HDV to DV vs. HDV to YUV
 

Maybe this seems a silly question.

I always shoot on the Z1 in HDV.
My NLE at the moment is only SD.
I can import DV with DV25 codec
or YUV with uncompressed codec.

Since I record in HDV I always import in YUV.
Seems to me that this should give better picture than importing in DV.

Of course, if I record in DV or DVCAM, importing in YUV does not improve anything.

Do you agree?

Camafe

Depends. What program do you use to compress to YUV? what process do you employ for scaling, etc?It could do more damage than getting the Z1 to do a very good DV downconvert for you.

Converting DV to YUV can have advantages as you're moving from a 4:1:1 or a 4:2:0 colourspace to a 4:2:2 colourspace.

I'm not converting DV to YUV.
I'm converting HDV (1440x1080) to SD YUV (720x576)

At my eyes HDV to SD YUV is better than HDV to SD DV.
Convertion made inside the Z1.

I capture with a TARGA 3K with DVcodec for DV
or YUV uncompressed codec for YUV.

Try this
 

Capture HDV to your hard drive, then resize to uncompressed 720 by whatever, using VirtualDub set to Lanczos3 filter interlaced. To my eyes it's an incredibly good picture.

If you don't mind, please come back and report on how this compares to your method.

Well it depends what you're seeing. I'm not familiar with the results of the Targa card, but if you're happy with the results then stick with them.
I'm not really sure what the question is.

If it's uncompressed YUV, file sizes & data rate must be huge. If it can handle uncompressed YUV are you sure it can't handle HD?

I know you're not converting DV to YUV, I'm just stating that there ARE occasions where converting DV to YUV can be advantageous. But if your source is HDV in MOST cases, HDV to YUV would be better than DV to YUV

Ervin, your result would achieve very good results, V'Dub is a great tool, but the capture would not be realtime of course, and you'd lose timecode if batch capture was important to you.

I'd agree you can get good results with downsampling HDV to a Standard Definition format BETTER than DV:

- Ingest HDV (I prefer Apple Intermediate Codec)
- Edit HDV (maybe AIC, or CineForm or something - or what your card can do)
- Deinterlace the HDV
- Export to a 4:2:2 codec such as DVCPro-50

(actually, the last two are one step in FCP)

That way, the 4:2:0 effect of HDV and DV can be reduced. 4:2:0 really mucks up things like threatrical coloured lighting and computer imagery with strong colours.

I've done some tests shooting HDV and DV side by side, then ingesting the HDV as DV and cutting it with the DV shots. I did this by switching mode on my Z1 from HDV to DV and back again on the same tape, recording high motion, high detail, tracking shots etc.

The HDV footage downconverted to DV exhibited some softening in detail through high motion, but this is to be expected. And I found it quite pleasing.

In-camera downconversion from HDV to DV is a lot better than some would give credit to. Charts show a minor but notable drop in quality, but images of real life subjects don't - I've run tests with jumbling up HDV clips downconverted in-camera to DV with the same clips imported as HDV then downsampled to DV. The only way you notice is the 1 scanline difference, and that was because I was watching on an underscan monitor.

So my bottom line is that I'll use my Z1 in a HDV-to-SD workflow for green screen shoots where the 4:2:2 effect from downscaling really helps, and for deinterlacing, where the 25% drop in resolution is hidden by the scaling down.

Progressive scan is very important for me as most of my work is either for the web or for playback in PowerPoint as a WMV. Interlacing makes compression inefficient - all those comb like edges drive the compression engine nuts - and no computer screen or data projector I know of uses interlacing. So best do without.

Thanks. I will report.

Anyway I have some news that are surprising to me also.

For the moment I have to edit on this machine:
Windows 2000 Pro SP2
Premiere 6.5
Targa 3000 with YUV in

Because of the Targa I can't go to WinXP or Win2K SP4.
The Targa has a FireWire port but never worked properly.

When I capture YUV from Betacam SP or Digital I have a TERRIFIC picture.
Capturing in YUV means around 72G /hour...

Recently I installed a card with FireWire and USB2 for external drives.
I've tried to connect the Sony Z1 to the PC via FW.

I always shoot HDV on the camera.
And the results are:
Z1 YUV out --> Targa YUV in / YUV codec ----- good image
Z1 YUV out --> Targa YUV in / DV codec--------not so good
Z1 DVout --> PC in captured WITHOUT the TARGA hardware---- better than the other 2 options.

The difference is nost noticeable on high details.
Quite a surprise for me.
Perhaps that in this way I have only 1 downconvertion from HDV to DV inside the camera. Than the signal is recorded on disk directly.
Maybe the 2 steps HDV / YUV and YUV to DV made by the targa is to much for the signal.

I'm starting to capture around 3 hours of material and I will do it this way.
The camera downconverts from the HDV on tape to DV on the FW port, and I capture the DV on Premiere WITHOUT the use of the targa.
The avi DV files are then played realtime with "good" quality by the Targa hardware.

I'm dreaming of my next MATROX AXIO system.

I will do some tests capturing the HDV signal on a Premiere CS2 to a HDV timeline and then export to PAL and see the diference.

One of this days I have to forget the TARGA forever...!

Thanks for all the answers
Carlos

Actually my own workflow is this: capture and edit natively HDV using PremPro 2; output HDV, resize with VDub and frame serve uncompressed to CinemaCraft for DVD encoding - so there is no loss of timecode.

Those results are no surprize to me. Targa might have been a good format back in it's days, but both software and hardware have evolved by leaps and bounds in recent years.