JVC HR-S7600AM S-VHS VS. HM-DH5U D-VHS Capturing a VHS Tape

[dellsam34]

Here is two 30 seconds lossless AVI samples of the same section of a VHS tape played back on both machines in analog mode via S-Video, Both set to "Edit" mode, The 7600AM B.E.S.T is set to on, The equivalent "Video Calibration" is set to on for the HM-DH5U, The 7600AM has a line TBC on, but the HM-DH5U don't have a setting for it, I guess it is automatically on, it does have video stabilizer but the manual states it's for EP tapes only.

Both VCR's hooked up to the same usual workflow, S-Video/YC -> BE75 -> SDI -> USB3 -> Windows 11 with mechanical HDD.

I leave the decision to the members to analyze and judge.

The HM-DH5U D-VHS deck has 4 ways to get video out of a tape, S-Video/Composite, Component, iLink (MPEG-2) and HDMI, Each output requires a compatible capture card, Maybe in the future I will do 4 samples from each output, but I don't believe anything will be superior to the S-Video out for VHS tapes.

[lollo]

Thanks for the data!

Quick view on my side, no deinterlace, just a preliminary impression.

I prefer HM-DH5U, even if it crops more on the top and has more head switching noise and has more crushed blacks.

Less soften video, more noise, maybe more details, better overall look.

https://imgsli.com/MTk5OTQz

https://imgsli.com/MTk5OTQ0

https://imgsli.com/MTk5OTQ2

[Skiller]

The HR-S7600AM has that very typical soft and "noise reduced" JVC-look, which some people love and others, like me, not so much.

I prefer the HM-DH5U, as it seems to give a less processed, more raw output – giving me more options for filtering myself in AviSynth to a result I like.

[Bwaak]

The DH5U is noisier, while not having significantly more detail. It is also greener, which is noticeable even without color diagram, BTW how do you generate it? The S7600 looks good out of the box.

[dellsam34]

In my opinion the S7600AM colors are more accurate than the DH5U, Plus line TBC is more pronounced in the S7600AM, I've only used it once for one YT video and I think I'm going back to the S7600. DNR softening doesn't bother me much, it saves me a step from doing it in software.

[lollo]

Looking again to your sample, I still prefer the HM-DH5U: more natural colors, more details, sharper (and noiser) image, same lineTBC correction (vertical and diagonal lines)

https://imgsli.com/MjAwMDcz

see also how the N.R. reduced the width of the diagonal lines in the 7600.

But you are the owner and can run multiple tests, so nobody better than you can decide

[dellsam34]

I specifically posted these samples because I was asked numerous times on how D-VHS machines handle VHS tapes vs conventional VCRs, So it's really not about me. I just wasn't expecting that much difference and if it is age related or just different processing.

[lollo]

I just wasn't expecting that much difference and if it is age related or just different processing.

Difficult to answer. Is true that in 2023 even the machines in best conditions are starting to provide lower quality output because aging. It is the case with my JVCs

[Bwaak]

Looking again to your sample, I still prefer the HM-DH5U: more natural colors, more details, sharper (and noiser) image, same lineTBC correction (vertical and diagonal lines)

https://imgsli.com/MjAwMDcz

see also how the N.R. reduced the width of the diagonal lines in the 7600.

The DH5U is noisy without adding more detail. On contrary, look at the headphone on the guy on the right: on the 7600 you can see the letter "S" (at least I think it is an "S"), on the DH5U there are just two blobs. Look at the T-shirt on the guy on the left: the horizontal chroma streaks are ugly, especially the ones to the right of the white print. Or look at his chin, you can see shade that looks like checkerboard on the DH5U. I don't know is it different interlaced chroma processing, or is it "just" noise reduction, but 7600 looks much better. I would like you see you processing the DH5U image to get rid of these streaks while preserving overall detail.

The 7600 is cleaner, better color, and in some cases even has more detail!

7600 all the way.

[dellsam34]

I can also make out the microphone grille in the 7600, I can always tone down the 7600 chroma in the proc amp of the BE75 and it wouldn't hurt anything, but if I pump the chroma for the DH5U it will come at a cost of more noise. This why I think the 7600 is better.

[lollo]

On contrary, look at the headphone on the guy on the right: on the 7600 you can see the letter "S" (at least I think it is an "S"), on the DH5U there are just two blobs.

You must be joking.

Look at the T-shirt on the guy on the left: the horizontal chroma streaks are ugly, especially the ones to the right of the white print.

No way. On the other hand, look at the shirt of the guy in the right of the picture I posted. More details, the "lines" almost disappeared in the 7600.

I stop here, you'll be able to argue and continue your non sense forever

[dellsam34]

It's okay, these are just personal preferences and opinions, no need to escalate guys.

