Hi8 capture—digitizing using Digital8 camcorder vs. a professional service

[stuffdvideo]

I've got a lot of tapes taken with a Sony Hi8 consumer camcorder during the 1990s. I've digitized all the tapes using a Sony Digital8 camcorder, which digitizes within the camcorder and then outputs DV format video via a FireWire/1394 port. However, on some tapes, certain colors tend to have low saturation (especially greens in certain types of background foliage, though other greens are often OK) and/or color shifting (blue skies can take on a purplish hue). I suspect this is due to the age of the tapes.

Anyone have experience with whether a professional transfer service can do a better job extracting high-quality video from Hi8 tapes than Sony's own Digital8 camcorders do? There's a shop within reasonable driving distance that has the equipment to perform the conversion, but they won't do a test of a small amount of video, only entire tapes. I'd like to solicit opinions from folks with experience before I spend the money, potentially only to find out their conversion's no better than what I've already done.

Thanks in advance for any experience you can share.

[aramkolt]

If you want to just see what better equipment can do, send me a tape or two and a copy of what you were able to get with your setup on a flash drive. I basically have all of the high end Sony players including the pro ones like the EVO-9850 and EV-S7000 in great working condition and if you send me a copy of what you got, it should be pretty apparent if there's much improvement to be had. PM me if interested. I'm always interested in direct comparisons.

[aramkolt]

Also, another thing you should do is ask specifically what decks and capture chain will be used on your tapes if you use them. If they won't tell you, don't use them.

[Alwyn]

There's a shop within reasonable driving distance that has the equipment to perform the conversion, but they won't do a test of a small amount of video, only entire tapes.

Just give them one of your poor tapes to compare, unless they are charging too much. I can't imagine one tape would be a lot.

[Sharc]

Thanks in advance for any experience you can share.

Not my personal experience but you may want to take a look at this discussion:
https://forum.videohelp.com/threads/420331-Overexposed-U-Matic-Captures

These "professional shops" are often clueless idiots, but as @Alwyn wrote you would have to find out yourself by giving them a (critical) tape and see what they deliver. You may then upload a snippet of their result here and collect comments from members.

The restauration of difficult cases is time consuming, and - if the shop has the capabilities and knowledge at all - the cost can be prohibitive.

Another option would be to try to improve your existing DV captures yourself using an NLE or similar. NLEs usually have powerful color grading tools. For a start you could upload a snippet of your current critial DV capture and collect comments or proposals from members here.

[rgr]

I've got a lot of tapes taken with a Sony Hi8 consumer camcorder during the 1990s. I've digitized all the tapes using a Sony Digital8 camcorder, which digitizes within the camcorder and then outputs DV format video via a FireWire/1394 port. However, on some tapes, certain colors tend to have low saturation (especially greens in certain types of background foliage, though other greens are often OK) and/or color shifting (blue skies can take on a purplish hue). I suspect this is due to the age of the tapes.

Anyone have experience with whether a professional transfer service can do a better job extracting high-quality video from Hi8 tapes than Sony's own Digital8 camcorders do? There's a shop within reasonable driving distance that has the equipment to perform the conversion, but they won't do a test of a small amount of video, only entire tapes. I'd like to solicit opinions from folks with experience before I spend the money, potentially only to find out their conversion's no better than what I've already done.

Thanks in advance for any experience you can share.

Has the luma level been corrected from 16-255 to 16-235?
If not, then the D8 camera captures may have a low saturation level - lowering luma values simultaneously increases saturation.
Post a sample.

[VWestlife]

You already did the hard part. No sense in doing it all over again.

Using any decent video editor, you can adjust the brightness, contrast, color saturation, and tint of the captured footage to make it look better.

Just make sure you use a program which still knows how to properly deal with interlaced, non-square-pixel DV video. Some apps (like Microsoft Clipchamp or NCH VideoPad) do not support interlaced video, while others (like OpenShot) require tedious manual settings to be applied to every clip you add to the timeline.

[dellsam34]

I've digitized all the tapes using a Sony Digital8 camcorder, which digitizes within the camcorder and then outputs DV format video via a FireWire/1394 port. However, on some tapes, certain colors tend to have low saturation (especially greens in certain types of background foliage, though other greens are often OK) and/or color shifting (blue skies can take on a purplish hue). I suspect this is due to the age of the tapes.

