As a bit of a continuation of my previous thread but I managed to get my hands on the Nintendo 64 "Change The System" Promo tape from 1996 to do a proper comparison between it and the vhs decode copy uploaded on archive.org back in July of 2023 (Was decoded by me a week ago)

Now, the tape was only ever produced in Extended Play mode sadly, so I'm wondering if a Panasonic 1980 would fair a better chance at getting a higher quality extraction compared to my JVC MV-45U. The JVC was set to EDIT mode too

That or I could acquire a different copy of the same tape. Had to do that for some blooper tapes were the HiFi audio track produced a lot of crackles, bought a different copy of the same tape with much cleaner audio.

Here are some image comparisons, in my eyes decode is a bit better in the sense that you can make out text more easily, but the ringing problems are at full force. The only thing I did was resize the images to a 4:3 resolution.

But, the more and more I'm looking at this, the more I'm thinking "how much of this is an actual boost in quality rather than just all the extra noise disguising itself as an overall sharper image"

-- merged --

And here are some video comparisons

Kind of looks like the difference between having gone through a noise reduction circuit versus not. Kind of akin to the degree of an edit mode leaving things "unprocessed" which will vary from brand to brand and model to model.

Can't lie, decode looks quite a bit better in this comparison specifically.

Ironically, I see there's a copy of that tape that is less than 20 miles from me on ebay, but I'd probably still pay to have it shipped if I were to do my own comparison between higher end machines. It's unfortunate that it is EP as that excludes the SVO-5800 and JVC BR-S series that'll only do SP.

The other issue I think is that the same machine didn't play both of those tapes, so it's not the greatest comparison. I am a little surprised that the decode copy still has quite visible line dropouts as that's one of the things it is supposed to be very good at covering up.

I do remember skimming through my decode of it and yeah, had lots of dropouts.

Honestly going to buy a different copy and play it in my JVC-SR-VS30U instead to see if that would help at all. Maybe, maybe not. From the small comparisons I've done between the two VCRs the image quality is basically the same, the SR-MV45U tends to have better linear audio quality.

Who knows, maybe your Panasonic would prove better results than my JVC. But yeah we have no idea what deck was used for the decode capture back in 2023.

You might find this slightly interesting, it's from the guy on ebay that sells rather expensive refurbed JVCs, but the comparison aspect is still interesting.

https://jvccompare.com

My favorite comparison would have to be the 7600 vs 9600. The 7600 looks much more detailed all else being equal. Oddly, he's capturing into DVCPro50, but that's not a variable here. My guess is capturing to ProResHQ would have looked better for the same bitrate as ProRes has more color depth and is visually lossless.

The black levels and saturation levels also vary quite a bit which makes it hard to tell if decode is better or just not crushing the blacks as much on certain captures. VS30U(Domesday) vs 9911 is also an interesting one where there's still technically more detail in the domesday, it's just that the chroma and luma levels make the 9911 look more appealing overall in most scenes unless the blacks are being crushed.

The other issue is that there aren't comparisons of say two or three different units of the same model to show what can be expected of machine-to-machine variation.

For example, could be the 7600 has less worn heads than the 9600 and that is why it appears sharper. Hard to say.

Well I do like how the site is setup, but I thought you weren't a fan of his comparison video?

Anyway, made more comparisons between my VCRs. Different Nintendo Promo tape, this time being Donkey Kong Country Exposed from 1994. It was released in EP mode as well.

First set (Gameplay) has the VS30 picture mode set to Edit with the Pinnacle 510's sharpness set to 2, The MV45 picture mode is also set to Edit, Pinnacle sharpness is at the default value of 0.

Second set (Developer) has the VS30 picture mode set to Norm, MV45 picture mode is set to Edit. Pinnacle's sharpness is left at the default of 0 for both. Though keep it mind the live action footage was filmed in choppy bit crushed slow mo.

Third set (Diddy Kong model) has both VCR's picture mode set to Edit, Pinnacle sharpness is left at 0

You also lack the best AIW. PM me when you're done getting your captures.

