Native 1080p (DVCPRO HD)?

"rbilsbor", Your question about using 720p chips and then do pixel shifting to get 1080p is an completely eligible one, a Canon rep. should answer to. And JVC should answer, too.
Are progressive recording 1280 x 1080 pixel CCDs so much more expensive than 960 x 720 ones? And why should progressive recording CCDs be so much more expensive than interlaced recording CCDs?
 
Barry_Green said:
960x720 is guesswork. Nobody can say that it will "probably" have any specific resolution, because that hasn't been commented on.
Click to expand...

Well, it isn't but it is likely. The price point, the lattitude, f8 at 2000, ...
It all indicates that it wouldn't be 1920x1080 (obviously) or 1440 x 1080 ...

Barry_Green said:
What we can say is that having more than 1280x1080 is pointless because the codec only records 1280x1080. In fact, there is just about no recording format that handles 1920x1080.
Click to expand...

Indeed, could be usefull in other circumstances, but not here...

Barry_Green said:
And pixel shift is completely immune to temporal differences -- interlacing causes temporal differences, pixel shift is a way of sampling the CCD to generate a higher-resolution image.
Click to expand...

But it remains (under)sampling. Actually it is 'tricking' the human eye, by constructing a CCD config that samples the human eyes' 'shortcomings'. On specific images the results are below par. It remains however the best solution if you don't have enough pixels (Canon and Panasonic always did this), but it is in no way a replacement for full resolution CCD's. On the other hand: you can't expect that at this price point, I'm guessing. I - personally - believe they should stick with (real) 720p. I don't want to open ant 720/1080 debate again, but in pro-video it seems reasonable to match the CCD's closer to the format resolution. Post can work wonders if you want to 'up'grade or step up to a higher resolution format. It isn't all that simple: hardware interpolation is rigid in 1 system of interpolation (disadvantage), but of course, then you have the benefit from the pixelshifting (advantage). Still - I'd gone for 1280x720 or 960x720 (since it's DVCPRO)...
 
Until we get the true spec all this is just speculation. Wait and see then we can discuss how the techical solution will work.
 
It will just ake one look at resolution charts to show that whatever Panasonic has come up with will work, so just wait on until they do so! It doesn't matter what the rez of the chips is, as long as the final image looks blinkin good.

Graeme
 
I havnt checked how much but im sure it will be cheaper to travel to the states buy the camera there and then pay extra VAT at the airport.

As example today you get the

HDR-FX1 for 4 504,38$
GY-HD101E for 7 658,99$
HVR-Z1E for 7 530,25$
AG-DVX100AE for 4 247,55$

My real hope is to buy it in another EU country. Then I dont have to pay extra VAT.
 
I didn't think Pixel shift "tricked" the human eye at all..

I thought it sample more pixels through an offset in the alignment of the ccd's... am I wrong?
 
mezelf27 said:
On specific images the results are below par. It remains however the best solution if you don't have enough pixels (Canon and Panasonic always did this), but it is in no way a replacement for full resolution CCD's.
Click to expand...
It's an excellent replacement for "full resolution" CCDs, and in almost all ways, a PREFERABLE replacement.

You say that "on specific images the results are below par." I presume that you're talking about single-color situations, such as shooting under solid red light or solid blue light, etc, where only one CCD is actually in play, thus defeating the higher-res sampling possible when using the multiple CCDs. And if so, then yes, maybe so, that one specific case would result in a shot that's below par. Versus the full-resolution CCD, in which EVERY image delivers results that are "below par", meaning lower latitude, lower sensitivity, and a higher signal-to-noise ratio than necessary, all in a fruitless chase for resolution that the codec's going to give away anyway.

There is one 1/3" HD camera out there that justifies the choice of native-resolution CCDs, and that's the Canon XL H1, and that's only because it has the HD-SDI output port.

For all these other cameras, using their 4:2:2 or 4:2:0 color sampling systems, native-res is wasteful overkill, and unnecessarily compromises all the images (from a sensitivity and latitude perspective) in exchange for chasing a resolution boost that is not observable in the extremely vast majority of situations, and would never likely be seen in the recorded footage.

This is a situation unique to small-CCD camcorders; in the bigger cameras where bigger pixels are the norm, there's enough CCD "real estate" that the pixels are big enough to deliver satisfactory performance anyway, and in the bigger cameras you're more likely to encounter SDI or HD-SDI or 26-pin connectivity (which could take advantage of the native chroma res). But in the smaller cameras, where the size of the tiny pixels is a very real threat to video performance, making them smaller than necessary seems like an unwise choice.

