Conventional vs. vhs-decode comparison! (cartoon + live action)
 

Good day, ended up getting my JVC S-VHS MV45U VCR refurbished from Aramkolt recently with him also adding in modifications in place to get the deck to be able to be used with VHS-Decode.

When he still had my deck in his hands he made a capture of the VHS tape "It's the Girl in the Red Truck Charlie Brown" mainly for me to grab the close captioning data from.

So with his help, I can now make the most unbiased comparison between both methods, same tape and same VCR. Two images and two videos are provided, with the decode versions being created from a raw data snippet Aram sent me that I decoded myself.

The only thing I did was slightly crop them to 704x480 for the conventional capture and 720x480 for the Decode capture along with adjusting the colors of the conventional capture to match the decode capture as close as possible.

Right now I can not make my own comparisons as I do not have any hardware to actually start making my own decode captures.

I also have a Google Drive with a couple of more photos between Aram's finalized decode file and a previous capture I did of the tape on a VS30U

At least since he installed the suggested amplifier the deck's conventional output quality has not been reduced, in fact it is better than what it was before as it no longer over sharpen halos (and make halos baked into tapes even brighter) so therefore it is more compatible to the quality I get out of my VS30U.

Overall I am not impressed and the quality is worse, I'd only stick with using decode to extract close captioning data at this time. I know the JVCs tend to suck but I thought it would suck in the sense of they output a nosier image (which means their Signal to Noise ratio is lower than the good amount of around 40)

And I've done comparisons, in the context of SP mode tapes my JVCs and Panasonic AG-1980 are pretty much the same quality wise, there isn't a big difference outside of frame positioning and the 1980 leaving more of the right edge of the frame intact. LP/EP speeds and mono audio is when the Panasonic is just better than the JVCs but that is besides the point.

Site Staff edit:
- I've added the before/after slider for your post.
- If the above before/after slide doesn't work, then open both above attached images in new tabs, and then flip back and forth.
- And sample from "It's the Girl in the Red Truck, Charlie Brown"

convention version on left <> vhs-decode version on right
Admin refused to unhide hidden content
nothing will help to see hidden content

That vhs-decode sample has so much noise, and is so soft and blurry. And that cyan cast? :sick:

It's crazy how vhs-decode actually makes a quality VCR look worse. vhs-decode merely make a low-end VCR look better, it's not at all "archival" or the best conversion method. This has been shown over and over again, using all sorts of samples, from multiple people online. The method had promise, but this has been the problem since day 1, and it's been almost 10 years now. That's just never changing.

I've added the before/after script to your post. :)

We've known that VHS-decode has softness, ringing and hallo issues for a very long time, We've been pointing it out all along. The troubling fact is, it is impossible to know for sure if this is an issue with the RF pickup hardware or the decode software, There is no possible way to visualize the RF signal without decoding first, So it's a two step process with no possible way to single out each step.

If they want to perfect the process I'm afraid they have to reverse engineer an actual VCR processing boards (RF preamp, RF demod, TBC, Y/C) and convert the signals into a simulated algorithm, No one did it better than JVC engineers.

The ghosting on the right edge of Spike's head under the whiskers (and also on the right hand side of the pole to his right) curiously only really appears on the JVC capture. Is Decode blurring it out? Interested in seeing how it would look without filtering (--sd? --y_comb?) if it's not already set that way.

Applied a slight luma noise reduction of 3 to get rid of ringing that decode captures tend to introduce.

Now the images on the Google Drive were from Aramkolt's decode of the video data so they do not have any luma noise reduction applied.

Edit - Here is a frame of the decode capture without any noise reduction applied

It's both.

Also, as I've said all along, NTSC is vastly worse than PAL, and I don't know why. It's really not the format, but something in the processing. PAL still has issues, but the NTSC is just awful, sometimes outright unviewable crap, with the severity of noise.

Yep.