It's awful. I can even see ringing where none usually exists. :huh1:

In the "decode2" image, you can actually see the outline of his face in the sky. Even low-end VCRs didn't make that mistake. In a tiny preview window, sure, you may think it's not a big deal. But view it large (as people normally do), and the ghosting is obnoxious.

That's it entirely. That's been the argument for 30 years now. Noise is noise, not detail. Lots of people are easily fooled by optical tricks, and they refuse to acknowledge it. Terms like "most aesthetically pleasing" are misused here as well, as a backup excuse when they're confronted.

I don't find noise pleasing anymore than I find a dirty tile floor pleasing. I never hear somebody say "the dirty floor looks good", unless they're trying to avoid mopping.

This is why it's not interesting at all. He's using non-suggested gear, and likely getting non-quality as a result. So the comparisons are invalid.

Too many variables, as you've correctly pointed out.

No, don't make excuses for them. Or rather, don't give in to their common backpeddling. :no2:

We were told "software will make it better later", but that was obvious BS, as I've correctly stated for years. Now the goalpost has been moved to "that was done with the old way, you have to use (this month's) new way, which is now final (it's not)".

That's almost an oxymoron statement. It looks "better", but has obvious problems? :huh1:

I'm just saying that the file was uploaded back in 2023, I myself decoded that file using the latest version of the software, keep all the settings at their defaults. But I guess the file was uploaded as a bit of placeholder, so we have no clue what deck was used to play back the tape

So it's an old file I decoded just recently, but I bet if they were saying "oh that's invalid now, go recapture it using these tools" then doesn't that go against their claim of "capture it once, just decode it over and over again the future it'll be fine"

I really don't care now to spend like around 2 hours downloading the file from archive.org which always has slow as balls upload and download speeds, and spend up to 2 more hours decoding only half of the video because I'm not using some ultra $2000 gaming PC but a $1000 mobile workstation laptop.

I'm ok with saying one looking better, but ultimately it is down to preference in what you prioritize.

Here's two quick examples:

This was probably the best capture I'd seen of Sequal's "I'm Over You" on YouTube:
https://youtu.be/ovxBT2X_W-o?si=LN6ifr_jk53aKzv3


But then I uploaded this U-Matic capture:
https://www.youtube.com/watch?v=ihw1aVfucQo


As to which is "better", depends what you look for. If you can't stand ringing or "grain type noise" the U-Matic capture probably isn't going to be your preference. However, if you prefer more detail and fewer dropouts, you'd probably say the U-Matic is better. Both could be further improved with avisynth wizardry.

Comparison is a little apples to oranges (both are NTSC at least), but the idea is if you played the same tape in two different VHS machines and got the two above results, you'd probably prefer the output of one machine over the other for that tape even though both have flaws.

Neither will be what the master looked like (as there's generational losses for both), butyou can only do as good as what is recorded on the specific tape you have. Decode should theoretically look "at least as good" as the next best option, but if it's basically indistinguishable from a traditional capture, the extra time/storage/complexity still wouldn't be worth it for me personally.

That's an outstanding video to show the weakness of Youtube (over)compression. :congrats:

Nope, just yours. That other video has too many flaws to get much help.

And that's the irony. It's lots more work, but for an inferior output. At least for the expense (fast "gamer"-quality desktop, lots of storage), and all the time involved, you'd think you'd at least get something better. But nope, that's never happened, and probably never will. At least for VHS.

Other formats may do better, such as Betamax. Maybe.

So I guess ringing isn't the norm for VHS, but the norm for U-Matic? I think if there was ringing from a VHS tape it'd usually be a TV recording that was recorded from a channel with poor reception, hm.

Guess on the topic of noise and grain, this reminds me, I captured footage from an NFL game on FOX back in 2002 that was recorded in SP Mode. You can easily tell that compared to most tapes, the footage here is extra noisy and extra grainy, that's just because of the recording itself.

So hey, a traditional capture was able to preserve this extra noise. Keep in mind I had the VCR picture mode set to Norm and a Panasonic ES15 in the chain, that was a leftover when a previous section of the tape was a copy of a camcorder tape, so it had flagging.

