SD vs down converted HD

[Robert M Wright]

I just did it again, using Panasonic's VFW DV codec for encoding the output file (literally converting HDV to DV). Speed was a little over 11fps.

[Pete Cofrancesco]

Originally Posted by Robert M Wright

I'm confused about the source of these images. Can you tell us exactly how (the process) each were generated?

We've got three possibilities:

1 - Shot as DV and captured as DV
2 - Shot as HDV and captured as DV (camera converting on the fly)
3 - Shot as HDV, captured as HDV and converted to SD by an NLE (or other software, on a computer)

Those images are also 853x480, so I assume they have been re-sized from 720x480, and also compressed (via JPEG). That does alter the image, but as long as the methodology is the same, and of high enough quality, for both images being compared, should still make for close enough to an apple-apple comparison (less than very high quality, in resizing to 853x480 or in JPEG compression, could potentially narrow the differences between the images though).

One was shot in SD anamorphic and the other HD and down converted from the camera during import. The screen captures were done in MPEGstream Clip. I'm not really sure why they aren't 720x480 because I didn't resize them. Btw, I'd like to see stills comparing other people's down convert so I can see for myself any quality improvement over SD.
I don't use VitualDub because I'm on a Mac. If I remember correctly I once did a test down converting one frame in Photoshop and it really did a nice job but you'd have to be crazy to use it to downconvert a movie.

[Shaun Roemich]

Originally Posted by Javier Gallen

you can get better color resolution by downconverting. Just like having a 4:2:2 SD camera.

Slight correction: the colour is noticeably less washed out on the Z1 downconverted from HDV compared to DV but HDV colour space is 4:2:0, NOT 4:2:2. This means that each "block" of colour information is 2 pixels by 2 pixels square, as opposed to DV which is 4 contiguous pixels on one line getting 1 unit of colour info.

4:2:2 colour has 2 pixels in a row sharing the same colour info, all on one row.

EDIT: Of course, the above assumes NTSC for DV. PAL DV uses 4:2:0 colour space, just like HDV does.

[Robert M Wright]

I think what he was trying to say is, even at 4:2:0, with HDV you have more chroma samples per frame than with DV. When you downsize the HDV to SD, you effectively get higher chroma resolution than DV records (somewhat similar to effectively getting 4:2:2 in SD - actually closer to 4:4:4, I think, but I haven't had my morning coffee yet so I'm still groggy).

[Shaun Roemich]

You don't get MORE samples per frame. You get a different PATTERN: a 2 pixel by 2 pixel square instead of a 4 pixel x 1 pixel line.

[Shaun Roemich]

Chroma subsampling - Wikipedia, the free encyclopedia

A great graphical representation of 4:2:0, 4:1:1, 4:2:2 and 4:4:4 can be found at the above address.

The number of SAMPLES of colour is identical in 4:1:1 and 4:2:0. The distribution is what makes the difference. And yes, 4:2:0 does have PERCEPTUALLY better colour representation (or at least less washed out) than 4:1:1, in my opinion.

[Robert M Wright]

Originally Posted by Shaun Roemich

You don't get MORE samples per frame. You get a different PATTERN: a 2 pixel by 2 pixel square instead of a 4 pixel x 1 pixel line.

When you shoot HDV, you get both more luminance samples and chroma samples per frame, than with DV. It's higher resolution video in both regards. With 1080i HDV, you get 1440x1080 luminance and 720x540 chroma resolution. With DV you get 720x480 luminance and 180x480 chroma. In theory, you can effectively wind up with 4:4:4 color sampling after downsizing 1080i HDV to SD.

[Shaun Roemich]

Robert: INTERESTING... I'd never thought of it like that before. I'm sure that "in theory" comes into play somewhere. But of course, in downresing, you'd need to go to an intermediate codec that would support the increased colour sampling. Going to DV (4:1:1) would "throw away" all your gains by re-sampling to the 4 pixels by 1 pixel matrix.

[Robert M Wright]

Part of why I say "in theory" is that I think (but I don't know absolutely for sure) the 2x2 block (in 4:2:0 interlaced footage) is by field (rather than by frame). It almost has to be. I'm not sure (hurts my brain to think hard enough to figure it out), but I don't think downsizing 1080i HDV (to interlaced SD properly) will really yield 4:4:4 color. It is safe to say though, that downsizing HDV (of any flavor) to SD should certainly yield better than 4:2:0 or 4:1:1 color. Yes, if you then "recorded" the downsized footage (saved the result) as DV (for example), it would then be 4:1:1 color (essentially, that is downsizing the chroma portion of the image yet again).

[Robert M Wright]

Another part of why I say "in theory" is in recognition that the imaging system of the camera is a vital part of determining effective resolution. Just because a camera records an image in a particular format, doesn't mean the camera actually acquires at the resolution of that recording format (as a practical matter, the effective resolution never reaches 100% of the potential of the recording format - in the neighborhood of 70-80% is fairly typical).

[Tom Hardwick]

Originally Posted by Peter Manojlovic

Proper downconverting is a tricky little devil..Especially for 1080i sources. And it's as slow as molasses. I'm coming to the realization that one should shoot for the intended output...

I reckon Peter's nailed it. Don't forget that the Z1 (and any HDV camera for that matter) always shoots HDV. If you film in the SD mode the downconversion is taking place in real time, from chips to tape.

If you downconvert from an HDV tape so that DV exits via Firewire, the (presumably the same, and pretty cheap 'n' cheerful) downconverters are working to effect this.

But shooting for the intended output is very good advice, and if you're only ever going to make DVDs (by definition SD) then it's far safer to film in the SD mode. This is because dropouts are far less damaging (and individual frames can even be worked on in Photoshop) and of course the audio is uncompressed.

I too have difficulty seeing any real benefit in downconverting after I've done the HDV edit. There may be minute differences but hey - one HDV dropout is FAR more damaging from a visual pov.

tom.

[Javier Gallen]

Originally Posted by Robert M Wright

When you shoot HDV, you get both more luminance samples and chroma samples per frame, than with DV. It's higher resolution video in both regards. With 1080i HDV, you get 1440x1080 luminance and 720x540 chroma resolution. With DV you get 720x480 luminance and 180x480 chroma. In theory, you can effectively wind up with 4:4:4 color sampling after downsizing 1080i HDV to SD.

That's exactly what I was talking about. Thanks.

My advice is to shoot always in HDV, edit in HDV and let the final step do the downconversion. This allows you to mantain the maximun quality of your original footage even with color correction, fx, and preview renders (second generations). Also, you can reframe when you needed up to a certain degree without loosing quality in your final SD master.

[Jason Selmes]

peter,

is the SD you shot showing up s 16:9?? I was wondering this myself because im buying a FX1 to shoot weddings with and want to know if its still widescreen 16:9 SD????? or anamorphic

thanks

Jason

[Tom Hardwick]

The FX1 has 16:9 CCDs, so everything you shoot will be widescreen unless you pillarbox it for 4:3 in the menu.

[Boyd Ostroff]

Originally Posted by Jason Selmes

im buying a FX1 to shoot weddings with and want to know if its still widescreen 16:9 SD????? or anamorphic

Maybe we're just getting confused with terminology, but all standard definition widescreen 16:9 is anamorphic. That is the only standard definition widescreen format. It is recorded at 720x480 (or 720x576 PAL), then it is stretched to fill the 16:9 monitor when you play it back. The Z1 conforms to this standard when you set it for DV (or use the camera to downconvert from HDV).