VHS over HDMI capture workflow with sample clip

[robertzombie]

Playing around with a colour bar generator going directly into the Panasonic's s-video input.

In the Startech's capture pin settings, there is the option to "Shrink output colour range", bypass, and expand. I did a test capture with this setting Bypassed, and Shrink, and I'm getting different white levels on my capture. I think "shrink" is correct, but it's been a while since I've played with proper hardware scopes so I could be mistaken.

[robertzombie]

VDub logs more metadata than ATV

How can that be accessed?

I use "Media Info" https://mediaarea.net/en/MediaInfo

[Sharc]

The YUV levels of your capture look fine and you can leave these alone and let the player/TV do its job. No need to proactively stretch or shrink the YUV levels. I don't have Startechs, but my best guessing is to select "bypass" (if this means leave it alone) ....

[Attachment 91029 - Click to enlarge]

[robertzombie]

I'm wondering if "bypass" output colour range is clipping 100% white? Then again, "shrink" looks too grey...

[dellsam34]

I would use bypass if you intend to do post processing, With shrink the levels are baked in and whatever lost cannot be recovered. This is why I told you stay away from Vdub and AMTV for HDMI ingest, too many things to go wrong if you don't know what you're doing, Most of those settings are to control the procamp in the ADC of the capture card, your signal is already digital.

[ozymango]

So it first converts analog input to digital, does its filtering in digital format, then converts back to analog for the s-video output.

I stand corrected. That being said, do you agree that the HDMI capture route gives better results than S-video capture? Though in my own case this isn't an option, the DMR-ES15 doesn't have HDMI output, only S-video (and composite). I'm still quite happy with the results as-is, anyway!

[Sharc]

I'm wondering if "bypass" output colour range is clipping 100% white? Then again, "shrink" looks too grey...

Look at the luma in the histogram of the picture in post#33: no clipping is visible. And with the luma staying within the 16...235 range you are reasonably safe. Even when using an expanding matrix like the standard Rec.601 for decoding, the YUV(235,128,128) reference white will be mapped to RGB (255,255,255) which is perfectly 100% white, no clipping. Similarly, the YUV(16,128,128) reference black will be mapped to RGB(0,0,0) which is perfectly black. No clipping.
Even less of a problem for TVs and editors using "studio" mapping as per ITU specs.

If you "shrink" you risk a low contrast greyish picture. If you "expand" you risk clipping/crushing. So unless you have a good reason to change it leave it at "bypass" and make any levels adjustments - if necessary - in post. Perhaps in Avisynth for your cat clip.avi some minor shifts like

tweak(hue=0.0,sat=0.9,bright=6,cont=1.0,coring=false)

or:

levels(0,1.0,227,6,235,coring=false,dither=true)

[Attachment 91037 - Click to enlarge]

[robertzombie]

So it first converts analog input to digital, does its filtering in digital format, then converts back to analog for the s-video output.

I stand corrected. That being said, do you agree that the HDMI capture route gives better results than S-video capture? Though in my own case this isn't an option, the DMR-ES15 doesn't have HDMI output, only S-video (and composite). I'm still quite happy with the results as-is, anyway!

I was also happy with my ES15 results, I was just like the neatness of keeping the signal digital once it's been converted. I could do a few test captures pitting HDMI against s-video (the Startech supports both). I can run some static signals through (my generator will do colour bars, greyscale, and crosshatch - would they be useful?) and see if there are any noticeable differences.

[robertzombie]

I'm wondering if "bypass" output colour range is clipping 100% white? Then again, "shrink" looks too grey...

Look at the luma in the histogram of the picture in post#33: no clipping is visible. And with the luma staying within the 16...235 range you are safe. Even when using an expanding matrix like the standard Rec.601 for decoding, the YUV(235,128,128) reference white will be mapped to RGB (255,255,255) which is perfectly 100% white, no clipping or crushing. Similarly, the YUV(16,128,128) reference black will be mapped to RGB(0,0,0) which is perfectly black. No clipping or crushing.
Even less of a problem for TVs and editors using "studio" mapping as per ITU specs.

If you "shrink" you risk a low contrast greyish picture. If you "expand" you risk clipping/crushing. So unless you have a good reason leave it at "bypass".

Great, thank you for explaining that. I will leave it at "bypass"!

[ozymango]

I could do a few test captures pitting HDMI against s-video (the Startech supports both). I can run some static signals through (my generator will do colour bars, greyscale, and crosshatch - would they be useful?) and see if there are any noticeable differences.

That would be interesting! Also you may have already said this, but what software are you using to capture HDMI? I see there's back and forth on whether VirtualDub and AmeRecTV are the best program for capturing HDMI, but if those aren't optimal, what *is* optimal? I may be giving away my own visual limitations in that -- for my own vintage home movie tapes, which weren't all that great to begin with -- even with a cheap $10 capture dongle, the results still look okay to me when played on a TV and I sit on the couch across the room. Mostly what I notice is dang, I sure was young back then!

[rgr]

So it first converts analog input to digital, does its filtering in digital format, then converts back to analog for the s-video output.

I stand corrected. That being said, do you agree that the HDMI capture route gives better results than S-video capture? Though in my own case this isn't an option, the DMR-ES15 doesn't have HDMI output, only S-video (and composite). I'm still quite happy with the results as-is, anyway!

I couldn't capture video from the HDMI input so that the frames matched 1:1. There were always some drops, but I also didn't have a super good device (only Avermedia, Ugreen dongle and poor $5 chinese dongle).

[rgr]

VirtualDub and AmeRecTV are the best program for capturing HDMI, but if those aren't optimal, what *is* optimal?

I had no problems with capturing HDMI through the Ugreen dongle, aside from adjusting the parameters. It worked the same way.