I actually did order a copy of that same promo tape and will play around with it just for the heck of it with a few different VCRs. I'll try it include a D-VHS machine, WVHS machine, AG1980, a few JVCs, and possibly decode if I can get it to play nice with my computer hardware. It could be that EP has more to be gained from decode than SP, haven't really seen specific comparisons on multiple tape speeds, but I'm sure they are some out there. I'll plan to post back with some of my captures if I can do any better than yours and the posted decode.

What is your logic there? :unsure:

W-VHS will be most interesting to me.

Use mine. We'll PM.

Honestly I'm most curious about the Panasonic, I've heard they are better for EP tapes, but I wonder how much of that is quality wise and not tracking wise.

And I think his logic is that I guess EP tapes have a much lower signal to noise ratio (SNR) so apparently decode could be able to extract more 'data' from those kind of tapes.

That’s a really good idea. If you use one of LS’s it takes out a lot of variables.

I’ve seen some examples where the rigging is extremely obvious. Crazy to me how dramatic that whole conversation gets.

After spending two hours downloading the ldf file and decoding around two minutes of it again, here are two raw screencaps to show how bad the ringing is

Ignoring most of the top rated comments dunking on the guy and saying that he's a stupid old cranky man. I feel like Oln's comment is the most pleasant to read. He seems rather grounded in reality.

Another interesting-ish comment would Iknowityoudont's where he made his own comparison between his thrift store quality setup and a decode capture he grabbed from archive.org back in 2024. The main takeaway being that the ringing is still present, it's most noticeable on the edge of the banana.

https://slow.pics/s/9kfrBqLt

I don't have the images anymore, but there is also that Wacky Adventures of Ronald McDonald decode capture Titan did back in 2022, the video files were oversharpened to all hell and had serious ringing problems. I myself decoded the same raw .flac file a couple of days ago, in which while it was no longer oversharpened, the ringing was still present.

There is also that Gatorade ad that was posted to the archive.org page in 2024 where the comparison was between decode and some composite only VCR hooked up to a Pioneer DVR-530H and a Black Magic Analog to SDI Converter.

When I went into the drama filled discord server there was a comparison that was posted like around month ago that had some ringing, yes it wasn't in your face bright as can be, but it was there, and that is when the excuse was "something something rf filtering" Man at least Latreche34 had some sense and said something was wrong with those comparisons, believing the conventional capture was too soft and the decode capture was too noisy.

So much for "capture today, decode tomorrow" huh? These aren't "random bad examples" no it's a pretty common problem...

oln = hodgey = Oyvind (his actual name) = the guy who actually forked/"invented" vhs-decode.

Notice that the Github has "oyvind" in it.

He's always been a pleasant chap. But my understanding is that life has intervened, and he's not active much anymore. That's what has created the current situation, where "the inmates are running the asylum".

He gave the keys to the house and the house is now a mess :laugh:

But yeah, I know he is the actual developer behind the project, it's the way he writes sounds more, professional to me and overall down to earth.

From what I've heard the decode quality output for video8/hi8 is around the same as a high end workflow. So for somebody like myself, I have no need nor desire to make the jump. It'd be a complete waste of my time.

U-Matic seems to be Aramkolt's thing so he'd be the judge on if Decode could be a benefit to that format specifically.

It's like, the only format we can all agree on was the one where the decode project was born from, Laserdisc, but that's mainly because Laserdisc and VHS, while analog formats, are wildly different.

Isn't Laserdisc a composite only format, so something like s-video would not be a benefit. So yes, bypassing composite by doing an RF capture of a Laserdisc is the right idea and does provide the best quality capture. But with VHS and 8mm we already have a way to bypass the quality loss of composite, an S-VHS VCR and Hi8 Camcorder respectively.

I understand s-video isn't the only reason why we use those tape players, but still.

My copy of that tape already arrived, so I'll see if I have time to dig everything out to try a few different captures with at least 4 different machines that can do EP. I still like capturing into ProRes422HQ without the use of a computer (AJA KiPro via SDI), so it won't be quite the same, but should give you a good idea if the capture can look better than the decode one that was posted.

Welp, that testing was kind of short lived, I think the tape was either not that great of quality or was slightly sticky as it frequently was clogging heads or would slow the drum way down at random points. I might try baking it later, but don't really want to subject my VCRs to more of that than I need to.

