digitalFAQ.com Forum

digitalFAQ.com Forum (https://www.digitalfaq.com/forum/)
-   Capture, Record, Transfer (https://www.digitalfaq.com/forum/video-capture/)
-   -   Hi8 with DCR-TRV230 - capture DV vs. HuffyYUV? (https://www.digitalfaq.com/forum/video-capture/10330-hi8-dcr-trv230.html)

nullb 02-04-2020 10:28 AM

Hi8 with DCR-TRV230 - capture DV vs. HuffyYUV?
 
I was deciding if I should re-record all my DV firewire recordings with HuffyYUV ones.

Screenshots are done with AvPMod + FFVideoSource + QTGMC(Very slow)

HuffyYUV:
https://i.imgur.com/YSjPVdu.png
vs
DV:
https://i.imgur.com/seFtsZT.png

Looks like the DV recordings aside from not being 4:4:2 and DV is a lossy codec.
Wipes away a ton of detail, either by DNR or the codec limitations, I wasn't expecting that much of an improvement.

If you look at the trees and the building a ton of detail is very clearly lost in the DV, it's more obvious on faces but didn't want post any photos of family.

Also looks like the DV output does some stretching for some reason, hopefully somebody else find this useful. Going HuffyYUV is worth the extra effort.
You do oddly seem to loose a few extra lines at the bottom and sides with a capture though, not all that sure why, if the recordings frame to frame match you can always attach those to the HuffyYUV in post, but probably not worth the effort.


Sony cameras all have the same TBC, so it should be a fairly universal comparison.
HuffyYUV was done with S-Video out + AVT-8710 + ATI 600 USB Clone

hodgey 02-04-2020 12:55 PM

When doing a DV example, did you adjust the levels when making a screenshot from the DV file? I don't remember exactly when it comes to analog -> DV conversion in the cameras, but miniDV and D8 frequently has luma higher than 235. Not really clear to me if DV is supposed to be full-range YUV, or normal 16-235.

The S-Video sample is a little sharper, with more noise, maybe the camera applies some low pass or noise filtering to avoid blocking artifacts. That said I would suggest doing a comparison with a raw frame rather than deinterlaced with QTGMC as it does some blending between frames. Alternatively Bob or nnedi3.

The cropping of top/bottom is a bit odd, not something I've seen in any PAL Hi8 camera at least. Don't have an operational analog-capable Sony D8 at the moment but I don't remember the image being cropped on the top/bottom on the one I used a bit in the past either (a DCR-TRV330). The TBC and capture card may crop a few lines on the left/right but they normally shouldn't crop top/bottom? My TBC-3000 is a bit off-center and crops a bit much on the right, black AVT8710 much less so, but neither crop top/bottom. I think some "pro" VCRs have a head switch masking function built in to the TBC that blanks out the head switch area, but I don't think that's something you would find in a consumer camera.

latreche34 02-04-2020 01:36 PM

I don't believe you would gain anything from the analog conversion of a DV footage compared to the raw DV file other than a fake sharpening with added noise, There are a lot of tools out there to do correction in the digital domain using software not an ancient DAC in a camcorder.

traal 02-04-2020 02:12 PM

Quote:

Originally Posted by latreche34 (Post 66406)
I don't believe you would gain anything from the analog conversion of a DV footage...

I was also confused at first but it's actually analog footage (Hi-8) and the camcorder can output DV, composite video, and S-Video. So the question is whether to use the DV output or S-Video.

The DV screenshot is surprisingly bad. I would have expected it to look almost identical to the HuffYUV screenshot.

sanlyn 02-04-2020 06:04 PM

[QUOTE=traal;66408So the question is whether to use the DV output or S-Video./QUOTE]
Use analog s-video. Why do you want to damage already crummy VHS with nasty ugly compression-riddled DV and its lousy baked color, data loss, and blown out levels? And you lose data that doesn't come back (that's what loossy means).

