested to leave audio ingest volume at default 50%, and fix it in post-capture as needed. Only in situations where it's really low would you want to increase capture volume. In most cases, correcting in Sound Forge will be equally good, and actually better (because lots of audio clips, at some point, even if briefly).

I was thinking of doing the same, but mainly because AVR is only on the battery backed up outlets... and 5 is not enough for me for both the computer and video equipment together.

Out of curiosity, why do you like to keep the computer equipment on an isolated UPS from the video equipment?

-- merged --

Okay, I now have only the video equipment (VCR and AVT-8710) running on the new UPS. (Not sure why you recommend keeping the computer equipment isolated from the video equipment, but I am going to follow your recommendation and try it like this for now until I get a second UPS for the computer.)

I am still getting very touchy behavior for that waviness/power interference pattern. I wasn't seeing it since my last post, until I moved cables around and reconnected things when setting up the UPS.

However, in the past I have been able to get rid of it by unplugging and replugging in cables.

Moving cables around (separating video from power cables) does not seem to have much of an effect. (Though I can introduce additional interference if I bring other electrical equipment near the video equipment wires. I just am not always able to easily eliminate all interference.)

In fact, up until a few minutes ago, I couldn't get rid of the power interference pattern at all... even after making sure all S-Video connections along the path were secure. (I even disconnected the audio cables, and removed the AVT-8710. I connected the VCR directly to the video card's purple adapter, and the waviness was still there. I tried different S-Video cables, and still no luck.)

Finally, I was able to get rid of the waviness again by simply unplugging the purple adapter from the video card and re-plugging it in. (Thankfully, adding the AVT-8710 and the audio cables back in continued to retain the clean behavior.)

Attached is an AVI of what I'm seeing now. Does this look properly clean to you? There's a very, very slight flicker, but I would guess that's because it's an analog signal?

And, if you think this looks good, hopefully I can maintain it looking like this.

Thanks!

-- merged --

Besides the questions about power interference, I have two more for post-capture and the editing/encoding workflow.

First of all, the captured video definitely looks better than anything I've been able to capture in the past... thanks to all of your help!

Now, I'm on to the editing/encoding.

#1. As a reminder, I plan to stream my edited videos off of an Oppo BDP-103 that supports MP4.

It seems to stream interlaced H.264 just fine. Since this is the case, wouldn't it be preferred to leave the video interlaced and let the Oppo/HDTV handle it, rather than using QTGMC to de-interlace before doing any editing/encoding?

#2. I'd like to use Adobe Premiere Elements for editing. (Trimming some of the frames, merging some captures together, adding subtitles, etc.) Since I'm using PE for that, I'm assuming I'll also use it to encode my MP4 H.264 files (unless you have a better suggestion).

In the past sanlyn has warned about PE having colorspace issues. I am trying to avoid them as much as possible.

Currently, PE does read in my Lagarith YUY2 AVI files just fine.

I am currently creating a project that is based off of their DV Standard NTSC preset (720x480, 29.97 frames/second, 48kHz audio), except that I edit it to set to Upper Field First (which is what my captured AVIs seem to be).

After editing, to encode I am using Publish and Share -> Computer -> AVCHD, and am using a custom preset based off of the MP4 - NTSC DV Widescreen preset, except that again I set the Field Order to Upper, check off Render at Maximum Depth... and I set the H.264 Profile to High, since the Oppo supports that.

However, I end up with a MP4 that has a color space of 4:2:0 instead of 4:2:2. I don't know if this is expected, or if it matters.

My 3.2 GB AVI file is also reduced to 195 KB, which seems pretty small.

Basically, I'm not sure if I'm killing quality here during the encoding.

#3. I noticed that MediaInfo says my AVI had a 3:2 display aspect ratio, but the MP4 is 4:3. GSpot says both are 3:2. I just found this strange that they were reporting different things, and I think that I am misunderstanding something here.

Thank you!

I've been under the weather lately but manage to cruise by every few days or so, and glad to see you're making progress. Nice work.

OPPOs ae great players, I've used them for years as well as a pricey DENON and can't make up my mind about which I like better. I can see differences in their playback image on a calibrated TV, but you'll never get two dissimialr models to output exactly the same thing. Still, both work better in every respect than typical players. I give OPPO the edge when it comes to motion handling and avoidance of moire or stair stepping with some videos.

The basic rule with interlaced video is that it should remain interlaced. True, there are some analog problems that absolutely require deinterlace or field separation for really tough cleanup needs, but the results should be properly reinterlaced. Denterlacing, like resizing, always has a cost. There's no getting away from it. Deinterlacing requires that interlaced half-height frames be resized to full frames. Those interpolations involve numerical rounding errors and other shortcomings, proper correction for field offsets, chroma offset differences, and other sticky factors. Reinterlacing involves the same shortcomings in reverse. Thus the breakdown and restore is done only when the original problems are worse than decent deinterlace/reinterlace. I know of no NLE of any quality that performs these operations as well as Avisynth and QTGMC, which have gone through years of engineering tweaks to ensure as few errors as possible.

I've seen a ton of pro-vs-con arguments and demos that debate deinterlace issues, as well as telecine issues (most but not all of my captures are telecined video). So far it's my experienbce that originally interlaced video at 60 fields per second plays more smoothly than deinterlaced video at 60 frames per second most of the time. I have to go with one video pro who noted simply that 30i plays that way because the original was created that way -- once the original relationship between interlaced scanlines is altered, something gets lost. I've also seen plenty of motion playback studies showing that a telecined video that has 2 interlaced (pulldown) frames for every 5 at 30i will render motion more smoothly than de-telecined video at 23.976 fps, if for no other reason than the display device itself is running at 60Hz, not 23.976 Hz, so flickery motion at 23.976 is inevitable on a 60Hz display. My final encode of film source video is encoded with telecine flags restored. As for PC monitor display, it's yet another can of worms with various PC media players giving different results.