However, I was able to grab several different captures with screenshots below. For EP, the 7600 seems best with R3 On. I'd say both the JVC HR-W5 (WVHS) and AG1980 do better though, I'm mainly looking at the clarity of the small text on the lower right and along the bottom. These screenshots are just in quicktime with the ProRes422HQ file still interlaced, so that's why you'll see some jaggies on all of them, though it doesn't matter too much since this is a no-motion scene.

For the AG1980, it is in "normal" mode with the sharpness slider set to the middle.

I'd say in this group of captures, the AG1980/HR-W5 look the best, followed by the 7600 in either mode with R3 on.

Now whether those are preferable over decode, that's up to the end user. I think the above are pretty good results to achieve in realtime for EP tapes without the use of a computer for the capture (Used SDI->AJA KiPro/ProRes422HQ).

Well thanks for trying, hoping the new copy I'll buy won't suffer the same fate as yours. Seems like they are all comparable quality wise.

I appreciate the help.

Actually, the quality is quite a bit better and they do vary noticeably, but I forgot that DigitalFAQ compresses uploaded images. Check out these in the uncompressed screenshot form.

Okay yeah, looking at all your examples the Panasonic 1980 looks to be the best imo, and with that I feel like Decode isn't all that better in comparison to it.

Feels like if you enabled Edit mode on the Panasonic all the extra noise would make it practically identically to the Decode capture.

Basically the 1980g sample is the best overall in terms of sharpness, so the decode sample just looks like it but with extra noise and ringing artifacts.

Starting to agree with Latreche/Dellsam with this comment he made on the videohelp thread when Brad posted his own comparison

"I see both samples have the same sharpness, The VHS-Decode (or tape-decode as they call it now since it captures different formats not just VHS) has an extra layer of noise which makes it look sharper but the actual video details are not."

No, that's not quite accurate.
- JPEGs are not recompressed, unless found to be defective, or over (about) 810 pixels wide.
- PNG is converted to JPEG. I don't like it, but few use PNG anyway.

That's the current limitations. It will be addressed on update.

Yuck. :mad4:

That device/workflow seems to degrade quality (blurring) more than normal. :hmm:

What I see in the decode sample is lots more noise, and boosted contrast. So it appears "sharper" when it really isn't. There's not much in the way it true added detail. (That's ignoring non-VCR device blurring.) And that ringing/halo and ghosting is truly awful. It may seem tolerable in a still image, but in motion the image is infested with distracting ghosting images.

Interesting, but none of it is surprising. :2cents:

I know nothing about possible video tape SSS and how they should be baked (for audio tapes there are a lot of technical bulletins). But you can try food dehydrator. I prefer that to remove SSS before digitization, however in Europe it is very rare, on some AGFA tapes only because BASF never used polyurethane binding.

Ah, those original uploads all were PNG's, and all also over 810 pixels wide since QuickTime scales to double the native resolution upon opening the file because 720x486 is only like 2 inches diagonally on modern retina displays. Either way, all of the captures were with the same hardware and display/screenshot method, so the variances should really just be the difference in how VCRs process the image.

Here's an unedited trimmed file of the AG1980 and the W5 captures for those that want to do a more detailed comparison when you adjust the chroma/luma/black levels in post compared to both decode and the original poster's traditional capture.

On the AG1980, you'll see some dropouts that it is masking whereas they really aren't as noticeable on the W5. Since the tape is crappy, I could also see there being variations on dropouts on different passes. There was a paper I read on U-Matic where dropouts were said to actually decrease with more passes, which is counterintuitive, but I will say that W5 capture was about 10 passes after the AG1980. My guess is it has it had more to do with an intermittent or short lived clog on the AG1980. If there are loose particles that can be swept away with passes, the "more passes, less dropouts" could be a real thing I suppose and perhaps the tape gets "less sticky" with more passes also, but I think that also presumes that the tape hasn't been physically damaged by being eaten like this one was.