Some D8 camcorders tend to blow out brightness levels with washed out colors out of analog recordings if used to convert the tapes, This has less to do with the DV codec and a lot to do with the internal ADC, The best way is to recapture them using the camcorder's Y/C output with an external capture device, Purple skies is a white balance calibration issue, Some consumer camcorders don't have an option to calibrate it for the scene being shot. Post a short sample to assess the quality, Maybe it's not worth re-doing.

[rgr]

Some D8 camcorders tend to blow out brightness levels with washed out colors out of analog recordings if used to convert the tapes, This has less to do with the DV codec and a lot to do with the internal ADC, The best way is to recapture them using the camcorder's Y/C output with an external capture device, Purple skies is a white balance calibration issue, Some consumer camcorders don't have an option to calibrate it for the scene being shot. Post a short sample to assess the quality, Maybe it's not worth re-doing.

D8 cameras output analog video after converting to DV, so that's poor advice.

[dellsam34]

Some D8 camcorders tend to blow out brightness levels with washed out colors out of analog recordings if used to convert the tapes, This has less to do with the DV codec and a lot to do with the internal ADC, The best way is to recapture them using the camcorder's Y/C output with an external capture device, Purple skies is a white balance calibration issue, Some consumer camcorders don't have an option to calibrate it for the scene being shot. Post a short sample to assess the quality, Maybe it's not worth re-doing.

D8 cameras output analog video after converting to DV, so that's poor advice.

Go learn circuit design basics. Chat GPT is not going to help unfortunatly.

[VWestlife]

D8 cameras output analog video after converting to DV, so that's poor advice.

At least according to my understanding of the block diagram, the analog playback is converted to digital, yes, but the DV compression/encoding is only used for the FireWire output. So analog playback through the composite/S-Video outputs is from uncompressed digital, not DV.

Page 46 of the PDF: Service manual for Sony Digital8 camcorders

[dellsam34]

That's correct, That's why I advised against certain D8 camcorders and suggested using the original Hi8 camcorder outputs, high end Hi8 camcorders tend to have better digital processing than cheap stripped down D8 camcorders.

[rgr]

Some D8 camcorders tend to blow out brightness levels with washed out colors out of analog recordings if used to convert the tapes, This has less to do with the DV codec and a lot to do with the internal ADC, The best way is to recapture them using the camcorder's Y/C output with an external capture device, Purple skies is a white balance calibration issue, Some consumer camcorders don't have an option to calibrate it for the scene being shot. Post a short sample to assess the quality, Maybe it's not worth re-doing.

D8 cameras output analog video after converting to DV, so that's poor advice.

Go learn circuit design basics. Chat GPT is not going to help unfortunatly.

It's possible, I'm basing it on what I've already found on the internet.
Even here: https://forum.videohelp.com/threads/388151-Capturing-via-DV-passthrough-or-with-S-vide...rd#post2512972

"1. Block diagram – analog inputs going into the digital path
In the OVERALL BLOCK DIAGRAM (4/4) (page 3‑7 in the PDF, section “3. BLOCK DIAGRAMS”):

The JK‑260 (JACK) board carries the A/V connectors.

On this board you see signals:

VIDEO IN/OUT

S‑Y I/O (luma for S‑Video)

S‑C I/O (chroma for S‑Video)

These signals are routed to the main board VC‑352, where the relevant blocks include:

VIDEO I/O

DV SIGNAL PROCESS

DV INTERFACE

This shows the signal path:
analog IN (VIDEO/S‑Y/S‑C) → VIDEO I/O → DV SIGNAL PROCESS / DV INTERFACE
i.e. analog → digital (DV) inside the camera.

The diagram also distinguishes:

Arrows labeled “VIDEO SIGNAL” (analog video) going from JK‑260 into VC‑352 blocks.

Then, within VC‑352, blocks such as:

DV SIGNAL PROCESS

DV INTERFACE

VIDEO IN/OUT

This confirms that the camera treats the analog input as a source that is processed in the digital (DV) domain.

2. Schematic JK‑260 – physical connections of the analog inputs
In the SCHEMATIC DIAGRAMS → JK‑260 BOARD (JACK) (page 4‑59 in the PDF):

Connector J701 is labeled:

pin 1: VIDEO/AUDIO

pin 2: LANC

pin 3: S‑Y

pin 4: S‑C

Signals from these pins are labeled:

VIDEO IN/OUT

S‑Y I/O

S‑C I/O

These lines then go through connector CN704 to the main board VC‑352 (CN1208), where they appear as:

VIDEO_I/O

S_Y_IO

S_C_IO

This is the “hard” connection: from the analog jacks → to the main board → into the VIDEO I/O / DV processing block.