Your mileage might differ, so I'm in no way debating those issues by adding to tens of thousands of previous posts. I can say without doubt that my OPPO and DENON players and my DENON upscaling/HDMI receiver can handle interlace and telecine better than lesser players, which vary widely in their ability to manage any kind of video. More "typical" players can't handle interlace or rescaling worth a damn, which anyone with two good eyes can confirm.

I'll add here that I don't encode final output from tape caps as mp4 containers. For PC demos in forums it's an efficient way of doing it, but my final encodes are to standard formats: MPEG/DVD or standard definition "regulation" BluRay. My reasoning is that any player from anywhere is designed to handle those formats properly because those formats have strict encoding requirements that generic mp4, mkv, etc., don't enforce. My encodes are MPG or BDMV/M2TS containers encoded to DVD or BluRay spec that I can also use for authoring to disc.

Well, actually, the colorspace cautions aren't mine but from tests by other users. I've tried all the popular NLE's and finally settled on lossless RGB with AfterEffects, converted via Avisyth, but used only when I need an NLE for things like subtitles and whatnot, which I seldom use, or for adavanced color correction, layer work, or Ken Burns type slide shows. I then use Avisyth for conversion back to YV12 for encoding. I think you can probably find better encoders than PE, but it's likely better than some overly picky users claim it is and is certainly adequate at decent bitrates.

Standard consumer DV field order is bottom field first, standard field order for everything else is TFF. So if your originals are TFF, then obviously you should not reverse its field order to BFF. In any case, always stick with the source field order if it's correct to begin with.

Standard MPEG and h.264/BluRay/AVCHD encodings are YV12 (4:2:0), just like the stuff you buy at the store and see on your TV. A more complex colorspace is used for a lot of processing, for reasons beyond this discussion. So your final encodes are normal.

Can't say. More playing time means larger files, higher bitrate means better quality and larger files. Lossy encoding is, well...it's lossy. Maybe that gives you an idea of what lossy means. The encoded version will always be smaller than unencoded, full-frame video. The encodes contain only one or a very few full-sized reference frames for each GOP (Group of Pictures). The other frames contain only the data that changes since the last reference frame. This is called interframe video encoding, where each picture in a GOP is based on other pictures and GOPs that come before and after it. Your Lossless AVIs are made up of complete, self-contained images.

Typically my 1-hour TV show captures come out to 26 to 30GB or so as lossless AVI, but encoded to 720x480 BluRay at 10MBPS Variable Bit Rate come to about 5GB.

Lossless AVi has no display aspect ratio (DAR) flags for playback. Its frame aspect ratio and its playback aspect ratio are the same as its physical ratio, which is 3:2. Your encoded video at 720x480 has a frame size aspect ratio of 3:2 and a DAR flag set for either 4:3 or 16:9. A PAL lossless AVI at 720x576 would show in MediaInfo as a "5:4" aspect ratio because that unencoded AVI has no DAR flag.

Thanks, Sanlyn! I hope you feel better.

I am definitely making progress. The final products already look so much better now that I understand more what's going on (all thanks to you guys!), and am using better equipment and techniques.

I appreciate all the responses. One thing in particular stood out:

I think this is where I have room for improvement. I left PE at the default settings of: 1-Pass VBR, Target Bitrate 3Mbps, Maximum Bitrate 6Mbps.

I am still thinking of leaving the final encoding as H.264 mp4 for the time being (since my player can play it, and it gives me an equal testing ground for now.... though, I may take your suggestion to make the final encodings more "portable" for friends and family).

But, it seems like it'd be a good idea to do 2-Pass VBR... and certainly to up the Bitrate. Is 10Mbps the suggested rate? (And would that be both for Target and Maximum?)

I tried that, and though I'm not seeing a noticeable difference in quality in this particular clip, the file size is 3x. (610KB vs 195KB, from the original 3.2GB file.)

A 3000 to 4000 1-pass encode sounds like the usual default for many encoders, but that has always sounded like a default designed for web posting. The standard def BluRay example I gave is for MPEG2 10MBPS target/15Mbps max 2-pass encoding for near-archival SD quality. One could go higher (I believe 15Mbps target is often quoted as "broadcast quality" MPEG). There are other settings in use, including the use of standard-spec GOP (30 frames for fast action sequences most of the time, 60-frame GOPs for talky shows lots of times, but frankly I think smaller GOPs look better every time something moves on my 52" plasma. On the 32-inch LCD the difference is very difficult to see). Audio is Dolby AC3 48KHz at 324kb.

For DVD -- yes, I still use it, as the majority of people I know don't own BluRay players and don't intend to get one -- my limits are about 6500 target and 9000 max MPEG2. The encoders are the old TMPGenc Plus 2.5, TMPGenc Mastering Works, and the HCenc encoder.

I'm also using x264 encoders for h.264 BluRay and AVCHD, with 8MBPS target / 12Mbps max for SD video. These make for pretty good archives, but interlace/telecine and what I call "real" color and detail subtlety don't hold up as well with h.264 and tends to look over filtered. These are BluRay containers for hard drives or disc authoring.