When the W5 does actually have a line dropout, it's much more poorly masked, so it probably isn't preferable for tapes with lots of dropouts but could also be that those individual lines are more easy to correct in post for those that are avisynth masters. I really have no idea how all the heads are used on the W5 when playing EP, there are a total of 5 pairs of video heads (10 total) plus one flying erase. It could be that it takes multiple passes at each line and then compares to the other reads of the same tracks to get an average or the best signal, but that does seem kind of an advanced technique for a machine that was released in 1995.

To be fair, SP is likely to have far fewer dropouts in general since the tracks are wider, EP here is a worst case scenario. These will all do much better with SP tapes.

Been able to upload 720x480 .pngs before, seems like if it's above 810 pixels then that is when they get downscaled.

Anyway it's funny how they tend to mention the head switching noise is perfectly clean, it isn't, it's cleaner but not fully gone. Worst of all, for this example anyway, is the head switching noise covers up a bit more of the bottom frame. With your Panasonic picture I'd be able to crop off 6 pixels, whereas with the decode capture I'd need chop off seemingly 8 or 10 pixels.

How many pixels to chop is a bit of a tougher subject, my captures are in the

I'm not sure if I've seen a claim about no head switching noise with decode. If it were to work, I think it'd need to be a first generation tape or at least be a higher end commercially produced tape. I think I've seen a JVC BR-S series not have any head switching noise with a commercially produced tape and I think the W-5 can do it as well. SR-MV's are also pretty easy to adjust the head switching point manually, but it'd be a little different for each tape. Conceptually I think it is up to the VCR as far as the head switching point and Decode only taps in after the two head signals have been merged by the VCR which decode would have no control over. Also with my captures, the AJA captures the full 486 vertical lines, so 6 of the lines are kind of a "free" clip. With other capture cards, they choose a fixed 480 of the 486 available lines, which is why you'll see some capture cards showing more of the top image. In your posted decode capture, there are 488 lines, and I'm not sure where the extra two are coming from. Could be that it grabs the few lines right above the standard frame in order to preserve closed captioning maybe, but I also don't see any closed captioning data up top either, but perhaps this tape doesn't have closed captions. I think certain SDI digitizers (snell and Wilcox products come to mind) have a way of preserving closed captioning data without line 21 being part of the capture, but I haven't really played around with it.

Not exactly related, but this guy was pretty impressed that his DVD player could preserve closed captioning when dubbing in both directions (from DVD->VHS and VHS->DVD): https://www.youtube.com/watch?v=1pKjur_tnQE

Yes, ringing was bad on U-Matic, particularly if the footage on the tape was copied via composite. Using dub connectors as well as the later "SP" formulation of the format helped somewhat.

Hello again, after getting a Panasonic AG-1980 from Aramkolt decided to capture a bit of the tape using said VCR, here are the exact same frames as the previous JVC-MV45U and 2022 Decode (Decoded by me in 2025) samples.

Like with the JVC, all I did was crop 8 pixels from the sides and resize it to a 4:3 aspect ratio. Same TBC and Capture Card were used, same settings used for both.

The Panasonic was set to EDIT mode.

I happen to have both the DK Country and N64 promo tapes as shipped from Nintendo Power. Guess I should pull out my decks and see what I get. If I recall, the N64 tape is one of those cheapo deals where they used small T120 reels (as opposed to the big plastic reels T30s have) with very little tape on them. This tends to cause problems with some VCRs when threading tapes as the tension can be off and confuse the VCR into thinking there is a problem.

Yeah, afterwards they started having their tapes recorded in SP Mode thank god. Minus the Pokemon one, that sucks.

That probably explains why it, and the Invasion of Nintendo tape I had, have tracking problems on the Panasonic. Yet an EP mode recording of HBO from 1989, in which it was partly taped over with family home video footage, plays back basically fine.

This ain't about decode specifically but I've been running some of my own comparisons between my Panasonic and JVCs (used Edit mode on both, I am really unsure on if I should stick with edit mode on the 1980...) On SP mode tapes (VHS or S-VHS) there doesn't seem to be any quality jump, at least nothing major.

With EP mode tapes, the Panasonic beats out the JVC. I think that is the general consensus? Use 1980 for EP/VHS-C, JVC SR series for SP.

When it comes to Decode, at least in the context of EP mode. It doesn't seem any better than a refurbished AG 1980.