3. LCD and outputs coming from the same digital path
In the same overall block diagram (pages 3‑2 to 3‑7):

Display-related blocks:

LCD901 / LCD902 (LCD and EVF)

PD‑213 board (LCD RGB drive, backlight)

Signals to LCD/EVF come from blocks on VC‑352 labeled:

CAMERA PROCESS

MPEG MOVIE / DIGITAL STILL PROCESS

DV SIGNAL PROCESS

This means the image on the LCD/EVF comes from the digital processing path (after ADC/DV processing), not directly from the analog input in raw form."


However, I haven't seen a D8 camera with separate analog input and output, so the thread author probably doesn't have one anyway.

[rgr]

D8 cameras output analog video after converting to DV, so that's poor advice.

At least according to my understanding of the block diagram, the analog playback is converted to digital, yes, but the DV compression/encoding is only used for the FireWire output. So analog playback through the composite/S-Video outputs is from uncompressed digital, not DV.

Page 46 of the PDF: Service manual for Sony Digital8 camcorders

Do any of these cameras have both an analog input and output?
I have one and it doesn't. Isn't that a LCD output?

[dellsam34]

Most consumer camcorders I've seen have one set of connections used for either input (VCR mode if equipped) or output (Playback), Most Prosumer and pro camcorders do have separate inputs and outputs.

[VWestlife]

When I have some time I can do comparisons, using a commercially pre-recorded Video8 tape, between:

Video8 camcorder composite output (no TBC) -> analog input of Digital8 camcorder (TBC on) -> DV out
Hi8 camcorder S-video output (TBC off) -> analog input of Digital8 camcorder (TBC on) -> DV out
Hi8 camcorder S-video output (TBC on) -> analog input of Digital8 camcorder (TBC off) -> DV out
Playing the tape in the Digital8 camcorder itself (TBC on) -> DV out

[rgr]

I would be interested in the results, although I suppose the quality will be similar (if the tape is in good condition), only the color reproduction will probably be different.

[stuffdvideo]

If you want to just see what better equipment can do, send me a tape or two and a copy of what you were able to get with your setup on a flash drive..

Thanks for the offer, though I'm a bit wary of shipping tapes considering how careless carriers can be. Are you located in the USA or a different country?

[stuffdvideo]

Also, another thing you should do is ask specifically what decks and capture chain will be used on your tapes if you use them. If they won't tell you, don't use them.

That's good advice. If they can't or won't give me specifics, I'll steer clear of them.

[stuffdvideo]

There's a shop within reasonable driving distance that has the equipment to perform the conversion, but they won't do a test of a small amount of video, only entire tapes.

Just give them one of your poor tapes to compare, unless they are charging too much. I can't imagine one tape would be a lot.

They charge $20 to convert a 1-hour tape to mp4 format, double that to output to ProRes. That's tolerable for a one-off test, but at those prices having them re-digitize all ~75 Hi8 tapes would get pricey fast, especially as my goal would be to have high-quality transfers suitable for upscaling (using Topaz Video AI).

[stuffdvideo]

[QUOTE=Sharc;2801137]

Another option would be to try to improve your existing DV captures yourself using an NLE or similar. NLEs usually have powerful color grading tools. For a start you could upload a snippet of your current critial DV capture and collect comments or proposals from members here.

I'm on a Mac and use Final Cut Pro X as my NLE, with which I've done some color grading before. I've read that Davinci Resolve's color grading tools are superior, but haven't tried them out. Maybe it would be worth my while.

Things get a little complicated as it's not a matter of all colors being desaturated, nor even all shades of a color. For instance, in some clips, foliage of one shade of green appears almost completely desaturated while another shade of green looks good. Attempting to compensate for this in FCP X has been extremely time consuming, at least at my (admittedly limited) skill level.

[Alwyn]

That's tolerable for a one-off test, but at those prices having them re-digitize all ~75 Hi8 tapes would get pricey fast

Fair enough, a one-tape test would verify whether your camera is dodgy or whether it is actually a poor tape. Dodgy camera? Buy/try another. Poor tape? stick with your current captures.

As far as the argument between DV Out and Analogue Out from a camcorder, here's a test I did (this was a Video8 tape, not Hi8); IMO no practical difference:

https://youtu.be/Ap59r45OmOw

The analogue capture, of course, requires a separate digitiser, which for a mac, could limit your options.

Doubling the cost for ProRes is a ripoff, unless they are using some whizzbang capture chain and not camcorder+Firewire.