Anything of importance that I have on external drives is backed up on other drives or optical discs. These are personal preferences, compromises between the hungry commercial bitrates by pros on some sites and the bitrate misers on many consumer sites who claim they see no difference between h.264 at 2000kbps and h.264 or MPEG at more generous bitrates. My preferences are used by a great many people who encode for quality over file size, and historically they likely go back to my upbringing in an environment where film was always the primary medium. Frankly my eyes don't agree with flowery claims about low bitrate encoding. But that's just me, I guess. By and large I'm only going to encode once and I can't save every lossless capture, so I go for the optimal encode. I'm one of those stuffy old dunderheads who still thinks film is better than digital and analog audio is more accurate than CD (but how many would get a $2500 turntable for their LP's, and how long ago was it that they saw a fine celluloid print projected onto the giant silver screen?).

Personal preference, and personal history as well, have a lot to do with this archiving business. The going theory among teachers and pros is that the current generation was brought up on low quality mass media, mis-tuned CRT's and factory-preset LCD's, and few know what first class work looks like. There might be something to that, I don't know. I do have a wife who can't stand watching most of the careless video work she sees today, so I guess I'm not alone. I was raised in movie houses and had many relatives with graphic arts and pictorial design backgrounds, so I guess that makes a difference.

Moral: what looks right to you is what you should use. The more you work with visual media hands-on, the more you'll learn and the closer you'll look. It's a natural progression toward getting what you want out of the memories you have on tape.

Thanks, Sanlyn! Super-informative, as always!

I would like to try the MPEG2 encoding (via Premiere Elements, since I'm editing in it)... to see how it compares to H.264 MP4, but I'm having trouble trying to find an equivalent to your settings.

#1. For DVD, PE has a NTSC DVD Standard preset for use with the MPEG2-DVD format. I took that and modified it.

Multiplexing is set to DVD. For video, it looks like the Codec is MainConcept MPEG Video. There is a quality slider, which I put at the max of 5 (though interestingly, that doesn't seem to affect file-size).

I select to Render at Maximum Depth, and I went with VBR, 2 pass.

Interestingly, this has Minimum, Target and Maximum bitrates. 9 is the maximum for all three (though I have seen postings that say not to go beyond 8).

For GOP Settings, there is a M Frames (2 through 8) and N Frames (9, 12, 15, 18). Defaults are 3 and 15, which is where I left them.

For audio, there is Dolby Digital Stereo, MPEG and PCM. I went with Dolby Digital, but I'm pretty sure that it's not AC3. Seems to be a limitation of PE. Also, I assume you meant 320kbps and not 324.

#2. For MT2S there's not a whole lot of options. Multiplexing is set to TS. It only does the MainConcept H.264 codec, and is fixed at Profile: Main for 720x480. Can also set a Target and Maximum bitrate, with the maximum being 12. I ended up going with a Target of 10, and a Maximum of 12. That's about all I can set for it.

Audio is once again Dolby Digital Stereo, 320kbps.

#3. I'm not so sure I get all the encoding options in PE that I really need to control (as in your example settings).

Maybe I should export as another format from PE, and then use some other encoding software (like the ones you mentioned). But, I wonder what format to export from PE. Maybe Lagarith AVI again, and then pull it into a proper encoder? (Once again, I'm wary of PE trampling colorspaces and the like during conversions.)

#4. I am still digesting the results, but it seems like the H.264 MP4 has an ever so slight jerkiness about it when there is movement, compared to the MPEG2 and MT2S encodings. It's very slight, and I'll have to sleep on it to see if I'm imagining it or not.

Thank you!!

-- merged --

Followup on #4. (Sorry, can't edit the post anymore.)

I notice that slight jerkiness on the H.264 MP4 only on the more recent encodings that use 2-pass VBR with the 10Mbps target/15 Mbps maximum bitrates.

I don't see the jerkiness on the H.264 MP4 that is 1-pass VBR and a lower bitrate.

The jerkiness is super slight, but I do keep noticing it on the higher quality H.264 MP4.

However, the image quality might still be the best on the 2-pass VBR with the higher bitrate (possibly even a little better than the MPEG2 and MT2S encodings I did... though it's hard for me to get a good frame comparison.)

You know, these posts get a little long when you have to keep taking a break every couple of minutes. Maybe it's my medifcation.

Jerky playback with high-bitrate encodes can be a problem if certain elements are a bit more than a specific PC can handle. The media player can be a problem, too, so when you mention playback problems note whether you mean PC playback or external playback. On any PC I have a love-hate relationship with VLC player, which looks great when it works but goes schizo now and then and forces me to use Media Player Classic, which seems to be one of lordsmurf's favorites.

The best way to get info about what might be happening is to either post a small piece of the encode, or post a sample of the original AVI and let one of us brave readers make an encode to give you some ideas. It could be that your typical NLE won't let you get away with a few things using those presets -- which is one reason why I don't use NLE's for everything. They're just too confining.

PE should be able to save its output as lossless AVI for input into another encoder if you want to use it that way. I do it with AfterEffects and even with TMPGEnc Mastering Works.

Sorry, sanlyn. I appreciate the help, but please don't feel the need to reply until you're feeling better (which I hope you do soon)!

I was comparing playback on the end-device... the Oppo BDP-103D.

I went ahead and narrowed everything down to a small clip. There is camera motion in the clip, and I still keep thinking there's something a little jerkier (even if it's slight) in the mp4 encodings when played on the Oppo.