In a 2/3" CCD camera, native res is fine and reasonable. In a 1/3" standard-def camera, native res is fine and reasonable. In a 1/3" high-def camera, or a 1/6" standard-def camera, native res gets in the way and results in actual compromises in every shot, and that's when you have to start taking a serious look at spatial offset/pixel shift in order to produce the best quality images.
 
no.. you not wrong.. its not an optical illusion.. its just a sampling technique. and in the end wysiwyg. its sampled as seen via the spatial offset. the information is less than 4:4:4 but since the codec doesnt record 4:4:4 to begin with its the same as if you had sampled with a native res CCD and tossed away the extra info.

the part that "fools the human eye" is the undersampling of the data by the codec which still looks good. its just using less info to construct an image that looks nearly the same becasue our eyes cannot precieve little changes in the areas the data is missing in that well.
 
Loki said:
I didn't think Pixel shift "tricked" the human eye at all..

I thought it sample more pixels through an offset in the alignment of the ccd's... am I wrong?
Click to expand...
You're not wrong, but in a way neither was mezelf. It does sample more pixels by an offset in the alignment. But inherently in doing so, it brings into question whether it can deliver full color resolution. The spatial offset/pixel shift system can deliver full native luma resolution, but the offset of pixels means that it can't deliver 100% of the chroma resolution -- I don't believe a spatial offset system could deliver uncompressed 4:4:4, for example. It could easily handle 4:2:2, but 4:4:4 may be beyond its reach.

What mezelf may have been referring to is the lower sensitivity of the eye to color resolution vs. luma resolution, implying that spatial offset works okay because the eye can't typically see that result anyway. However, I'd take it a step further and say that whatever codec you're going to use to compress the footage is going to throw away that extra chroma resolution anyway, so it's of little to no benefit to have it in the first place. (unless you're working with uncompressed 4:4:4 over HD-SDI, in which case it would retain that res).
 
Barry_Green said:
Versus the full-resolution CCD, in which EVERY image delivers results that are "below par", meaning lower latitude, lower sensitivity, and a higher signal-to-noise ratio than necessary, all in a fruitless chase for resolution that the codec's going to give away anyway.
Click to expand...

Well, before encoding the image has to pass the DSP, so it can't be overkill as by definition.

Better to have a good head-section and a mediocre format, then a mediocre head-section and an exellent recording format...
 
Loki said:
I didn't think Pixel shift "tricked" the human eye at all..

I thought it sample more pixels through an offset in the alignment of the ccd's... am I wrong?
Click to expand...

In simple terms: It is the green CCD that's positioned so to maximize the 'effect' of resolution, because that wavelength is exactly where the human eyes' best performance is.
 
So, try a wide shot of a forest or some nature (all greens) with pixel shift... It is a well documented dip in performance....
 
Right - so switch to CMOS, deal with the noise (boo), exploit the dynamic range (yay), make bigger sensors (yay - cheaper optics, better DOF), get higher yeilds on chips (cheaper! yay!), go native resolution (yay!), say goodbye to streaking artifacts (hip hip! hooray!), hello to easily adjustible variable frame rates (really!? yay!), and hallelujah for the new technology.

One can only hope they start to milk sensor technology for what it's worth. I'm betting the next generation of sub-$10k HD cams will have a lot to offer - almost purely from improvements in the sensor departments.

-Spiff
 
Yes, going to big cmos chips would be lovely, but they're not fast enough at affordable prices yet, but that day will come.

When dealing with affordable technology, I think Panasonic's pixel-shift approach is the right one for overall picture quality.

Graeme
 
but they're not fast enough at affordable prices yet
Click to expand...

I think that day is coming pretty quickly. Sony has already got a 4:3 1920x1440 CMOS chip that does 1080i60 on their HDR-HC1, which is the cheapest camcorder on the market. How much of that particular model is the cost of the chip I can't guess, but I'm wagering it's not that expensive.

The CMOS sensors in the digital still cameras obviously fit in the pricing constraints you're talking about - but the problem there is that they're 5-8 megapixel devices - which is waaaaaay more than we need for video. Dropping the resolution of the chips would probably increase yeilds and make them even cheaper. I think most of the cost is in getting the fab lines up and going, and dealing with the signal processing end of it right now.

CCD will remain the king of low light performance... which makes pixel shifting the correct solution this year... But by next year, it may not be something to brag about at all.

-Spiff
 
I saw a 3 CMOS Sony the other day... anyone know anything about it? I was under the impression that 1 CMOS was enough... can anyone explain?
 
You must log in or register to reply here.
Back
Top