HDMI vs. SDI Analog Tape Video Capture

[dellsam34]

It turns out the TBS800 was switched to scaling mode by mistake when I did the test captures, that's why it looked dull, I switched it off by unchecking "Sync Mode" in Image tab, Unbelievable! I have replaced the files once again, Sorry guys for wasting your time.

I removed the HR-S7600 sample with TBS800's wrong setting, The VCR is on the workbench now, I will upload one later after it has been serviced.

[Sharc]

No problem. It happens to the best uf us
The winner so far is now the HM-DH5U-TBS800-SDI, for my eyes (levels, sharpness, details, contrast)

[dellsam34]

I'm guessing you used the non legalized sample, Right?

I would say at a frame level all devices are pretty close, But I think the TBS800 has some extra tools that help produce better captures if used correctly when needed, I like the analog DNR, it seems to work better than the VCR's one.

Before we declare the TBS800 a winner, I would like to see more opinions from other members.

[Sharc]

I'm guessing you used the non legalized sample, Right?

updated Post#11, last sample. (I think this is non-legalized).

Here the parade for comparison.

[dellsam34]

I've updated post #12 that contains the two extra samples as well.

[Sharc]

All 7 interleaved. The legalizer eliminated a few more "hot" pixels, but the non-legalized was also nearly perfect IMO.
(Note that all have been converted to 4:2:0 for this comparison. I don't think this has an impact on the ranking).

v1=AVISource("HM-DH5U-HDMI.avi").assumetff().tfm().tdecimate().subtitle("HM-DH5U-HDMI.avi").crop(0,2,-0,-4)
v2=AVISource("HM-DH5U-BE75-SDI.avi").assumetff().tfm().tdecimate().subtitle("HM-DH5U-BE75-SDI.avi").crop(0,2,-0,-4)
v3=AVISource("HM-DH5U-Miranda-SDI.avi").assumetff().tfm().tdecimate().subtitle("HM-DH5U-Miranda-SDI.avi").crop(0,2,-0,-4)
v4=AVISource("HM-DH5U-USB-500-USB.avi").assumetff().tfm().tdecimate().subtitle("HM-DH5U-USB-500-USB.avi").crop(0,0,-0,-0)
v5=AVISource("HM-DH5U-TBS800-SDI.avi").assumetff().tfm().tdecimate().subtitle("HM-DH5U-TBS800-SDI.avi").crop(0,2,-0,-4)
v6=AVISource("HM-DH5U-TBS800-SDI_Legalizer.avi").assumetff().tfm().tdecimate().subtitle("HM-DH5U-TBS800-SDI-Legalizer.avi").crop(0,2,-0,-4)
v7=AVISource("HM-DH5U-TBS800 ALL DNR.avi").assumetff().tfm().tdecimate().subtitle("HM-DH5U-TBS800 ALL DNR.avi").crop(0,2,-0,-4)

last=interleave(v1,v2,v3,v4,v5,v6,v7).histogram().assumefps(2)
return last

[dellsam34]

Nice work, thanks.

[lollo]

My impressions:

HM-DH5U-TBS800-SDI_Legalizer.avi versus HM-DH5U-HDMI.avi -> small preference to HDMI capture

https://imgsli.com/MzE0OTYw

HM-DH5U-TBS800-SDI_Legalizer.avi versus HM-DH5U-BE75-SDI.avi -> difficult to choose, small preference to TBS800

https://imgsli.com/MzE0OTY1 (no equalization here)

Pinnacle USB-500 suffers from gaps in the histograms, so is out of contest; however, but its output is not that bad:

https://imgsli.com/MzE0OTcy

[Sharc]

My observations:

HM-DH5U-HDMI.avi
- Elevated darks
- Brights exceed luma 235, but not clipped => risk of loosing details if expanded to full range RGB
=> "foggy" look due to the elevated darks
=> all above can be corrected in post though

HM-DH5U-BE75-SDI.avi
- Clipped brights at luma 255
=> clear loss of details in the brights, not recoverable

HM-DH5U-Miranda-SDI.avi
- Soft/blurry picture
- Gaps in the REDs (RGB histogram)

HM-DH5U-USB-500-USB.avi
- Hard clipping at luma 235 => loss of details in the brights, not recoverable
- Spiky histogram

HM-DH5U-TBS800-SDI.avi
- nearly perfect, few out-of-gamut

HM-DH5U-TBS800-SDI_Legalizer.avi
~perfect, even less out-of-gamut

HM-DH5U-TBS800 ALL DNR.avi
~perfect, similar

Bottom line:
- Winner is TBS800 overall
- the benefits of the legalizer might be more prominent for poorer sources with more illegal YCbCr
- Second comes HDMI after post processing (levels adjustment)

[dellsam34]

To add:
- The HDMI route incur a step of adjusting luma level after it has been digitized already, It also clips the top 6 scan lines of the 22 lines reserved to data (so does any consumer capture device), sometimes useful for extracting caption and teletext, And for me this is a NTSC deck only, so other formats tapes are out of luck.

