Essential "raw"
Jan 18, 2018 12:05:30 #
I can't find exactly what I want already, and I don't think it would be productive to hijack another thread - so I'll ask my question here: essentially, what is needed for a file to be truly "raw"?
For example, Nikon has several different sizes of NEF; I understand that some keep different precision of bits, but don't others keep different numbers of pixels?? If some type of interpolation is used to get the smaller pixel count, I would think that would be mild "cooking".
Then, of course, there are the manufacturers, such as Sony, Fuji, and Pentax, that are rumored to do Noise Reduction on their "raw" files.
For example, Nikon has several different sizes of NEF; I understand that some keep different precision of bits, but don't others keep different numbers of pixels?? If some type of interpolation is used to get the smaller pixel count, I would think that would be mild "cooking".
Then, of course, there are the manufacturers, such as Sony, Fuji, and Pentax, that are rumored to do Noise Reduction on their "raw" files.
Jan 18, 2018 12:18:03 #
My understanding; Bits are elemental parts of a digital file. Pixels are elemental parts of an image. RAW data is as it comes off a sensor in digital form, and isn't an image until a conversion is done. If, indeed, there are different precisions or noise reduction processes done to the RAW data, it is between truly raw and an image and can be called anything.
Jan 18, 2018 12:26:29 #
In principle, a raw file is the binary data created by exposing the sensor to light. This data is nothing but a stream of 0s and 1s. The camera firmware must convert that binary data into image data by some algorithmic means. A pixel is just a name for a cluster of solid state charge traps of some number (14 in my Sony a6300) and so occupies some physical area of the sensor. There is much electronics technical detail written in technical argy bargy. The manner by which any camera converts all that first order sensor charge state data into image data is not standardized which leads to other complications. Don't know if this helps or befogs.
Jan 18, 2018 12:27:41 #
rehess wrote:
"For example, Nikon has several different sizes of NEF;"
It is just the processor creating different files sizes, which require lower resolution, 8256x5504 (56 Mb files), 6219x4128 or 4128x2752. I shoot landscapes, so leave things BIG.
"Then, of course, there are the manufacturers, such as Sony, Fuji, and Pentax, that are rumored to do Noise Reduction on their "raw" files.
"For example, Nikon has several different sizes of NEF;"
It is just the processor creating different files sizes, which require lower resolution, 8256x5504 (56 Mb files), 6219x4128 or 4128x2752. I shoot landscapes, so leave things BIG.
"Then, of course, there are the manufacturers, such as Sony, Fuji, and Pentax, that are rumored to do Noise Reduction on their "raw" files.
Lossless compression is an algorithm that applies a slight compression on the file size, similar to creating a ZIP archive on your computer, to create a smaller file size without loosing definition. These are all menu settings on cameras, allowing user control.
Jan 18, 2018 12:41:23 #
Jan 18, 2018 12:46:45 #
rehess wrote:
I can't find exactly what I want already, and I do... (show quote)
All RAW files from the same camera have the same number of pixels. I don't know of a camera that save RAW files with fewer pixels than the maximum size. The sensors in most cameras can only capture 1 single color per pixel either R, G or B. Generally there are 2 green pixels per 1 red and 1 blue. After the exposure the camera would read the analog value from each of these pixel and digitize them in either 8, 12 or 14 bit each and some cameras allow option to select this. When you convert an analog value into a digital value you divide it into some number of steps. 8 bit has 256 step, 12 bit has 4096 steps and 14 bit has 16384 steps so you can see more bits can record finer tonal gradation. If the camera use a noise reduction technique mainly for long exposure to subtract the values from a dark noise values from the image values then it can be applied to the RAW data. The RAW converter will then create 3 color values per pixels from 1 color values coming from each of the pixel using the Bayer algorithm or some other improved version.
Jan 18, 2018 13:00:17 #
be seeing youJan 18, 2018 13:56:23 #
rehess wrote:
I can't find exactly what I want already, and I do... (show quote)
The small NEF (sNEF) are cooked and do no make any sense. A marketing thingy that is slowly back firing. Please read this article on sNEF (similar to sraw - canon)
Compressed NEF use a lossless compression similar to one used in rar, cab and the like. The data is untouched. Two different approaches.