First, attached are the following small, uncompressed clips:

1. original-lagarith-avi - direct stream copy via VirtualDub of the original capture
2. exported-lagarith-avi - lagarith export from PE just to see if this is viable to then do the final encoding in other software, if needed. (The only thing done to this clip is that the head noise on the bottom had a "Transform Clip" effect applied to "clip" the noise and replace it with a black background)

And then the following four encodings, done in PE:

1. mp4_vbr2pass-10mbps-15mps - H.264 MP4, VBR 2-pass, Target 10Mbps, Maximum 15Mbps, Profile: High
2. mp4_vbr2pass-3mbps-6mbps - H.264 MP4, VBR 1-pass, Target 3Mbps, Maximum 6Mbps, Profile: High (included only because I originally thought it didn't have the jerkiness of the better quality mp4 clip... but now I think I was wrong)
3. mpeg - MPEG2, VBR 2-pass, Quality 5, Minimum 8Mbps, Target 8Mbps, Maximum 8Mbps, GOP: 3 M Frames, 15 N Frames
(was an attempt to match your suggestion for a high quality DVD)
4. m2t - M2T, VBR 2-pass, Target 10Mbps, Maximum 12Mbps
(was an attempt to match your suggestion for a near-archival quality standard-def blu ray, but I don't think I could get PE to match your settings)
(had to zip this one, because the site wouldn't let me upload a .m2t file)

Audio is Dolby 48KHz at 320kbps... but, I believe PE only let me use AAC for the mp4 encodings.

I'm still not 100% sure I'm seeing additional jerkiness in the mp4's... but I think I am (on the Oppo).

I'm curious as to what you think of the encodings in general, and the uncompressed export from PE.

And, back to the original need... I would think that in an ideal world, my final encoding for my needs would either be a H.264 MP4 or a high quality standard-def DVD mpeg2 or a blu-ray encoding like M2T, so that it could play on more devices. But, obviously my primary concern is to get a high quality final encoding that I can stream to my Oppo.

Thank you!!

Thanks for the new samples. Dandy stuff, mostly good points but still a few things to learn from, so I'll take a closer look and post later.

I loaded them into a hard drive on my OPPO and DENON, no shakes or jerks. The camera shake doesn't help, though. One of the mp4's came out to constant bitrate (or so it looks in some software reports) instead of VBR, but that usually won't cause choppy playback. You had, I think, a "minimum" bitrate of 8000 on one of the encodes -- that was the MPG, I believe. While there's no official rule on min. bitrate, for DVD it's usually been quoted as 2000 kbps. Setting a high minimum would waste a lot of data bits and could give some decoders a fit.

I think I see why you wanted to deinterlace. Many consumer cameras just didn't interlace well, so there's aliasing and line twitter now and then. Will give some suggestions on how to handle it.

More later. Good work.

Thanks, Sanlyn! I was actually watching the full clips today, and noticed that the MP4's are a lot worse than I thought in other portions of the video. I am going to make new samples of some other portions of the clips which show the issues a lot more. I definitely appreciate you looking at these, but might be worth holding off anything further until I post the new samples.

And, thanks again, for the kind words. This has been an adventure, to say the least. =P Thanks to you, lordsmurf and the rest of the forum members for helping me learn my way through this!

-- merged --

Okay, I was able to get some new samples made. These are from another tape of a fashion show.

Both of the H.264 MP4s are particularly bad on my Oppo. As the guys walk by, they almost seem to jump forward on both of the MP4s.

However, I don't see that behavior when watching on my computer (though I'm watching interlaced frames there, and not seeing the full video... so maybe that's why I don't see the issue on the computer).

I am now certain the MP4 encodings I am getting out of PE definitely have this jerkiness/jumpiness in motion when played on the Oppo. All of the clips that I have encoded this way exhibit the problems whenever there is motion.

I do not see any issues with MPEG-2 or M2T. They seem to be a lot smoother overall.

Here are the new sample files:

First, attached are the following small, uncompressed clips:

1. fashion-original-lagarith-avi - direct stream copy via VirtualDub of the original capture
2. fashion-exported-lagarith-avi - lagarith export from PE just to see if this is viable to then do the final encoding in other software, if needed.

And then the following four encodings, done in PE:

1. fashion-mp4_vbr2pass-10mbps-15mps - H.264 MP4, VBR 2-pass, Target 10Mbps, Maximum 15Mbps, Profile: High
2. fashion-mp4_vbr2pass-3mbps-6mbps - H.264 MP4, VBR 1-pass, Target 3Mbps, Maximum 6Mbps, Profile: High (included only because I originally thought it didn't have the jerkiness of the better quality mp4 clip... but it still does)
3. fashion-mpeg2 - MPEG2, VBR 2-pass, Quality 5, Minimum 5Mbps, Target 8Mbps, Maximum 9Mbps, GOP: 3 M Frames, 15 N Frames
(was an attempt to match your suggestion for a high quality DVD. I changed the Minimum and Maximum from my last sample, but couldn't quite bring myself to lower the minimum below 5Mbps, since the whole video is actually pretty small.)
4. fashion-m2t - M2T, VBR 2-pass, Target 10Mbps, Maximum 12Mbps
(was an attempt to match your suggestion for a near-archival quality standard-def blu ray, but I don't think I could get PE to match your settings)
(had to zip this one, because the site wouldn't let me upload a .m2t file)

Audio is Dolby 48KHz at 320kbps... but, I believe PE only let me use AAC for the mp4 encodings.

Actually, the only reason I mentioned deinterlacing was because of lordsmurf's original suggestion that I do my final encodings for streaming as deinterlaced H.264 mp4s. But, I wasn't sure the deinterlacing was necessary, since the Oppo plays interlaced H.264 mp4s just fine.

Thank you yet once again!