No way. On the other hand, look at the shirt of the guy in the right of the picture I posted. More details, the "lines" almost disappeared in the 7600

Yes you scored on that one.

[Bwaak]

On the other hand, look at the shirt of the guy in the left of the picture I posted. More details, the "lines" almost disappeared in the 7600.

I agree, the guy on the right has better defined wrinkles and shades on his T-shirt. It is a matter of taste, I'd prefer to get rid of the chroma streaks at the cost of losing some shading.

[dellsam34]

Keep in mind, this is the 90's DNR, Not bad for the era, I'm pretty sure there are better filters out there if someone needs to go that route.

[Brad]

Thanks for the samples.

Can you do an EP mode comparison?


See here for an example of JVC S-VHS being much better than D-VHS for EP: https://www.digitalfaq.com/forum/video-capture/11964-vcr-observations-usage.html

I don't have a JVC S-VHS to compare to, but I ruled out using my HM-DH5U for anything other than SP, as even in EDIT the EP output is mush.

[dellsam34]

I don't doubt it, By that time in the VHS era when JVC planned to move forward with D-VHS, VHS was stripped down to a basic playback let alone low speed VHS, Those decks don't even have a flying erase head, That tells you that JVC was focusing on digital recording not analog.

[Skiller]

Let's put it this way; if I were to watch a tape on the HR-S7600AM on a TV and that's it, I would prefer it for it's low noise. Especially the chroma is much cleaner.
If I were to capture a tape, I would prefer the HM-DH5U. I don't mind the increased noise – I am going to do filtering in AviSynth in any case anyways, and having a somewhat less processed and potentially ever so slightly more detailed picture to start with seems desirable to me in that case.

[Bwaak]

@Skiller, I would like too see how you process the DH5U video to get rid of chroma streaks. Will the result be comparable to what the 7600 produces out of the box.

[lollo]

Let's put it this way; if I were to watch a tape on the HR-S7600AM on a TV and that's it, I would prefer it for it's low noise. Especially the chroma is much cleaner.
If I were to capture a tape, I would prefer the HM-DH5U. I don't mind the increased noise – I am going to do filtering in AviSynth in any case anyways, and having a somewhat less processed and potentially ever so slightly more detailed picture to start with seems desirable to me in that case.

Yes.

The potential drawback in HM-DH5U is that the common JVC TBC/N.R. defect of the temporal chroma denoise affecting the frames at scene changes
https://forum.videohelp.com/threads/401232-JVC-S-VHS-settings-for-capture-%28again%29
https://forum.videohelp.com/threads/395451-Chroma-mess
seems more evident compared to HR-S7600AM, especially in case 2.

In this sample the defect stays for 1/2 frames only so is not that important anyhow.

case 1 hat visor on the right contamination




case 2 overall beige color contamination

[Skiller]

@lollo Good observation!

@Skiller, I would like too see how you process the DH5U video to get rid of chroma streaks. Will the result be comparable to what the 7600 produces out of the box.

Well yes, are you serious? This is 90's technology filtering. I and others can easily reach and surpass that. Maybe I will give it a go later.

I understand it is a personal preference whether someone wants to do all this in software or would rather have an acceptable result out of the box and not worry about it.

[Sharc]

The potential drawback in HM-DH5U is that the common JVC TBC/N.R. defect of the temporal chroma denoise affecting the frames at scene changes.

Doesn't the chroma bleeding exist between all frames? It's just most obvious at sceen changes, no?

[dellsam34]

I understand it is a personal preference whether someone wants to do all this in software or would rather have an acceptable result out of the box and not worry about it.

This is me, I don't have time to sit and tweak and re-tweak parameters, This is one of the reasons I haven't jumped on the vhsdecode bandwagon. Very few extreme cases that I go that route, My main focus is quality capture not restoration.

[Skiller]

Well, vhsdecode is way above my head as well.

But it's not like you have to re-invent the wheel with every capture and make a new script tweaking 50 parameters. I find that I can basically get away with copy and paste most of the time.

@Skiller, I would like too see how you process the DH5U video to get rid of chroma streaks.

There you go. Chroma noise gone. Very, very mild filtering on the Luma as well. Looks quite good for VHS – in my opinion.

The first frame of some scene changes is somewhat messy, but it was like that in the source as pointed out by lollo and is likely because of some noise reduction/processing inside the DH5U (which is apparently not very effective).


https://imgsli.com/MjAwNDk2

https://imgsli.com/MjAwNDk3

(Doesn't show well in still images though.)

[dellsam34]

I still prefer the primitive DNR of the S7600AM on the right.

https://imgsli.com/MjAwNTg5

[themaster1]

@Skiller, I would like too see how you process the DH5U video to get rid of chroma streaks. Will the result be comparable to what the 7600 produces out of the box.