[rgr]

There's a shop within reasonable driving distance that has the equipment to perform the conversion, but they won't do a test of a small amount of video, only entire tapes.

Just give them one of your poor tapes to compare, unless they are charging too much. I can't imagine one tape would be a lot.

They charge $20 to convert a 1-hour tape to mp4 format, double that to output to ProRes. That's tolerable for a one-off test, but at those prices having them re-digitize all ~75 Hi8 tapes would get pricey fast, especially as my goal would be to have high-quality transfers suitable for upscaling (using Topaz Video AI).

Never upscale SD footage from VHS (or Video8) using Topaz. Ideally, you shouldn't upscale it at all, but doing so with Topaz is the best way to ruin your recordings (regardless of the specific model).

Furthermore, depending on the model, Topaz does NOT upscale video - it uses ffmpeg's simple Lanczos algorithm for upscaling. And people pay $300 for that...

[rgr]

[QUOTE=stuffdvideo;2801197]

Another option would be to try to improve your existing DV captures yourself using an NLE or similar. NLEs usually have powerful color grading tools. For a start you could upload a snippet of your current critial DV capture and collect comments or proposals from members here.

I'm on a Mac and use Final Cut Pro X as my NLE, with which I've done some color grading before. I've read that Davinci Resolve's color grading tools are superior, but haven't tried them out. Maybe it would be worth my while.

Things get a little complicated as it's not a matter of all colors being desaturated, nor even all shades of a color. For instance, in some clips, foliage of one shade of green appears almost completely desaturated while another shade of green looks good. Attempting to compensate for this in FCP X has been extremely time consuming, at least at my (admittedly limited) skill level.

If you want to process footage from the tape in Resolve, you first need to convert the matrix from 601 to 709; Resolve doesn't support 601. Final Cut probably doesn't either.

[stuffdvideo]

Never upscale SD footage from VHS (or Video8) using Topaz. Ideally, you shouldn't upscale it at all, but doing so with Topaz is the best way to ruin your recordings (regardless of the specific model).

Furthermore, depending on the model, Topaz does NOT upscale video - it uses ffmpeg's simple Lanczos algorithm for upscaling. And people pay $300 for that...

That's a matter of opinion. Topaz isn't perfect by any means, and can make video look "plastic", but I've found my audiences (mostly friends and family) far more tolerant of slightly-unreal looking (but clean) upscaled video than they are of grainy SD video blown up to fill the screen. This is especially true when viewed on computer monitors, which don't have all the built-in processing TVs use to make SD interlaced video look better on modern screens. And as my goal is to produce something my audiences will want to watch, their satisfaction with the end result is what matters.

Also, I'm often cutting between old (recorded in analog SD) and new (recorded at 4K) video so the SD video's going to be deinterlaced/upscaled by my NLE anyway, and other tools are better at it (QTGMC/Hybrid at deinterlacing and Topaz at the upscaling).

And yes, Topaz does rely on ffmpeg but also applies proprietary filters on top. Opinions online vary about whether it adds enough "special sauce" to be worthwhile. For me, the fact it puts a GUI on top of ffmpeg so I don't have to do everything from the command line makes it worthwhile, especially as I bought a license on sale prior to it going to a subscription basis.

[stuffdvideo]

You already did the hard part. No sense in doing it all over again.

Using any decent video editor, you can adjust the brightness, contrast, color saturation, and tint of the captured footage to make it look better.

Just make sure you use a program which still knows how to properly deal with interlaced, non-square-pixel DV video. Some apps (like Microsoft Clipchamp or NCH VideoPad) do not support interlaced video, while others (like OpenShot) require tedious manual settings to be applied to every clip you add to the timeline.

My NLE is Final Cut Pro X, which does support interlaced video (though it no longer directly supports .dv format files; I have to use ffmpeg to put the video into a .mov wrapper for FCP X to be able to use it). However, my experiments so far have shown attempting to correct the color issues inside of FCP X is slow and tedious. As the desaturation and color-shifts aren't across the board, but typically affecting specific shades of various colors more than others, applying corrections that improve some colors without negatively impacting any others is difficult. I've been successful at stacking copies of the clip in the timeline, applying different corrections to different layers, and then applying masks to cut out sections negatively affected by corrections but this method is way too time consuming to make regular use of.

[aramkolt]

DV content really isn't the greatest for editing as it is lossy.