More details as to what I've tried and what I can visibly see as the least amount of interference were in the earlier post #22 above.

But, just wanted to add that I am still seeing that waviness pattern on occasion, and it's unbelievably frustrating trying to get rid of it. I do seem to be able to get rid of it by moving cables around, plugging and unplugging things, and restarting equipment... but, there doesn't seem to be much rhyme or reason. I think I've just been getting lucky somehow.

I keep thinking something I did was the cure, and the interference seems to disappear for a while. But, often, when I come back the next day (or even hours later), I see some level of interference again.

Eventually I can get rid of it by trying to isolate video cables from each other. I even have a Trifield EMF meter, and have tried to use it to keep equipment away from anything that is generating an electric/magnetic field. But, this whole thing just seems like chance at the moment.

Yep, it can be mighty frustrating. The earlier "waviness" post shows what are called hum bars and hum interference -- not always from
some electromechanical radiation but often from poor ground connections (maybe). That could be in any cable or input socket. I had that in an s-video plug some years back and it took weeks to finally isolate it.

Your latest samples from post #33 are quite good, and some nice levels control (for more tweaking if you like, although most people overdo the tweaks). I have some notes for those caps later.

Back to your earlier samples from post #30: Black levels are a bit high. They keep the dark background from clipping, though, and the camera's autogain is obviously a headache. BTW, there are some Avisynth filters that can help work with those fluctuating levels and luma "pumping" but a longer sample with that problem would be needed to help you with that. Sometimes the filters work, sometimes not so well. Depends on the video.

I did some chroma noise correction on the attached samples. There's chroma bleed and shift that discolors the central figure's shirt edges, his arm, and on the face of the smaller figure. I did these in Avisynth with the original YUY2, so thanks for including the original. The A_00_30i_Plus25_DVD.mpg attachment is a plain vanilla MPEG encoded to DVD spec. I should have done more color correction in RGB, but never got around to it.

Sloppy interlacing is a bane of consumer cameras. It can be fixed but at considerable cost. For these samples I deinterlaced with QTGMC using "very fast" presets and the vInverse plugin that do almost zero denoising but helps clean up interlace twitter and combing a bit. There's already too much denoising in the captures, leading to wax-face contouring that you can see mostly in facial contours. But I do understand the sloppy interlace problem, and it's a pain in the eyeballs. Hopefully the interlaced DVD sample eases some of it.

For the other samples I did a little more color correction but wanted to demonstrate that sometimes deinterlacing done correctly is the major fixup for what budget cameras do. Many of my projects required progressive output. The downside is that DVD/BluRay can't be used with 60fps. Dropping fields to get back to 29.97fps is a serious problem with motion, especially with a camera playing acrobat every few frames (my sister uses kamikaze camera moves you wouldn't believe), frame blending is nothing more than permanent damage. Fast motion, and especially jerky cameras, requires more bitrate -- most of it wasted trying to render smooth motion, and denoising is almost impossible.

The BluRay_encodes_m2t.zip attachment is 2 BluRay encodes, one MPEG and one h264, to give you an idea of what TVMW5's MPEG and x264 encoder would do with the originals. Both exhibit posterizing clay face effects, mostly the fault of the original. There are ways around it in Avisynth, but they're a big headache and I didn't have time to go there.

The A-00_60p_Normal_and_DeShaker.zip attachment has three 60p h264 encoded mp4's. Likely they look better because I did more work in Avisynth and VirtualDub, and even added more noise in the form of ordered film grain to help tame contouring and rough gradients. There's an MP4 with the normal frame dimensions, an MP4 that used DeShaker in VirtualDub to help stop the jiggles, and an MP4 with a Normal-vs-DeShaker comparison (left image = Normal, right image = DeShaker).

With all the samples I cleaned up borders and centered the image in Avisynth with this code:
This left the image untouched, cleaned up head switching noise, protected the borders from color correction, and centered the image vertically without touching the core picture. DeShaker, however, moves the frame around so that it loses 6 pixels on each side -- one has to decide whether steadying motion is worth the trouble. I've seen worse, including such bad camera movement that 32 pixels was lost on side borders. With borders keep in mind that even today's HDTV's use overscan unless you disable it.

Just some more ideas and different software. The DVD was encoded with TMPGEnc Plus 2.5 and needs more color work, the others were done with TVMW5 and a little contrast boost with TMPGenc's filters......

I'm having a hard time right now. Offline matters, forum posts, site improvements. It's taking a toll. I'm really glad the sanlyn is able to help you with this. I'm reading and posting as I can.

Power is all about "same circuit". Something noisy on a circuit can interfere with others on the circuit. Something really noisy can jump circuits. This is why grounding is so important. But not everything can be 100% grounded, especially when you're talking internal computer components. So that's why computers are rarely on the same UPS (circuit) as my video gear.

You may also have cabling issues. Cables go bad, and I throw several away yearly. Cheap or expensive, does not matter.

The purple ATI box can also go bad. That doesn't happen often, but I've seen it before.

If that Oppo will stream interlaced MP4, and it doesn't look bad, you can do so. Leaving it interlaced is the best solution, but I always worry about future-proofing since most devices hate interlaced H.264 content. They do understand it, thanks to our progressive Youtube world. Remember there's no rule to do things just one way. You can archive an MPEG, make an interlaced H.264 copy for your Oppo, and if needed, go back to that MPEG to make deinterlaced MP4 in the future.

Adobe Premiere uses the MainConcept SDK to encode, and it looks fine. You don't get the advanced options of the full MainConcept, but you probably won't need those. Just keep it simple.