I've posted a script before

avisource().assumetff()
converttoyv16(interlaced=true)
orig=last
ev=orig.assumetff().separatefields().selecteven()
od=orig.assumetff().separatefields().selectodd()
ev
ue_chroma = UToY(ev).SpotLessUV() #do not use YV12 for spotlessuv
ve_chroma = VToY(ev).SpotLessUV() #do not use YV12 for spotlessuv
YToUV(ue_chroma, ve_chroma)
MergeLuma(ev)
ev_filtered=last
od
uo_chroma = UToY(od).SpotLessUV() #do not use YV12 for spotlessuv
vo_chroma = VToY(od).SpotLessUV() #do not use YV12 for spotlessuv
YToUV(uo_chroma, vo_chroma)
MergeLuma(od)
od_filtered=last
interleave(ev_filtered,od_filtered)
assumefieldbased().assumetff().weave()

Also, CCD can be handy, sometimes both CCD & spotless are necessary

ConverttoRGB32(matrix="rec601",interlaced=true)
separatefields()
LoadVirtualDubPlugin("C:\Program Files (x86)\Virtualdub\plugins32\ccd_sse2.vdf", "CCD", 0)
CCD(10,1) # 0 to 100 # Défaut =30, retail: between 7-15
weave()
converttoyv12(matrix="Rec601",interlaced=true)

[Skiller]

I still prefer the primitive DNR of the S7600AM on the right.

The S7600AM does what I would call rather strong denoising on the Luma, which I purposefully did not replicate, just because I like it better with some noise left. But that's a matter of preference of course.

However, there is still some chroma noise in the S7600AM capture! What about that?



In any case, this is the script I used. It works in 16 Bits per channel to avoid even the slightest introduction of banding due to the denoising, especially on the chroma.
(Keep in mind I still use AVS 2.6, so that's why there's all the lsb stuff.)

AviSource("HM-DH5U.avi", pixel_type="YUY2")

AssumeBFF()


#some bob-deinterlacer here. I used a dumb bob because the result is going to be interlaced
#and the video is telecined so QTGMC is not really appropriate for that kind of motion.
Bob(0,1)


#interleaved to planar (lossless, YUY2 to YV16), also 8 Bits per Channel to 16 Bits
ConvertToYV16()
dither_convert_8_to_16()


#spatial chroma denoising
dfttest(Y=false, tbsize=1,sigma=16, lsb=true, lsb_in=true)


#temporal chroma denoising
SMDegrain(chroma=true, plane=3, contrasharp=false, tr=3, thSADC=550, lsb_in=true, lsb=true, lsb_out=true)


#temporal luma denoising
SMDegrain(chroma=false, plane=0, contrasharp=false, tr=2, thSAD=180, lsb_in=true, lsb=true, lsb_out=true)


#optional, warps the chroma to luma boundaries to reduce chroma bleed, conservative settings
aWarpSharp4xx(chroma=6, thresh=90, type=0, blur=6, depth=28, lsb=true)


#dither back to 8 Bits per channel
ditherpost(mode=6) 


#re-interlace, requires padding because the video height is 486 which results in an uneven field height (243)
AddBorders(0,0,0,2)

AssumeBFF().SeparateFields().SelectEvery(4,0,3).Weave()

Crop(0,0,0,-2)

#planar to interleaved (lossless, YV16 to YUY2)
ConvertToYUY2()

The four parameters in bold are the only really important ones.

[lollo]

Doesn't the chroma bleeding exist between all frames? It's just most obvious at sceen changes, no?

Yes and no. The JVC N.R. on chroma is a temporal filtering (my guess), so it suffers more at scene change because there is contamination from previous frame. IMO there is no "intrinsic" bleeding between frames, just the radius is higher than 1. Just speculating.

In any case, this is the script I used.

I quickly did something similar, and obtained this. I did not compare against yours, just versus originals; no competition, just experimenting

JVC-HM-DH5U 5 versus HM-DH5U 5 restored:

https://imgsli.com/MjAwNjY0

https://imgsli.com/MjAwNjY1

JVC-HR-S7600AM versus HM-DH5U restored (sorry, I forgot to shift one of the 2 images and align for a better comparison):

https://imgsli.com/MjAwNjY2

https://imgsli.com/MjAwNjY3

The output file: HM-DH5U_restored.avi

What is missing for a complete picture is a restoration on JVC-HR-S7600AM (which can use for example a temporal radius = 1 or 2 for denoising, because its capture ic "cleaner").

I did not adress specifically addressed the chroma aspects, if not included in the filters I used.

Edit: the restored clip has been kept interlaced, or better "lossless deinterlaced / filtered / interlaced back" for better comparison. Working on a deinterlaced clip would allow the filtering to be more effective