fomoDVXpal
Active member
hey guys. supposedly it will have pixel shift. that makes it 1540x984.
does pixel shift have any serious limitations?
does pixel shift have any serious limitations?
Reel-Stream Andromeda @ NAB05
- Thread starter khill
- Start date
fiercecurry
Well-known member
Fomo,
Isn't it just another form of interpolation/upresing from what you stated on the Reel-streem forum?
I am not sure about the limitations either. But from what i heard its a more accurate way of interpolating.
Isn't it just another form of interpolation/upresing from what you stated on the Reel-streem forum?
I am not sure about the limitations either. But from what i heard its a more accurate way of interpolating.
khill
Member
These are great points you've made. Not to mention, yesterday Jeremy (of Reel-Stream) mentioned in the online radio interview that they are looking into developing (at a later time) hardware (possibly like Firestore FS-4) that would be able to attach to the hip maybe. I like that idea especially if the drive space is there and and it's light like Firestore FS-4. Man, I like the FS-4. We had it demo'd before NAB. It would be a great move if FocusEnhancements would support the andromeda system. Maybe someone will suggest it to them. Seeing that there would be so much raw data available for manipulation I'm hoping that maybe Reel_Stream's SculptureHD software might take on some attributes like SeriousMagic's DVrack. That way the look can really be nailed in the field/studio before post, but maybe SculptureHD does do this and with a FCP5's wavefore/vectorsopes, etc. we'll have success.
thisiswells
Retired
FCP4 does have waveform/vectorscope, but it isn't realtime while you're recording like DVRack...
It's basically for when you're playing back from HDD. Email DVRack and have them get in contact
with Reel Stream. The developer of that system seems very open and willing to port new cameras
into the system. He basically just needs a windows driver for the camera and it seems like Reel-Stream
has already done the leg work in making the driver.. I think it would be worth a shot to email
them about it.
It's basically for when you're playing back from HDD. Email DVRack and have them get in contact
with Reel Stream. The developer of that system seems very open and willing to port new cameras
into the system. He basically just needs a windows driver for the camera and it seems like Reel-Stream
has already done the leg work in making the driver.. I think it would be worth a shot to email
them about it.
fomoDVXpal
Active member
pixel shift according Juan:
"The CCD's in the DVX are not aligned. There is a half-pixel shift in both axes on the green CCD which carries about half of all the image information. This half-pixel shift captures details which lie in between sensor elements of the red and blue sensors. Utilizing this knowledge, a mosaic can be built which is similar to what is obtained from a single color HD CCD. The image can be demosaiced in a similar manner to obtain the HD image, which is 4 times the size of an individual sensor, 1546x990."
so does anyone know what that means? are they real pixels or simply guessed?
"The CCD's in the DVX are not aligned. There is a half-pixel shift in both axes on the green CCD which carries about half of all the image information. This half-pixel shift captures details which lie in between sensor elements of the red and blue sensors. Utilizing this knowledge, a mosaic can be built which is similar to what is obtained from a single color HD CCD. The image can be demosaiced in a similar manner to obtain the HD image, which is 4 times the size of an individual sensor, 1546x990."
so does anyone know what that means? are they real pixels or simply guessed?
I am not the expert on this, but from what he is saying, it looks to be "simply guessed." Pixel shift is being employed to obtain the larger fram size ... The DVX CCDs simply do not contain that amount of pixels, but the way the green CCD is shifted, it allows it to capture enough information to interpolate this 1546x990 image.
The thing is....a 6mp digital camera has only 2 million photosites of each color. So it's really an interpolated 2mp image.
Since the green pixels aren't aligned exactly, it makes sense that extra information could be extracted in this same way (basically, bayer interpolation.) It wouldn't be as good as the real thing, but, then again, Foveon-based dSLRs are being outsold by a lot of cameras using bayer interpolation.
Since the green pixels aren't aligned exactly, it makes sense that extra information could be extracted in this same way (basically, bayer interpolation.) It wouldn't be as good as the real thing, but, then again, Foveon-based dSLRs are being outsold by a lot of cameras using bayer interpolation.
fomoDVXpal
Active member
xl1 uses pixel shift.
http://www.truecopy.com.au/canon_XL1.html
"3 CCD - The XL1's high-quality color recordings come from its use of a 3 CCD system with pixel shift. Three separate one-third inch CCD's are each dedicated to a different primary color red, green or blue. After passing through the lens, the light is split into its primary components and sent to its own CCD.
This gives natural color reproduction together with low color noise and low aliasing. Each CCD sensor offers 320,000 pixels, with each pixel about 150% larger than normal for this type of camcorder. The result is an improvement in sensitivity of about 4dB, allowing shooting in extremely low-light conditions.
Pixel Shift - Human eyes are most sensitive to green. The green component of a video signal contains about 60% of the picture detail. Canon has physically shifted the green CCD the distance of half a pixel horizontally in respect of the red and blue CCDs, and then electronically shifted the green signal vertically by the same distance. This increases the sampling points, giving a resolution which rivals that from 470,000 pixel systems. Picture quality is exceptional, with low light performance and minimal vertical smear. The larger light capturing area of each pixel also gives a higher signal-to-noise ratio and greater dynamic range. In addition, amazingly clear still images can be obtained from videos of moving subjects."
http://www.truecopy.com.au/canon_XL1.html
"3 CCD - The XL1's high-quality color recordings come from its use of a 3 CCD system with pixel shift. Three separate one-third inch CCD's are each dedicated to a different primary color red, green or blue. After passing through the lens, the light is split into its primary components and sent to its own CCD.
This gives natural color reproduction together with low color noise and low aliasing. Each CCD sensor offers 320,000 pixels, with each pixel about 150% larger than normal for this type of camcorder. The result is an improvement in sensitivity of about 4dB, allowing shooting in extremely low-light conditions.
Pixel Shift - Human eyes are most sensitive to green. The green component of a video signal contains about 60% of the picture detail. Canon has physically shifted the green CCD the distance of half a pixel horizontally in respect of the red and blue CCDs, and then electronically shifted the green signal vertically by the same distance. This increases the sampling points, giving a resolution which rivals that from 470,000 pixel systems. Picture quality is exceptional, with low light performance and minimal vertical smear. The larger light capturing area of each pixel also gives a higher signal-to-noise ratio and greater dynamic range. In addition, amazingly clear still images can be obtained from videos of moving subjects."
ddh
Well-known member
As I understand it(?), HD in the HDX200 will be getting it's picture off 4.3 CCDs (considering the CCDs are 16.9 shaped) by reading the center line and masking off the upper and lower elements of the picture. If that's the case, wouldn''t reel-stream be doing the same thing but getting 4.4.4.?
thisiswells
Retired
....and to think p2 is hard to grasp.......