Output 4:2:0 is fine. You won't be re-encoding it from that.
4:2:0 doesn't have loss like DV 4:1:1.
DVD is 4:2:0.

Gspot is old and sometimes get confused. It's best with AVI and MPEG only. MediaInfo is for newer formats. As noted, lossless AVI doesn't have DAR, as it's a raw video.

General rule for MPEG-2 encoding:
352x480/576 = 3500-5000 VBR
720x480/576 = 5000-9000 VBR
720x480/576 for archive/BD = 15000 (15k) CBR

^ H.264 encoding is about 25-33% of those numbers. In general. H.264 has more options, and is even more content-based.

Jerkiness can often be your computer. H.264 needs more CPU. The real test for motion is non-computer, using specialized MP4 player for the TV. For example, BD player or WDTV. Based off your descriptions, this is exactly what you're observing.

When I'm sick or just not feeling well, I get bored staring at the ceiling, or watching too much TV. I need mental exercise. So I know how sanlyn feels. This is our interest, and keeps the ol' noggin refreshed. It's a good escape from what ails us. Remember, the web is also an entertainment venue, even when the venue is for productive means!

Many consumer cameras interlaced oddly. Not badly, just oddly. It depends on the format, and the age (ie, early VHS from 70s, etc).

That's all for now.

Holy cow, sanlyn. That was way more than I expected!!

I am going to go through your work in detail, and try and learn from what you did. Thank you so much for that!!

Thanks for your kind words on the captures. They are definitely so much better than where I started... again, all thanks to you and other members of this forum!

As an aside, do you think the encodings from PE are pretty bad? In particular, I really thought I was going to use its H.264 MP4 encoding, but the jerkiness is just so noticeable on my Oppo.

For now I've been doing M2T from PE, which seems to result in nice playback... albeit it at an increased file size. (Though that also probably has to do with the 10Mbps target/12Mbps max I am using.)

From the sounds of it, I think I might be better off doing my timeline editing in PE and then exporting a new Lagarirh AVI (assuming that it looks ok to you and that PE didn't trample something when exporting Lagarith) for the final encoding in some other software (like you mentioned).

For the interference, I just ordered some s-video cables from Blue Jean Cables to see if they work out better than my mono price ones. I'm starting to think the interference is either from the VCR or the PC though. I might try some ferrite clamp-on cores to see if they help at all.

Thank you!!!

-- merged --

I'm sorry to hear this, LS. Thank you for reply! I definitely appreciate it, but also didn't mean to push you when you weren't feeling well. I hope you feel better soon!

Got it! Thanks.

Yeah, at this point I'm thinking it's either the PC, the VCR, or the cabling/purple ATI box. I just ordered replacement S-VIDEO cables, so we'll see if those help. I'm not sure there's much I can do if it's the PC or VCR.... and, I hope it's not the purple ATI box, but I may see if I can pick up another one off ebay just to see.

It's just weird that with some fidgeting I can make it go away. However, time also seems to be a big component. Just as often as that humming pattern comes after some time... it also goes away. I also see it every time I restart the computer. So, I can't hep but wonder if it's the PC and something it's doing at different times that causes the interference.

I'll just keep trying different things, replacing different things, and trying to narrow it down.

Did the "good" sample up above look like what you'd expect, though? Or do you still see interference issues with it? It's hard for me to achieve that "good" state, but I do get it on occasion... and when I do, that's when I've been capturing.

It looks good! Except for the jerkiness.... argh. I will try using something other than PE to encode, and see if the jerkiness disappears.

Thanks for that! Unfortunately, PE's export limits the max bit-rate for MPEG-2 to 9Mbps. (Well, at least it does for VBR. I'll have to check CBR). Otherwise, I might do that 15Mbps CBR and call it a day, since the MP4 is exhibiting jerkiness.

Or, continue to use MTS.

I definitely get jerkiness at the beginning of clips on the computer. On the Oppo, I get it only during motion. But, I only get it with the H.264 MP4. Not with the high bit-rate MTS encodings.

Are you saying my questions are like medicine? =P Well, I won't feel so bad for asking for so much help from you guys then! Seriously, though, I do hope both of you feel better, and VERY MUCH appreciate you both taking the time to help me.

This has become a great mental exercise for me, too, and I have enjoyed learning the right way to do things... and am seeing great results! I don't have a ton of footage to convert, but I do want to do it right, and I have enjoyed learning how from you guys.

(Except for the interference issues. Those, I am not enjoying. =P)

Thanks, guys!

PE is a consumer version, so it may max out at DVD, not higher specs.

You'll end out wasting time and money to fix power issues. You may grab a few unneeded cables (for now), etc, but eventually it will be found. We've all been here -- sanlyn, myself, many others at this forum.

MTS isn't a format. MTS = MPEG Transport Stream (m2ts, mts, ts, etc). In fact, H.264 can be put in a transport stream as well, so it's not just MPEG anymore. So saying MTS is sort of like saying AVI or MP4 -- ie, saying nothing. What format and secs of actual video are in that container.

Again, most players hate interlaced H.264, and I have a feeling your Oppo is no different.

And some devices hate some containers. MP4 and MKV are the two most popular ones. MP4 is falling out of favor for MKV. Although it has more overhead, and is not web streaming (yet), it has more abilities.

Everything I've ever played in my OPPO's over the years has been interlaced or telecined, except for 23.976 or progressive video I've made myself. so I can't agree with that. Most new OPPO players for the past few years have a "24fps" switch for progressive film speed and streaming sites that transmit film speed progressive video. Some OPPOs have that switch on by default. It's somewhere in the settings menu, I'll see if I can look it up.