It's really tough to tell anything from still images. You see, still images don't move. If they don't move, you can't see noise levels or patterns. Perhps you don't know how to make, analyze, or post video instead of screen caps? You do realize that you lost more data by converting those images to RGB.


When your offline png images disappear, your thread will pretty much be useless anyway for later readers.

nullb 02-04-2020 07:24 PM

1 Attachment(s)
Quote:

Originally Posted by hodgey (Post 66403)
When doing a DV example, did you adjust the levels when making a screenshot from the DV file?

No that's as is, I didn't want to alter it.

Quote:

Originally Posted by hodgey (Post 66403)
The S-Video sample is a little sharper, with more noise, maybe the camera applies some low pass or noise filtering to avoid blocking artifacts.

I was thinking the same when it does a DV conversion, so it compresses better.

Quote:

Originally Posted by hodgey (Post 66403)
QTGMC as it does some blending between frames

I don't think that's true, but I'll look at it again.

Quote:

Originally Posted by sanlyn (Post 66416)
It's really tough to tell anything from still images. You see, still images don't move. If they don't move, you can't see noise levels or patterns. Perhps you don't know how to make, analyze, or post video instead of screen caps? You do realize that you lost more data by converting those images to RGB.

Yes I realize all of that thank you, it was just a quick comparison after I clean the heads of the camcorder and convert everything I will probably do some more.
It very clearly gets the point across, there is a large noticeable improvement in detail.

Quote:

Originally Posted by sanlyn (Post 66416)
When your offline png images disappear, your thread will pretty much be useless anyway for later readers.

I really doubt imgur is going anywhere, but I've attached the bmp files zipped to this post.

msgohan 02-04-2020 08:22 PM

Please do attach short Direct Stream Copy video samples. I'm interested. :)

There are halos on the edges of the S-Video capture not present on the DV. The default Proc Amp settings of the ATI 600 USB add sharpening. I can't speak for the "clone" you're using, but if it uses this same default, you should lower the Sharpness setting down to 0 to reduce this artifact.

The right edge being slightly masked with Digital8 S-Video output of analog tapes is something I noticed with my NTSC cam as well. [EDIT: It's actually the left edge, on my caps.] It happens even with TBC & DNR disabled, because the cam actually does A-D-A regardless.

Quote:

Originally Posted by sanlyn (Post 66416)
You do realize that you lost more data by converting those images to RGB.

Quote:

Originally Posted by nullb (Post 66417)
It very clearly gets the point across, there is a large noticeable improvement in detail.

Sanlyn made an important point here that I think you're missing. The side of the building hits RGB 255,255,255 (i.e. full, clipped white) on the DV screenshot. This is arguably a misrepresentation of the DV; it appears to be missing more detail than it really is. As alluded to by hodgey, the DV likely has Y > 235 that you didn't recover before making the screenshot.

Quote:

I really doubt imgur is going anywhere
Yeah, I would've thought the same thing about ImageShack last decade, before they abruptly announced that they were deleting all images posted on free accounts. Here is one random old forum post about it, if you never went through that.

Here is my own comparison (NTSC): https://forum.videohelp.com/threads/...uto-load%21%5D

nullb 02-04-2020 08:34 PM

Quote:

Originally Posted by msgohan (Post 66418)
Please do attach short Direct Stream Copy video samples. I'm interested. :)

I will once I do some full tapes and find a good spot

Quote:

Originally Posted by msgohan (Post 66418)
There are halos on the edges of the S-Video capture not present on the DV. The default Proc Amp settings of the ATI 600 USB add sharpening. I can't speak for the "clone" you're using, but if it uses this same default, you should lower the Sharpness setting down to 0 to reduce this artifact.

I did notice that and I've turned Sharpening off. Didn't think it had that much of an effect at such a low value. It defaults to 2.

Quote:

Originally Posted by msgohan (Post 66418)
The right edge being slightly masked with Digital8 S-Video output of analog tapes is something I noticed with my NTSC cam as well. It happens even with TBC & DNR disabled, because the cam actually does A-D-A regardless.

