Good question. I assumed it was cropping the image.
Thread: who said f2.8 for the zoom lens
Results 31 to 40 of 54
10-19-2006 08:16 PM
10-19-2006 08:27 PM
I was under the assumption that the extra ".4K" was overscan to see what is just outside frame."I've heard that this project is impossible... hehe" ~ Jim Jannard
The Pursuit of Happyness
10-19-2006 08:35 PM
- Join Date
- Jun 2006
The full pixel array is listed as 4900 X 2580. The active pixel array is 4520 X 2540.
2540p (RED-speak) converted to 4K is still a downrez by my math.
10-19-2006 08:57 PM
- Join Date
- May 2006
- sydney, australia
at 2540p you can on crop, not downrezzing about it!
10-19-2006 09:10 PMOriginally Posted by david farland
- Join Date
- Jun 2006
Can anyone confirm? 2540p to 4K - a downrez or a crop????
10-19-2006 10:54 PM
4.9K --crop--> 4.5K (look around in viewfinder) --crop--> 4K
As correctly indicated by others no resampling / resizing in RAW recording mode (only cropping). RGB recording modes have some resizing options (2K, 1080p, 720p)
2K in RAW would be cropping (or windowing the sensor is what basically happens).
10-19-2006 11:32 PMOriginally Posted by Muttondraw
Most people I've found that complain about Bayer patterned sensors "not being full resolution" don't understand that just because a pixel has a bias, it doesn't mean the pixel stops doing it's job.
We use very sophisticated algorithms to maintain maximum resolution and sensitivity. It wouldn't surprise me if the folks at Red do something similar.
Hot Rod Cameras
My hilarious twitter twocks @hotrodcameras
10-20-2006 01:55 AMOriginally Posted by Illya Friedman
I would like to underline the point that I am not complaining about this issue though, I just thought it was worthwhile pointing out that there is a difference between the number of pixel you count on a sensor and the actual resolution you can achieve from that sensor if you are using the Bayer approach, rather than using something like the Foveon sensor. In the particular case of the Red camera with an active pixel area of 4520 x 2540 I would assume they are probably going to achieve 3k to 3.5k resolution. Maybe Graeme would contribute to this thread, I am sure he could put some more accurate figures to this,
As soon as a comparison is made to imaged resolution then we have to be comparing to the imaging resolution achievable by the sensor.
10-20-2006 06:39 AM
It's pointless to compare bayer to foveon when foveon's have neither the resolution nor the fps needed for a digital cinema camera. And although foveon cameras never seem to come with an OLPF, they certainly do need a hefty dose of anti-alias filtering as you can see from the severe aliassing artifacts in any foveon image. And 3 chip systems don't work with the size of sensor we're using, so they're out of the running too.
Say you're shooting 4k raw. You're recording 4096x2304 pixels. And that's the resolution of the final RGB image also. What the measured resolution will be is something smaller due to the optical low pass filtering which is necessary to avoid nasty aliassing artifacts occuring. It's that filter that puts an upper limit on the measured resolution, and the bayer demosaicing algorithms can get as close as you want to that, given enough render time :-) Practical algorithms can get very close though.
I state that the figure is > 70% which I think is accurate enough, because one of the nice things about a good demosaic algorithm is that it doesn't look uprezzed and artifacty the way it would look if you took a 30% smaller image of "full resolution" and blew it up to 100%. A good demosaic can look rather nice and smooth, and film-like in a good way, as the images we showed at IBC demonstrated.
10-20-2006 07:18 AMOriginally Posted by Graeme_Nattress
Thanks for that Graeme. I saw the footage at IBC and it looked stunning
So going back to the original point. Assuming the > 70% figure. Would it be fair to say that the maximum resolution you could capture using the 4K on the RED one would probably be about equivalent to the top end of what you could get off a 4K scan of 35mm film in ideal circumstances (assuming Jims figure of 3.2K for 35mm)
Would there be much of an issue upon the use of a B4 lens for 1080p, assuming the lens is only going to be able to use 1080 x 1920 pixels, or are the effects of the mosaic going to be swamped by the anti alias filter again. Would there be a significant resolution advantage in using S16 lenses at 2K instead and then downrezzing to 1920?
Last edited by Muttondraw; 10-20-2006 at 07:24 AM.