My Denon has that feature as well, but last I looked the DENON was on "Auto" for film speeds.

Like any expernal media player, any of them will have problems with oddball encoding structures, which I try to stick closer to standard formats. 1-second or 2-second GOP's, usually closed-GOP output, etc.. Even the priciest of players is not as accomplished as a PC media player is when it comes to accepting an unlimited range of formats, and none of them have anything like the codec storage on a PC.

Still, you shouldn't get jerky playback on motion. Does it do the same thing on commercial discs? I also have an HD PVR that records high-def h264 off cable TV, which play with no problems except for occasional slow audio sync because of the PVR's encoding of cable4 transmission noise (and they're all encoded as interlaced). But a little once-over with free software fixes audio in 20 minutes.

Look in OPPOs Video setup menu. "1080p/24fps" should be OFF, deinterlace should be "Auto", 4K/2K should be either Auto or OFF, 3D should be OFF. Check other settings as well. Sometimes their damn online firmware updates change the playback settings. Hopefully you have "NTSC" TV System selected instead of "Auto" or "PAL". I know some folks who aren't even aware that OPPO has dozens of settings inside.

Sorry, I should have said the M2T file was in fact H.264. So, it seems like the Oppo may be okay with interlaced H.264 in the transport stream... but not in the MP4 container.

So, I still need to find the right final encoding for watching on my Oppo (which is my primary need). If I continue encoding with PE, then it's probably either H.264 in the M2T (also uses the MainConcept H.264 codec), or possibly the MPEG-2 which maxes at 9Mbps bitrate....

Or, I just have to give up encoding in PE, because those seem to be the only viable options. =P

If as you say you see power problems as soon as your PC turns on without running other apps, that points to a power supply -- which could also mean your PC's power supply. But juggling external cables shouldn't turn it on and off: a PC supply either works or doesn't. Not that intermittent trouble isn't possible, but its not common.

I didn't see any of this noise in your latest captures, which looked pretty nice. I haven't had time to respond, as we're in hurricane mode at my New York home now and power keeps blowing out. Par for the course, they buy it from ConEd which has to be the biggest ripoff since NoFat Twinkies. I should just go back to Tennessee all the time but the wife works up here a lot.

As for MP4 containers, my MP4 encodes aren't always high bitrate but I don't think I've ever used or recorded h264 below 4000 kbps. I don't think PE's encoder is bad news throughout, plenty of folks seem happy with it, although I think it can look noisy at times. h264 and MPEG alike are both contrary: they act differently with different material -- so I'm glad we still have a choice between them.

[OOPS! Power just flickered on and off, shutting down my relays elsewhere. The PC's are on battery but I had to stop and reset all the clocks in here).

[OOPS. There it goes again. I give up, I'm not resetting clocks again until I think it's all over. A little difficult for me to get around for the next few days anyway).

When Mama Nature settles down I'll give more notes on your later captures. Sorry for the delay. Doing good work so far.

-- merged --

Oh, I forgot...better do this before Hermine or whatever her name is acts up again:

I mounted every video in the last several posts on a Canvio USB drive and plugged it into my OPPO and then the DENON. None of my posted encodes has stutter on my OPPO or Denon. Even the 1280x480 compare MP4 plays OK. However, I was sorry to see that all of your MP4's stutter. Hmm. I must have gone through them too quickly the first time.

I hereby conclude that all of my samples have black levels that are too high on a calibrated TV. Will try to fix those tomorrow. I must be getting careless.

I actually have mixed feelings on this. On one hand, it's unfortunate that PE's MP4 encodes have the stutter. But, I'm also relieved you're seeing the same behavior on your devices.

I may stick with H.264 M2T for now then.

I have tried to wait until the noise goes away before capturing.

Sometimes I wonder if it was coincidence that fidgeting with the cables worked... and that in reality it was the PC doing something different (like maybe being less resource intensive after calming down after boot up) that really helped.

I'd be up for trying a different power supply, but I'm not exactly sure what to look for in an alternate power supply that might lend itself towards less interference.

New s-video cables arrive today. I'll also try and pickup some ferrite cores to place on the s-video cables (and maybe the purple ATI box?)

Good luck on riding out the bad weather. Stay safe!

Pointing a finger at the PC power supply is really a guess. More likely, if you're jiggling cables and it clears the noise it points to connections or the cable itself. Personally I've replaced a few as I said, and there are now BlueJeans cables and British QED's all over the place. Also replaced a few monitor cables over the years.

Here's an idea of some of the color and noise cleanup in the sample video original-lagarith-avi.avi, from post #28. The image below is reduced-size frame 32 from the original:

http://www.digitalfaq.com/forum/atta...1&d=1473353374

Some color bleed and shift throws visible discolorations into skin an other areas. Not pretty look at. Sometimes this can be cleared completely, sometimes just enough to be unnoticeable. The image below is a full-size crop from the original lower right corner, with arrows pointing to unsightly uglies in several areas: the greenish overflow on the top-left finger, further down some bad chroma noise on the arm's wrist. There's orange bleed in the neckband and beside the boy's cheek onto the shirt, and further down are green uglies on the boy's folded forearm. The right-hand figure has some brownish bleed into the edge of the shirt sleeve from the dark sofa, some discoloration along the side of the face, and some green bleed on the smaller boy's arm.

http://www.digitalfaq.com/forum/atta...1&d=1473353521
f32 original avi crop

Here are two examples of the chroma cleanup. The image below is the same frame deinterlaced (it's is frame 64 in the deinterlaced version) with chroma cleanup:
http://www.digitalfaq.com/forum/atta...1&d=1473353669

