RAW ISO measures are inferior to manufacturer ISOs: is this a problem?
The RAW ISO measured value is almost always inferior to the ISO that you decided to use with your camera. Take a Canon EOS 60D, for instance. When you select ISO 200, the measured RAW ISO sensitivity is 160. At ISO 800, the measured value is 632… and so on. This simple fact compels some users to complain that the manufacturers are cheating [lying?]. Sure, this is a trick, and we are about to explain it, but it is not cheating.
ISO sensitivity compared: In grey, the nominal sensitivity. Some cameras, such as the Nikon D70 (orange dots), have a RAW ISO corresponding to the JPEG value. Many others, such as the Canon EOS 60D (red dots), have a much lower RAW ISO than the nominal value.
In fact, it is precisely the JPEG ISO value that all the manufacturers publish. They do so because JPEG (or any RGB) output is the visible output that photographers use. So when you select ISO 800 on your camera, you’ll have a JPEG ISO at 800, but the RAW ISO will be at (for instance) 550. The JPEG results are achieved by playing with the tone curve shape. This is absolutely legitimate: the ISO standard allows manufacturers to use this JPEG value. They are not cheating.
An old custom
Not cheating—okay. A trick? Maybe. The RAW-to-JPEG conversion allows cameras to achieve improved speeds, and to boost ISO values. And this is nothing new. What happens today with digital cameras also happened in the past with film cameras. When manufacturers produced an 800 ISO film, for example, they often used revamped 400 ISO film, and simply “asked” that it be processed differently. So when lab operators received a film labelled 800 ISO, they treated it differently (keeping it twice as long in the chemicals, for instance) in order to produce a stronger signal on the film. Digital camera manufacturers have simply implemented this same strategy in the camera body, and so could be seen as a “good old tradition.”
Moreover, underexposing the RAW file allows manufacturers to use their own complex algorithms to obtain a better output for the highlights while retaining good medium tones.
And this is also the reason why everyone should use genuinely efficient RAW converter software.
回复 ( 4 )
对于官方宣传口径就是我们APSC比全幅强的公司 你还想怎样
业界头号吹牛逼王，用索尼淘汰下来的阉割原生iso100的残幅CMOS的命，YY逆袭全幅的心，靠着虚标ISO+RAW涂抹忽悠智障富士粉，到处外面吹牛逼忽悠小白，jpg偏色片的连它妈都不认识，还管那叫直出色彩好，1600万像素最低iso200的cmos实际水平就索尼已经停产淘汰下来最低端现在只卖800块的NEX3N水平，卖的很人家A7一个价，简直就是典型的人傻钱多速来
DXO的ISO sensitivity不是测试虚标与否的，你们很多人究竟看不看DXO的测试说明，究竟是懒癌晚期，还是英文水平太野鸡，根本看不懂？
ISO sensitivity测试的无非厂商的ISO策略，即更偏重模拟增益还是数码增益，就那么简单。至于ISO有没有cheated，通过那个图是反应不出来的，再说了，Fuji的大部分机器，现在DXO根本不测好么？（当然这里绝对不是说Fuji是干净的）
至于怎么测ISO cheated，很简单，同曝光三角出JPEG比较亮度（忽略F值可能有的差异），明显偏暗的就是作弊。
RAW ISO measures are inferior to manufacturer ISOs: is this a problem?
The RAW ISO measured value is almost always inferior to the ISO that you decided to use with your camera. Take a Canon EOS 60D, for instance. When you select ISO 200, the measured RAW ISO sensitivity is 160. At ISO 800, the measured value is 632… and so on. This simple fact compels some users to complain that the manufacturers are cheating [lying?]. Sure, this is a trick, and we are about to explain it, but it is not cheating.
ISO sensitivity compared: In grey, the nominal sensitivity. Some cameras, such as the Nikon D70 (orange dots), have a RAW ISO corresponding to the JPEG value. Many others, such as the Canon EOS 60D (red dots), have a much lower RAW ISO than the nominal value.
In fact, it is precisely the JPEG ISO value that all the manufacturers publish. They do so because JPEG (or any RGB) output is the visible output that photographers use. So when you select ISO 800 on your camera, you’ll have a JPEG ISO at 800, but the RAW ISO will be at (for instance) 550. The JPEG results are achieved by playing with the tone curve shape. This is absolutely legitimate: the ISO standard allows manufacturers to use this JPEG value. They are not cheating.
An old custom
Not cheating—okay. A trick? Maybe. The RAW-to-JPEG conversion allows cameras to achieve improved speeds, and to boost ISO values. And this is nothing new. What happens today with digital cameras also happened in the past with film cameras. When manufacturers produced an 800 ISO film, for example, they often used revamped 400 ISO film, and simply “asked” that it be processed differently. So when lab operators received a film labelled 800 ISO, they treated it differently (keeping it twice as long in the chemicals, for instance) in order to produce a stronger signal on the film. Digital camera manufacturers have simply implemented this same strategy in the camera body, and so could be seen as a “good old tradition.”
Moreover, underexposing the RAW file allows manufacturers to use their own complex algorithms to obtain a better output for the highlights while retaining good medium tones.
And this is also the reason why everyone should use genuinely efficient RAW converter software.
多翻翻DXO的测试就会明白，目前绝大部分厂商都偏向比nominal ISO更低的策略，如Pa，S和N都是差1/2stop，Oly直接差1stop甚至更多，为啥呢？
原因很简单，偏向保护highlight细节，实际上目前根本也没多少机器的ExposureBaseline是0。
X100就是早年DXO测过的机器，超过一定程度的ISO就全部是通过数码增益完成的，就是这么回事。其实厂商哪怕不用模拟增益，全程数码提亮也不叫作弊，什么叫作弊，本文开头已经阐明了。
ISO本质上是根本无关痛痒的东西，无非一个亮度标尺，和信噪比本身都不存在关系，厂商的ISO作弊与否说白了就是有没有影响测光，判别准则就这么一个而已。
富士的色彩确实很棒，不影响用户实际拍摄体验的话，虚标iso也不会太让人在意，顶多快门慢一些。
可如果虚标了还打着“超越全画幅的高感”这样的宣传口号，确实有失偏颇了。（最近看到有测试出来标称iso3200实际iso才900多）