forum.optyczne.pl

[sigmiarz]

hmm.. wynika z tego co on pisze że Foveon to nic specjalnego.. i Bayer jest od niego 4x efektywniejszy jesli chodzi o wykorzystanie fotonów. Najlepszy jednak jest komentarz innego usera dpreview:

Cytat:

I'm amazed how nerdy and technical photography has become. I bet the likes of Cartier-Bresson wouldn't give a flying about photons and pixel pitch.

[hijax_pl]

No tak, ale co z tego? Zaletą Foveona to ułożenie w pionie trzech senseli i niemal natychmiastowa informacja o WB

To, że nie rejestrują RGB tylko WYR to nic wielkiego.
To, że mniej światła łapią - od tego zależy efektywne ISO.
A że się rozpraszają fotony po różnych warstwach tworząc szum - no cóż - w Bayerze też nie jest idealnie

[ghost]

w sumie nic , tylko fani foveona często twierdzą, że jest odwrotnie, ze Fv jest wydajny i dobrze oddaje kolory. no to gość pisze że nie jest.

[ Dodano: Pon 01 Lis, 2010 11:46 ]

[hijax_pl]

[quetzatl]

Napisałem maila do Imatestu w sprawie dziwnych wyników z ich programu, które wg Optycznych sugerują wyostrzanie:

Cytat:

Hi, I need your opinion about results from Imatest 3.0 Master. Optyczne.pl, in their review of Sigma SD15 says, according to graphs from yours software that: (Google translation:) "Looking at the profiles on the border of black and white will not notice the sharpness reveals a local extremes on the chart. The course is smooth, but on the second chart is striking high value of the MTF for the Nyquist frequency. This suggests that artificially increase the gradient and interfere in the image by sharpening the highest spatial frequencies. However, we are unable to say whether this is due to camera software, or RAW converter." "The second issue is the value MTF30, which is often equated with the old (traditional) definition resolution. In the case of RAW files that caused the original software SD15 MTF30 LWPH reaches 2100 or clearly higher than the Nyquist frequency (in this case LWPH 1760)." link to review: http://www.optyczne.pl/13...5%9B%C4%87.html Do these results indicates that there is some kind of software sharpening done by Sigma firmware(or RAW converter)? Or does it mean that your software makes wrong calculations, because 2100 lines per image height(at MTF30) clearly have no sense when camera have only 1760 rows of pixels. Best Regards, Grzegorz Maj

I odpowiedź jaką otrzymałem:

Cytat:

Hi Grzegorz, The Imatest results look correct. The reason for the large reponse at Nyquist is that there is no anti-aliasing filter (or a very weak one) in cameras with Foveon sensors, where each pixel is sensitive to all colors (unlike Bayer sensors, which require interpolation), and hence color aliasing is rarely a problem. Anti-aliasing filters that reduce the MTF to near zero at Nyquist are found in all (to my knowledge) DSLR cameras with Bayer sensors-- that means all DSLRs except Sigma. This is almost certainly NOT the result of software sharpening or any artificial signal processing. I've seen similar results on earlier Sigma cameras. Software sharpening is usually visible as a bump in response in middle spatial frequencies-- 0.15 to 0.4 cycles/pixel. No such bump is visible. I would guess that the SD15 image is essentially unsharpened. Some sharpening would make the images look even sharper. The Sigma/Foveon camera clearly outperforms Bayer cameras with similar pixel count. Here is a result from an old Sigma image, downloaded from dpreview.com. This image seems to have a small amount of software sharpening, but it shows similar high response at Nyquist. The 2100 LW/PH number is not unreasonable. Remember that one line width = 0.3 line pairs, so this is the same as 1050 LP/PH. Numbers like this are not common, but the Sigma/Foveon cameras use unique sensor technology. I would very much like to try one of these Sigma cameras. I'm not sure about how well they operate, but the sharpness is very impressive. Regards, Norman

Napisałem maila do Imatestu w sprawie dziwnych wyników z ich programu, które wg Optycznych sugerują wyostrzanie:

Cytat:

Hi, I need your opinion about results from Imatest 3.0 Master. Optyczne.pl, in their review of Sigma SD15 says, according to graphs from yours software that: (Google translation:) "Looking at the profiles on the border of black and white will not notice the sharpness reveals a local extremes on the chart. The course is smooth, but on the second chart is striking high value of the MTF for the Nyquist frequency. This suggests that artificially increase the gradient and interfere in the image by sharpening the highest spatial frequencies. However, we are unable to say whether this is due to camera software, or RAW converter." "The second issue is the value MTF30, which is often equated with the old (traditional) definition resolution. In the case of RAW files that caused the original software SD15 MTF30 LWPH reaches 2100 or clearly higher than the Nyquist frequency (in this case LWPH 1760)." link to review: http://www.optyczne.pl/13...5%9B%C4%87.html Do these results indicates that there is some kind of software sharpening done by Sigma firmware(or RAW converter)? Or does it mean that your software makes wrong calculations, because 2100 lines per image height(at MTF30) clearly have no sense when camera have only 1760 rows of pixels. Best Regards, Grzegorz Maj

I odpowiedź jaką otrzymałem:

Cytat:

Hi Grzegorz, The Imatest results look correct. The reason for the large reponse at Nyquist is that there is no anti-aliasing filter (or a very weak one) in cameras with Foveon sensors, where each pixel is sensitive to all colors (unlike Bayer sensors, which require interpolation), and hence color aliasing is rarely a problem. Anti-aliasing filters that reduce the MTF to near zero at Nyquist are found in all (to my knowledge) DSLR cameras with Bayer sensors-- that means all DSLRs except Sigma. This is almost certainly NOT the result of software sharpening or any artificial signal processing. I've seen similar results on earlier Sigma cameras. Software sharpening is usually visible as a bump in response in middle spatial frequencies-- 0.15 to 0.4 cycles/pixel. No such bump is visible. I would guess that the SD15 image is essentially unsharpened. Some sharpening would make the images look even sharper. The Sigma/Foveon camera clearly outperforms Bayer cameras with similar pixel count. Here is a result from an old Sigma image, downloaded from dpreview.com. This image seems to have a small amount of software sharpening, but it shows similar high response at Nyquist. The 2100 LW/PH number is not unreasonable. Remember that one line width = 0.3 line pairs, so this is the same as 1050 LP/PH. Numbers like this are not common, but the Sigma/Foveon cameras use unique sensor technology. I would very much like to try one of these Sigma cameras. I'm not sure about how well they operate, but the sharpness is very impressive. Regards, Norman

[jaad75]

[dr11]

jaad75, z Leiką M9 jako DSLR chyba troszkę przesadziłeś

[jaad75]

dr11, no fakt, to tak z rozpędu...