Below, reinterlaced frame 32 after cleanup (obviously the reinterlaced version has edge elements from both fields in the same pic):
http://www.digitalfaq.com/forum/atta...1&d=1473353709

The "after" pics also show some softening of false contouring in faces, caused by aggressive dnr in the original and which looked worse after sharpening during cleanup. The filters used were:

vInverse was used to help tame interlace combing a little. The chroma cleaners are ChronmaShift, FixChromaBleedingMod(),and SmoothUV(), plus a little help later from Virtualdub's CamcorderColorDenoise. awarpsharp2 further tightened chroma bleed along edges. GradFun2DBmod was used to soften gradient edges with help from AddGrainC to add some natural fine film-type grain to reduce an over-filtered look. The sharpener was LimitedSharpenFaster applied after smoothing with GradFun2DBmod.

The deinterlaced version was re-interlaced using this standard procedure as last step:

You can do some chroma cleanup using SeparateFields alone instead of deinterkacing, but in some cases it doesn't work as well. Cleaning up edges is ineffective on interlaced frames. You still have some halo effects from the original, but cleaning them would look worse than leaving them as-is and they aren't that bad anyway. It's unlikely that all of the chroma problems can be fixed entirely without creating a muddy mess.

-- merged --

Some notes on the Fashion Walk captures in post #30:

The original and the PE export version look identical to me, no changes. There would likely be no changes unless some RGB processing in PE had been applied. 4.2.2 and 4.4.4 color changes in Adobe's NLE's aren't the problem, 4.2.0 and 4.1.1 (YV12) to/from other colorspaces that's not very correct.

The smoothest motion with less disturbed interlace effects is the MPEG version. The lower-bitrate mp4 shows chroma noise and rough edges and mottling in the large smooth areas of solid color, such as the background wall panels and overhead panel in the last part: these are compression artifacts. Smooth areas of solid or gradually changing hue require higher bitrates, especially during motion. On motion the lower bitrate MP4s tend to look rough and grainy. I'd say that maybe at least 5000 or so would do a cleaner job than a 3000kbps target in h264. 5000 would be something like the equivalent of 6500 or 7000 in MPEG. To say that h264 can use 1/2 to 1/3 less bitrate than MPEG is a myth: it all depends on the content.

These are pretty nice captures, notably because you have better lighting with that camera and there appears to be little or no AGP activity/pumping. Or maybe you used a better camera??? Red is a bit oversaturated and tends to "glow" and smear somewhat. You'll note that consumer cameras have many more problems with the earlier captures in available room light (also popularly called "available darkness"!).

To clean up chroma problems I used pretty much the same code used in the earlier captures.

The picture below is cropped from a frame of the original fashion-original-lagarith-avi.avi (left image). Arrows point to red blur and smear toward the left, and general blur that softens the lettring in the sign. The image at right is the cleanup, looks cleaner and sharper.
http://www.digitalfaq.com/forum/atta...1&d=1473354434

The picture below is cropped from another frame of the original fashion-original-lagarith-avi.avi (left image). Arrows point to red blur and smear and a slight darker edge halo on the left side of the shirt. Arrows also indicate dark gray bleed into the red area at right. The image at right is the cleanup.
http://www.digitalfaq.com/forum/atta...1&d=1473354495

The idea is to get better clarity, cleaner color, and try to avoid that "typical VHS crummy noise" look.

Yeah, I'm hoping it's just the cables. I will try the new BlueJeans cables tonight. I also ordered a new, sealed replacement purple ATI box, just in case it's that, too. I know it's not the AVT-8710, because when the noise is present... it's present both with and without the AVT-8710 in the chain.


I both like and don't like you for this! =P I am just getting started on learning cleanup techniques, and I didn't even notice some of these bleeds. But, now I do, and I'm sure they will be very noticeable to me going forward. I think you just cursed me. Hah!

I'm pretty picky on this stuff, though, so I would have noticed eventually. I've just been so focused on getting the captures right up until this point.

This is great stuff! I am going to get very familiar with all of this (and your suggestions in past posts).

I do wonder, though.... as scenes and lighting change on a tape, do you find yourself applying certain scripts to certain scenes, and other scripts to others? Obviously if it's the same camera, I'd assume a lot of the same fixes would be constant. But, I can imagine that you might find yourself having to apply different fixes as the content and lighting change as well?


That's great news! So, I now know I can export a Lagarith AVI from PE, and then encode it into a final format with a better encoder without losing anything due to PE.

Thanks for looking at the the other exports from PE. If I continue to use PE to encode, we know the MP4 is a no-go. The MPEG-2 seems decent, but I'd probably use a different encoder to get a higher bit-rate.

I'm curious what you think about the H.264 M2T from PE, though. PE allows a high bit-rate, and it seems to look good on my Oppo. If it passes your scrutiny, I might just stay with that and continue to use PE to encode... just for the sake of streamlining things.

The first set of captures were from my parent's camera. The fashion show capture was from my wife's parent's camera. I guess we know who had the better device, now. =P

WOW!!! This is incredible. I cannot believe the improvement here. Were there any additional steps used (or subtracted or modified) on this vs. the script you posted for the first capture?

I am blown away by how much cleaner that looks!

It does seem like the cleaned up version might have a little more noise/pixelation in some areas... almost as if it was sharpened slightly. (For example, the top part of the red shirt that is just to the right of the shadow from the arm.) It's small, but is that a side-effect of one of the filters in particular? Either way, the improvement is substantial!

Thank you, sanlyn!!