- The TBS800 is bulky, half rack size, noisy fan, bulky D-Sub custom made audio cable.

- The BE75 size is perfect, has YPbPr input for up to 1080i which the rest lack for capturing analog SD and HD materials.

- Both the TBS800 and BE75 have controls on the front panel or via a computer app, the rest rely on the capture software controls, so they can't be used with an external recorder.

I would love to try the Singmai capture device but not willing to fork a grand for just this purpose.

[Sharc]

- The BE75 size is perfect, has YPbPr input for up to 1080i which the rest lack for capturing analog SD and HD materials.

- Both the TBS800 and BE75 have controls on the front panel or via a computer app, the rest rely on the capture software controls, so they can't be used with an external recorder.

Some way it should be possible to adjust the levels and avoid the brights clipping of the BE75. It's the worst capture of your samples now, IMO.

[Attachment 83306 - Click to enlarge]

I would love to try the Singmai capture device but not willing to fork a grand for just this purpose.

Same here, apart from the risk of a disappointment with the remake.

[Bogilein]

I dragged the 7 files into my NLE and then adjusted the levels to the 16-235 range.

At first glance, my favourite would be the HDMI recording followed by the TBS800DNR & TBS800Legalizer.

The worst recording for me is the Brighteye75 capture, followed by the Miranda and the USB500.

The most disappointed capture for me is the one with the Brighteye75. Blurred video, clipping, too strong colours in contrast to the other captures.

Were the levels adjusted during capturing to prevent the clipping on the USB500 and the Brighteye75?

I would love to try the Singmai capture device but not willing to fork a grand for just this purpose.

The Singmai SM03 device is often mentioned when people are hoping for even better captures. Nobody would know about the Singmai unit if dellsam34/Latreche hadn't posted about it on various video forums. If you're lucky there might be 10 of both versions sold worldwide and I'm not sure there were that many.
Since July you can no longer pay via Paypal. The shop has been closed since mid-August due to the owner's ‘health problems’. Since mid-October the Singmai websites and also the private pages of the owner are offline. It is better not to worry about it anymore.
Basically, the device was never alive.

[dellsam34]

No, I did not adjust any levels for any capture device, I did not check the BE75 setting, It may have some setting off the default, I will follow up with a post when I get time to check it.

[Sharc]

I dragged the 7 files into my NLE and then adjusted the levels to the 16-235 range.

Ok, but this is already a first step in post processing. Without this, I preferred the TBS800 variants and ranked the HDMI 2nd, as some NLEs treat HUFFYs interlaced 4:2:2 badly by converting and expanding them to full RGB upon import, AFAIK, loosing everything outside the 16...235 range (trying to find the posts on this subject from poisondeathray ...... ).

[aramkolt]

BE75 does have a luma adjustment in the brighteye control app, but I don't think it does anything in terms of legalization or automatic gain control.

I'm not totally clear on this, but it seems that different VHS tapes have wildly different luma levels with commercial tapes going as high as 130IRE or more when played out of a standard VCR. This could partially be some sort of extra copy protection maybe?

So the question is whether you'd need to just find the brightest scene in each tape you're capturing with the BE75 and adjust it so that brighter values aren't clipping for that specific tape, or it could be that if you are using it with a specific VCR that it'll have a max luma output and you just set the BE75 so that luma gain is such that it doesn't get over 100IRE for that specific VCR maybe.

Some VCRs will allow you to adjust luma as well using a potentiometer - AG1980 in particular is well-labeled as to which potentiometer it is, others may be changeable via service remotes, or you could just use an analog proc amp prior to the BE75.

In realtime, you could look at an SDI waveform monitor viewing the output of the BE75, though I'm guessing there are software SDI scopes as well, but may vary based on the SDI capture card you have in use.

[Sharc]

Worst case you have to insert an attenuator pad into the analog signal path before the signal enters the device which causes the clipping. It should be a '75 Ohms pad' (standard terminating impedance) rather han just some potmeter or resistor in order to prevent unnecessary signal reflections.