Gah. I didn't even see the DOGG (Snoop Dogg?) on the food box. It's so faint on the normal capture, but obvious heavy ghosted on the vhs-decode.

And remember -- these are small 720x previews. Put this on a normal large TV, and it's just awful. You could put a ruler on the screen, and it would measure by at least 1" of halo gapping everywhere. Double-vision.

Feels like any sharpness difference you can just do in Avisynth. I'd rather take a slightly less sharper image than one that has more artifacts that would need even more filtering to take care of.

Haloing and what not being baked into the tape is one thing..

Decode is not necessary to capture the closed captioning. Many DVD recorders and other MPEG2 capture devices will preserve the closed captioning and sometimes even XDS.

Hi have you test "Filter Tune" i did test it maybe 2 month ago i remember that it did have many filter knobs you can tweek

i have the Domesday Duplicator + the ADA4857 amp i did buy it about a year ago i still do not have any laserdisc player only done some VHS captures and Video8

i did do a custom Domesday Duplicator app thanks to Ai(https://claude.ai) i can not code so it's all Ai :)

it does downsample to flac on the fly and start the audio capture to at the same time

https://github.com/Mattiasn7788/Dome...ses/tag/v1.2.2

Hello, only Aramkolt can answer your question as he was the who made the decode capture. I do have not have any means to utilize the deck mod he made for me right now.

ok sorry did misunderstand did read to fast :)

Really do not want to make another thread, but this post still shows off a decode comparison, this time with a U-Matic tape.

Aramkolt did both captures, I am not into U-Matic as he is.

Here is what he had to say in terms of what he used

Admin refused to unhide hidden content
nothing will help to see hidden content

That Umatic "decode" looks like shit. It's covered in a veil of pattern noise --and-- typical halo/ringing.

Even an Easycap is better. :sick:

The Google Drive links aren't working. Can you fix them?

Thought Aramkolt still had them available, guess not. He'll have to provide new ones.

Here's the decode link for U-Matic (no audio), haven't done enough research on how to best synchronize audio just yet within FFMPEG alone. Guessing probably would be easiest to trim a traditional capture and the decode capture so that they start on the first frame of video, then paste the audio over:
https://drive.google.com/file/d/1vVa...ew?usp=sharing

Here's my "traditional" capture via SDI/ProRes422 (this does apply some slight compression), not sure if there's really much to be gained by going to ProRes422HQ for lower effective resolution SD formats, but maybe:
https://drive.google.com/file/d/1D1R...ew?usp=sharing

You'll have to download the clips to view them as google drive does not support FFV1 playback.

A better comparison would probably be captures all made on the same machine via a variety of ways. I can think of at least 5 ways to do it for U-Matic, but option 5 is only possible on specific machines. Some U-Matic machines also do not have DUB output, so that might not be an option in some cases either.

1. Regular composite output
2. External DUB timebase corrector or Transcoder to S-Video
3. Modified S-Video output (grabs luma and chroma before they are mixed to composite)
4. Internal TBC option of a BVU-9x0 series machine which bypasses usual processing, but is composite output only.
5. Decode
.

I've done comparisons on different clip of methods 1 through 4 in the past and I will say that they all produce a different "look" with differing degrees of noise/grain/ringing/haloing. Some may prefer the look of one over another. Decode likely does represent how the video actually is stored on the tape before it goes through various noise reduction stages within the machine. There are noise reduction settings that could be enabled during video export, so probably should play around some with that as well. Depends on the look you prefer and what equipment you're able to obtain for older equipment.

DUB transcoders and TBCs are notoriously hard to find these days (other than the Keystrobe Dub Optimizer which is also quite expensive and yet still grabs chroma from the composite signal).

The internal optional BVU-9x0 TBC is even harder to find (since not all of those machines have them) and the BVU-9x0 machines themselves are somewhat prone to shipping damage as they are rather large and weigh around 70lbs once all boxed up.

A better comparison might also be a first generation U-Matic recording as that'd have the least ringing to begin with.