Not saying this is the "right" way to do it, but if you already have DV files and don't want to recapture (and all you're really worried about is color correction etc), I'd first do some slight upscaling/deinterlacing/set aspect ratio to 4:3 (with square pixels)during the upscale/color space conversion from Rec601 to Rec709. Something like 960x720 or 1440x1080 would work. You have to decide if you want to turn each field into its own frame resulting in ~60fps (actually 59.94) or go with ~30fps (29.97). The former takes up twice as much space and some prefer that look, but you are being more true to the original content that way.

It would have probably been better to just capture into ProRes originally if you want to use an all Mac friendly capture chain. ProRes hardware encoders do typically allow capturing of illegal levels which is a plus as you don't run into clipping of whites/crushing of black levels as much, but once the information has been passed through a DV converter and output as DV, I would suspect that the clipping could be permanent. ProRes captures typically involve a TBC that goes from analog to SDI (these don't tend to be consumer units as SDI was never really used outside of the professional realm - it's kind of like HDMI over a single BNC cable and also can carry audio) and then the capture device captures the SDI signal.

I still like the AJA KiPro myself for this task (writes ProRes files to a removable hard drive), but the only actual analog input it has is component, so you need a TBC with at component output, or you can always use the SDI/HDMI inputs as well if you have a TBC that happens to output those. I've found something like the For.A FA-310 does pretty well at conversion to component from S-Video while also acting as a frame TBC. Main disadvantage of going with the FA-310 it uses internal 8 bit processing whereas more modern TBCs can do 10bit that ProRes can benefit from. You'd get no such benefit from DV or the typical Virtualdub lossless captures though as those are all 8 bit as well, so It's really "no worse" than those. Snell and Wilcox units tend to be pretty good at handling consumer formats as well and are generally more modern and are 10 bit.

It's kind of amazing that the FA-310 still holds up these days when it was released all the way back in 1989 - I guess that is also an advantage - incoming signals probably weren't expected to be very stable back then!

[rgr]

Never upscale SD footage from VHS (or Video8) using Topaz. Ideally, you shouldn't upscale it at all, but doing so with Topaz is the best way to ruin your recordings (regardless of the specific model).

Furthermore, depending on the model, Topaz does NOT upscale video - it uses ffmpeg's simple Lanczos algorithm for upscaling. And people pay $300 for that...

That's a matter of opinion. Topaz isn't perfect by any means, and can make video look "plastic", but I've found my audiences (mostly friends and family) far more tolerant of slightly-unreal looking (but clean) upscaled video than they are of grainy SD video blown up to fill the screen.

Just as Topaz should never be used to upscale SD material, it should ABSOLUTELY never be used for image denoising -- it's one of the worst tools for this purpose. Tools like AviSynth or Neat Video (many times cheaper, many times better) are used for this purpose. That's not an opinion -- it's a fact. Just look at the tests on YouTube.

This is especially true when viewed on computer monitors, which don't have all the built-in processing TVs use to make SD interlaced video look better on modern screens. And as my goal is to produce something my audiences will want to watch, their satisfaction with the end result is what matters.

I don't know how something like this can be satisfying... And this is typical Topaz behavior, sometimes it's better, sometimes it's worse -- but overall it's destruction, not improvement. And this material was heavily denoised anyway (because Topaz's degraining sucks) before being run through Topaz. And even then, it turned out terrible.
https://files.fm/u/k3vxag24fa

And yes, Topaz does rely on ffmpeg but also applies proprietary filters on top. Opinions online vary about whether it adds enough "special sauce" to be worthwhile. For me, the fact it puts a GUI on top of ffmpeg so I don't have to do everything from the command line makes it worthwhile, especially as I bought a license on sale prior to it going to a subscription basis.

Which filters are not suitable for SD at all? Neat and/or Avisynth filters are used, including for upscaling. Much cheaper, much faster, and the effect is many times better.

[stuffdvideo]

Just as Topaz should never be used to upscale SD material, it should ABSOLUTELY never be used for image denoising -- it's one of the worst tools for this purpose. Tools like AviSynth or Neat Video (many times cheaper, many times better) are used for this purpose. That's not an opinion -- it's a fact. Just look at the tests on YouTube.

The point of this thread really isn't Topaz and whether or not it's a good way to upscale video, so I'm not going to continue that part of the discussion. The question's whether there's much chance a video transfer service could extract significantly better quality from old tapes recorded on a consumer Hi8 camcorder than the transfers I've already done using a Digital8 camcorder. Hopefully the thread can return to the intended topic.

[rgr]

How are we supposed to know, since you didn't include a sample, what did I write about 25 posts ago? Secondly, if you want to run it through Topaz, it doesn't really matter anyway.