Interesting, at least I know there's no magic setting to get it back now.

Quote:

Originally Posted by msgohan (Post 66418)
Sanlyn made an important point here that I think you're missing. The side of the building hits RGB 255,255,255 (i.e. full, clipped white) on the DV screenshot. This is arguably a misrepresentation of the DV; it appears to be missing more detail than it really is. As alluded to by hodgey, the DV likely has Y > 235 that you didn't recover before making the screenshot.

I am aware of all of that but I just wanted to post a quick comparison, I haven't actually done a full tape yet I just quickly checked. The whole thing is a misrepresentation, it's RGB, then imgur has compressed it with PNG.
But it's enough to quickly illustrate a point, without having to download and playback two files.

Quote:

Originally Posted by msgohan (Post 66418)
Yeah, I would've thought the same thing about ImageShack last decade

That's a fair point.

nullb 02-05-2020 02:33 AM

I stumbled onto this post https://forum.videohelp.com/threads/...es#post1391493

Seems like the TBC could be destructive in these cameras but everybody here recommends using the in-built TBC.

Might be another case to compare, unfortunately those decks are way too expensive or worn down so I wont be testing that as well.

latreche34 02-05-2020 02:46 AM

The built in camera TBC is just a line TBC, unless you know what to look for you wouldn't notice any difference, But if you look closely where there are vertical lines such as door frames, light poles or similar objects you will notice the difference between TBC on and off, I will not give you a hint I'll let you find out for yourself.

nullb 02-05-2020 02:49 AM

Quote:

Originally Posted by latreche34 (Post 66443)
The built in camera TBC is just a line TBC, unless you know what to look for you wouldn't notice any difference, But if you look closely where there are vertical lines such as door frames, light poles or similar objects you will notice the difference between TBC on and off, I will not give you a hint I'll let you find out for yourself.

Interesting, I will look into it more but to that point what do you recommend. Having it on or off? I use an external AVT-8710 as well mind you.

latreche34 02-05-2020 02:58 AM

It's up to you to make that decision after you see the difference.
The AVT-8710 is a frame TBC it fixes the frames timing signal it has no bearing on the video signal itself although it has a Proc amp so I guess you could say it changes the video parameters, Unfortunately only the green cypress box known to work properly, if you have the black box you may as well take it out the workflow, Also its proc amp default parameters are wrong and you will have to set them manually every time you disconnect the power cord.

nullb 02-05-2020 03:09 AM

Quote:

Originally Posted by latreche34 (Post 66445)
It's up to you to make that decision after you see the difference.
The AVT-8710 is a frame TBC it fixes the frames timing signal it has no bearing on the video signal itself although it has a Proc amp so I guess you could say it changes the video parameters, Unfortunately only the green cypress box known to work properly, if you have the black box you may as well take it out the workflow, Also its proc amp default parameters are wrong and you will have to set them manually every time you disconnect the power cord.

Yeah aware of all that I got the AVT-8710 from LS so all good there. Proc Amp I've been playing with using LS's recommended settings all good there.

Fair enough just curious to hear what other people think as well.

nullb 02-05-2020 09:41 PM

Without the in-line TBC the color pops out a bit more and in some shots it looks better, a little more detail retained.
But without it, I saw frames dropped at random, due to that alone I'd keep the TBC on.

Also with the TBC enabled, on average more detail is retained, and prevents the color from getting completely out of control like below.

https://i.imgur.com/jmbyGyf.png

Ideally I guess you could record with it both on and off and cherry pick frame regions to create a hybrid of the two during post. But that's more work than it's worth for me.

Conclusions: Keep the in-line TBC on, turn DNR off it does nothing and may wipe away detail if it does in-fact do anything. HuffyYUV > DV but you will loose some lines.


All times are GMT -5. The time now is 04:21 AM

Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.