Note that there is no 'true' raw despite all the claims. The sensor array produces a analog signal that is transformed into digital through a process that depends on the technology used as well as how a manufacturer wants to extract the data. The result is that a raw is cooked, just much less than anything else.
Let's describe digital raw as is cut meat at room temperature..
Nikon has been accused of hiding some data that is only available if you use their software. There is no proof of that but it shows that folks have questions at the low level of an image capture.
Jan 18, 2018 14:00:12 #
If one shoots raw, set the camera for the highest bit depth, lossless, and set the raw post-processing for 16-bit. Keep viable raw files and .xmp files, unmodified. When saving final jpg files, convert to 8-bit and use an 8 or higher setting. You won't regret it.
Jan 18, 2018 14:06:09 #
rehess wrote:
I can't find exactly what I want already, and I do... (show quote)
I think RAW is what the camera makers want it to be. They decide how they want the files to be in order for the user to be able to work with the data and get the needed or most useful info from the camera data.
Jan 18, 2018 14:14:15 #
BebuLamar wrote:
All RAW files from the same camera have the same n... (show quote)
So, this user is misinformed when talking about "raw" files??
http://www.uglyhedgehog.com/t-508181-1.html#8592213
Bill_de wrote:
They are stored with compression to save card and drive space, but with all the available pixels. What I am referring to is the camera's ability to actually provide an image with few pixels.
What I was trying to get at but didn't ask properly was the quality of the final image. Would it make a difference if I shot full size and reduced it on the computer after editing, as opposed to choosing a smaller size in the camera and then editing on the computer?
What I was trying to get at but didn't ask properly was the quality of the final image. Would it make a difference if I shot full size and reduced it on the computer after editing, as opposed to choosing a smaller size in the camera and then editing on the computer?
Jan 18, 2018 14:24:56 #
Canon says their mRaw and sRaw are downsampled from RAW... to suit the requirements of different users. Now we all know there is NO free lunch as some info has to be lost in the smaller RAW files. Canon even states that for utmost picture quality you should use the largest RAW file available. SOMETHING has to be forsaken in the smaller files; whether it makes a diff. is prob. up to the user through testing. I doubt whether any OEM is going to go into details as to what data they are cutting back on.
Jan 18, 2018 14:38:30 #
Rehess, somehow I can't quote your post but Bill De didn't give misinformation. He said the compressed RAW is smaller but it still contains the same information and the same number of pixels. Of course you know there are ways to compress file without losing any data right.
Jan 18, 2018 14:45:15 #
rehess wrote:
I can't find exactly what I want already, and I do... (show quote)
I did not see even one single reply that got it right!
A RAW file is sensor data encoded using a Bayer Color Filter Array. No more and no less. (It necessarily must be demosaiced to use.)
An image file is an RGB encoded bitmap. (There are other kinds of image files, but DSLR's make RGB bitmaps. All TIFF and JPEG image files are bitmaps, but BW might not be RGB.)
Note that a RAW file actually defines nearly an infinite number of perfectly "correct" image files. A bitmap file defines exactly one and only one image.
Either type of file is made up of just 1's and 0's. Compression has nothing to with it either, nor does noise reduction.
Jan 18, 2018 14:48:03 #
BebuLamar wrote:
Rehess, somehow I can't quote your post but Bill De didn't give misinformation. He said the compressed RAW is smaller but it still contains the same information and the same number of pixels. Of course you know there are ways to compress file without losing any data right.
No, he was the Original Poster in a thread about "small Raw"; when further explaining his Original Post, he said
Bill_de wrote:
What I am referring to is the camera's ability to actually provide an image with few pixels.
He is clearly talking about a "raw" file with a reduced number of pixels - there is no other way to read his restatement!
If you want to reply, then register here. Registration is free and your account is created instantly, so you can post right away.