Can anyone tell me how footage compares between a digibeta (sony DVW970P / Sony DVW-709WSP etc) and a EX1?

I like the 2/3 chip of the digibeta , combined with their low compression,

but they don't have under/overcranking abilities for fast/slow motion

Anyone?

Btw, downconverted to SD the ex1 gives 4:4:4.

Can anyone tell me how footage compares between a digibeta (sony DVW970P / Sony DVW-709WSP etc) and a EX1?

I like the 2/3 chip of the digibeta , combined with their low compression,

but they don't have under/overcranking abilities for fast/slow motion

Anyone?

I've an editor friend who was involved in testing the DVW970 compared to the Varicam and HDW 750 for a BBC drama. For a SD production, the DVW 970 gave the nicest looking pictures, he wasn't really that impressed by either the Varicam or the HDW 750.

Btw, downconverted to SD the ex1 gives 4:4:4.

Hi Dominik, can you explain further....

thanks

The ex1 records 960x540 chroma-pixels at 1080p, which is more than 720x480 luma-pixels of NTSC-SD. So after resampling to 720x480 you have unique chroma-information for each luma-pixel. And this is what 4:4:4 means.

The only HD format that turns into 4:4:4 SD is 4:4:4 HD. You do not benefit from the larger pixel count when you scale your footage.

The only HD format that turns into 4:4:4 SD is 4:4:4 HD. You do not benefit from the larger pixel count when you scale your footage.

If you start with more distinct samples per channel than you get in 4:4:4 SD how could you not end up with the equivalent of 4:4:4 SD when scaled assuming you go to an uncompressed SD format? Certainly not all applications scale equally, but if you're using software with a decent scaling algorithm I would be surprised if you didn't end up with better results than native 4:4:4 SD due to oversampling.

The only HD format that turns into 4:4:4 SD is 4:4:4 HD. You do not benefit from the larger pixel count when you scale your footage.
Wrong.
Here's an example of scaling ex1-footage down to sd:
http://www.dominik.ws/ex1tosd.png
The first row is dv25, the second dv50 and the third no further compression.
The first column is rgb, the second is the a-chroma-channel and the third the b-chroma-channel (I used Photoshops Lab-Color-Mode and increased the contrast for better visibility).

It's a nice example. But it doesn't match what I see in my scopes, so I will stick to my statement for a while.

If I export a 4:2:0 HD (XDCAM HD) source to a 4:4:4 SD (AJA RGB) source I don't gain any chroma information when I review the footage in color's scopes. So either I am missing a step here or... We are talking about two different issues.

Please enlighten me.

It's a nice example. But it doesn't match what I see in my scopes, so I will stick to my statement for a while.

If I export a 4:2:0 HD (XDCAM HD) source to a 4:4:4 SD (AJA RGB) source I don't gain any chroma information when I review the footage in color's scopes. So either I am missing a step here or... We are talking about two different issues.

Please enlighten me.

If you dump 3/4 of the chroma information when recoding it, there's no way you're going get it back. You're just downscaling the information that's there and massaging it, but you're still missing the full information required for true 4:4:4.

My point exactly.

But if there was an easy way to keep the extra information in the chroma channel, I would love to hear about it.

And to get back on topic. Down converting or up converting are in their very nature destructive and will have a negative impact on image quality. So no, the EX1 does not produce better SD images, than for instance a sony DVW970P. It is widely accepted that if your final master is SD and the footage will never be used in HD, then shoot in SD.

But if you are caught in the middle, like the rest of us, then HD is the only way and there the EX1 really shines as a price performer. As long as you are aware of its shortcomings.

I have tested all of the scalers/format converters on the market and none of them really delivers. The best hybrid is to master to HDCAM SR or D5 and use their internal scaler boards to create a digibeta dub. It is the most consistent result with very few nasty surprises.

