網路會員
|
Canon出名出機快，Nikon慢D，D200一定就來出，等下la。
如果5D個對焦系統與20D同級我都未必買，似乎用Canon要用1系才能有較好的低光對焦準度。
其實1DMkII N用12Mp FF 再加個8Mp crop mode就有睇頭D。
20D高ISO低noise偏偏AF低光對焦準成不足，FF機正但wide angle頭質素又一般，始終欠DD。
|
網路會員
|
以前網上睇到有人提過，canon full frame邊位唔得，唔係因為Ｄ鏡有問題，而係sensor前面仲有好多filter，1.6fov仲未有大問題，但邊位由比較斜既角度進入，filter嚴重影響sensor感光。有Ｄ人比較過，菲林機無呢個問題，但我冇canon菲林機，更加無full frame機，所以都唔能夠肯定真相係點。我反而希望邊位唔理相係因為原廠超廣角鏡品質有問題，canon eos個mount咁鬼大，加個adpater就機乎乜野135鏡都插得落，原廠唔好仲有大把選擇，遇掂super wide冇自動對蕉都唔係問題，只係收光圈後測光會唔準姐，但都就到丫。。。 |
網路會員
|
不是filter問題,而是CCD/CMOS的基本問題,感光元件像是一個直身水杯,是有高度的,如果光線不是垂直射進感光元件,所收集的光線相對減少,從而失光(只有olympus E系所用的Zuiko Digital鏡頭有針對這個問題作出調整),如果感光元件密度不高,即口徑大一點,問題應不會太嚴重. |
網路會員
|
CCD/CMOS imager前面的microlens對直射光要求高D，但wide鏡邊位質素差D會令情況更壞。Rob Galbraith專業數碼網站曾經批評Canon wide邊位唔係好得，好多專業攝影師用被逼改用Leica wide鏡。但攝記就唔方便了。
Nikon早著先機發展專用DX鏡，Canon率先發展FF，只是wide未能有突破。
或許將來會解決microlens或wide鏡缺點，而成本又不高，不過要耐心等。 |
網路會員
|
The window opening is one of the issue.
The microlens is another issue, microlens is used to minimize the small window opening issue but causes another issue.
Although the resolution of lens resolution is usually much higher than current imager, but the Bayer pattern of the cells and the decoding of the signals will enlarge any imperfect light focus on them, that's why DSLR needs higher resolution lens than film. |
網路會員
|
>Bayer pattern of the cells...
So, it requires 4 times more pixels to generate a RGB pixel on the image??? |
網路會員
|
Not this issue.
The bayer pattern (RGBG)decoding need to use the other data surrounding a pixel (each pixel only detect one colour) to consider what exact colour and brightness it is, it is OK if all the pixels receive the correct light, but due to the imperfection of the lens, like sharpness, abberation etc., the light received by each pixel may not be exactly, and the computing of the colour and brightness signal based on these not perfect data will make the result worse than original imperfection. This is specific to bayer pattern imager, so a not so good lens will perform adequately on film but may be not satisfactory on CCD/CMOS imager.
The Foveon x 3 imager should be better but this type of imager has its own weakness, high cost and lower ISO rating. |
網路會員
|
Oh, that's why the number of effective pixels is always smaller than the total number of pixels on the sensor. The quality of the image depends mainly on the algorithms used in interpreting the light signal received from the adjacent pixels... |
網路會員
|
>全倒SENSOR以相同製程之下, 比APS所產生的熱力為高, 長時間曝光更不利, 這是電子學上的現象, 無得拗. Did you distinguish the difference between Sony F717 and F828??  |
網路會員
|
>全倒SENSOR以相同製程之下, 比APS所產生的熱力為高, 長時間曝光更不利, 這是電子學上的現象, 無得拗.
這是ccd的問題，相同長曝情況下cmos產生的熱力較少，所以eos20Da是天文界至愛。canon也因此比其他廠方更有能力發展10M Pixel以上的full frame d-slr，也是nikon在d2x即使是dx format sensor也要轉用cmos的其中一個原因。
>Did you distinguish the difference between Sony F717 and F828??
這又是另一原因了:同是2/3"ccd，但f828放進了800萬pixels，比f717的pixel密度高很多，令pixel size小了，noise也多了，但這和sensor size無關吧？更與full frame vs aps的比較無關。 |
網路會員
|
問題:
雙倍sensor是否雙倍熱力? 如果時, 雙倍熱力又會否做畫質下降? |
網路會員
|
可否從另一角度理解：
製程／pixel size相同，每單位面積所受到的熱能並不會隨imager大小而改變，因此從雜訊的角度而言，大imager的雜訊問題不會較嚴重。 |
網路會員
|
咁phaseone, imacon同fuji D digital back實無人買攞 ^^'' |
網路會員
|
Pixel density = number of pixel / AREA '>http://forum.hkgolden.com/faces/clown.gif"> |
網路會員
| |
網路會員
|
So? Please elaborate. |
網路會員
|
suppose number of pixel remain unchange.
Area of full frame > Area of APS
=> Pixel density of full frame < Pixel density of APS
( => "Heat" density of full frame < "Heat" density of APS )
=> Noise level of full frame < Noise level of APS
|
網路會員
|
Yes, I understand your formula, but just wondered if it has anything to do with the
preceding discussion.
Maybe I can try to add the missing link - perhaps we're talking about/rebutting the same
thing (presumably Superman's "雙倍sensor" meant doubling the size of the imager, rather
than doubling pixel count).
My logic is that when pixel size (remotely means pixel density, in my case) remains unchanged, no matter how big the imager is, the "heat density" remains constant, so as the noise level.
|
網路會員
|
>presumably Superman's "雙倍sensor" meant doubling the size of the imager, rather than doubling pixel count From 上文下理, his meaning is "doubling the number of CCD cells" |
網路會員
|
But, how about 相同製程 if he meant doubling pixel count?
BTW, anyone saw the sample pics of 5D? I think Canon is rather honest - the corner sharpness is simply suck. |
