HV20 HDV to DV conversion quality test
 

For some like myself editing HD really isn't feasable so what I have been doing is simply setting the HV20's output to DV locked and capturing my HDV material to the computer as DV. That made me question the quality of the HV20's scaler so I did some tests.

First of all I captured a DV stream using the HV20's built-in scaler. The quality is pretty good but you can get better quality by capturing the HDV stream and resize it with high quality software scalers. I used Digital Anarchy's Resizer as well as an HDV artifacts filter to get the absolute best results. Using a specialized scaler plug-in will give you slightly better results than After Effect's built in scaler. I'm not sure you'd ever see it though unless you were looking at the video with a magnifying glass or if you were using the video for some pretty serious FX work.

My test is in native square pixels so it may appears stretched. I also show the blue channel only since that showed the clearest scaling artifacts. Taking a full HDV file into QuickTime Player and exporting it to DV yielded less quality than the HV20's scaler/encoder which surprised me.

Shot it at...
Cine mode
24p
1/48 shutter
F5.6
Contrast -1

100% (hard to see differences)
http://farm2.static.flickr.com/1315/...effd5365_o.jpg

800%...
http://farm2.static.flickr.com/1287/...c7f943ab_o.jpg

The footage was captured to tape as HDV and then...
Sample box 1 - Captured to computer as DV using the HV20's built-in scaler
Sample box 2 through 6 were captured to the computer as an HDV stream, I then exported that from MPEGStreamClip as uncompressed TIFFs and exported the TIFF sequence from there.

My conclusion is unless you want super high quality footage for high quality DVD or you are using the shot for FX work at DV resolution then the HV20's scaler/converter is good enough.

Thanks for that comparison Wes,
To me the photoshop one on on the upper right appeared the best.

Just to add to this, I recently compared the down conversion (of A1 footage) done by the HV20 with the downconversion done by the XHA1 and they are identical.

To my eye the Photoshop scaling is adding more of a sharpening effect also and that helps the video look sharper.

There's no color? One of the biggest differences is that dv codec reduces color. If you capture in hdv and then scale down and don't encode to dv you will get more color information.

There is color, but I simply sampled the Blue channel for this since it showed the most detail in scaling. The only DV compression is from the HV20 and the QuickTime player. The others are simply scaled from the .MT2 file.

i thought there were all very close ... and from past testing i've done i would call them all a draw... try taking a look at 5-6 frames - you might find frame 1 might look better in photoshop and frame 2 might look better in AE - frame 3 might look better in QT etc ...

This reminds me of how bad Quicktime's DV codec is for encoding ;) Thanks for the test.

Don, you are right about changes along a sequence of frames. I analyzed each clip plus there was no motion in my video so I'm confident these frames give a good idea of the actual difference in scaling results.

Euisung, tell me about it! I remember way back when QuickTime added DV and how horrible it was, then Apple updated it and it actually was decent but it's interesting that it's 2007 and it's still pretty bad.

HDV artifacts filter
 

Hi,

I would like to know wich HDV artifacts filter did you use.

Thanks

Just wondering if there is an equivalent HD to DV resizing process in Vegas?

Many thanks ~ John