If you dump 3/4 of the chroma information when recoding it, there's no way you're going get it back. You're just downscaling the information that's there and massaging it, but you're still missing the full information required for true 4:4:4.
The ex1 throws away 3/4 of the chroma-information captured by its sensors. But even then it's still more than 4:4:4-SD can hold, so for what do I need that thrown-away-information (assuming that the ex1 does proper filtering before throwing-away to avoid aliasing in the chroma-channels)?!
Down converting or up converting are in their very nature destructive and will have a negative impact on image quality. So no, the EX1 does not produce better SD images, than for instance a sony DVW970P.
Downconverting gives very smooth results, because if well done you can approximate the mathematical optimum given by the Nyquist–Shannon-sampling-theorem even closer, than physical sensor-chips with their sub-optimal layout concerning that theorem (there are no optical sinc-lowpass-filters) can do. Also the lowpass-filtering needed before downsampling eliminates noise. So downconverting if well done is very constructive concerning the final resolution.

The example I posted shows pin-sharp chroma-channels. What do want more (ok, the mpeg2-compression has some impact, but that's a different story)?!

It's a nice example. But it doesn't match what I see in my scopes, so I will stick to my statement for a while.

If I export a 4:2:0 HD (XDCAM HD) source to a 4:4:4 SD (AJA RGB) source I don't gain any chroma information when I review the footage in color's scopes. So either I am missing a step here or... We are talking about two different issues.

Please enlighten me.

You're asking the wrong question - it's not 'do I gain any chroma information when down scaled?' because of course once data is gone it's gone. The correct question is 'do I end up with more chroma information in my downscaled SD version than I would have in an SD source recorded at 4:2:2, 4:1:1, etc?'. Or you could phrase it 'do I end up with the same amount of chroma info when I downscale XDCAM to SD as when I start with an SD 4:4:4 source?'

Dominik's sample answers this question very clearly and is probably the best demonstration of it that I've seen. To test it with the scopes in color as you're doing Nils what you need to do is export an XDCAM source to both uncompressed 4:4:4 SD and an uncompressed 4:2:2 SD to represent what you get from digibeta (or try shooting the same subject on digibeta for the most accurate comparison). Compare those and see if you don't see an improvement in chroma information in the 4:4:4 SD export vs. the 4:2:2 SD digibeta source.

You're asking the wrong question - it's not 'do I gain any chroma information when down scaled?' because of course once data is gone it's gone. The correct question is 'do I end up with more chroma information in my downscaled SD version than I would have in an SD source recorded at 4:2:2, 4:1:1, etc?'. Or you could phrase it 'do I end up with the same amount of chroma info when I downscale XDCAM to SD as when I start with an SD 4:4:4 source?'

Dominik's sample answers this question very clearly and is probably the best demonstration of it that I've seen. To test it with the scopes in color as you're doing Nils what you need to do is export an XDCAM source to both uncompressed 4:4:4 SD and an uncompressed 4:2:2 SD to represent what you get from digibeta (or try shooting the same subject on digibeta for the most accurate comparison). Compare those and see if you don't see an improvement in chroma information in the 4:4:4 SD export vs. the 4:2:2 SD digibeta source.

Digibeta is 10 bit compared to 8 bit XDCAM HD, plus the compression differences, so you really need to compare the moving images and see how well they handle grading and manipulation in post before they break down.

...
And to get back on topic. Down converting or up converting are in their very nature destructive and will have a negative impact on image quality. So no, the EX1 does not produce better SD images, than for instance a sony DVW970P. It is widely accepted that if your final master is SD and the footage will never be used in HD, then shoot in SD.
...

Something you are not considering: the DVW-970 actually is a high definition camera, because its CCDs have 980x1164 effective pixels (in the PAL version, NTSC version has 980x988 effective pixels). So the camera's DSPs must downconvert the high resolution signal from the CCDs to standard definition before recording in Digital Betacam. And that's just why the images from a 970 are so sweet: because they are oversampled, the number of lines of the CCD is double than the TV standard they record on, so they produce a much better image than a camera which has the same resolution in it's imagers than its TV standard.

So don't came and tell me that downconverting is not good, because that's just what the best SD cameras available at the moment are doing to perform the best quality. The choice of doing it before or after recording is more depending of personal preferences than of technical evidence.