RAW, same settings, same camera- why file size differences? (Page 3): We might have to define "size." Are we talking about the file size in MB or are we talking about the file size in...

russelray Loc: La Mesa CA

We might have to define "size."
Are we talking about the file size in MB or are we talking about the file size in pixel height and width?
If MB is the topic, all of my RAW files—and I shoot an average of 500 pictures a day—are different.
If the topic is pixel height and width, all of my raw files are the same 6000 pixels by 4000 pixels.

Linda From Maine Loc: Yakima, Washington

warrenvon wrote:

Linda, Are all of your camera bodies in uncompressed RAW?

According to my Olympus manual, it's a loss-less compression orf (raw) file, with "approximate size 17.3 MB." I can't find any reference to compression in my Panasonic manual except for jpg. I no longer have the T3i.

For anyone else who quoted me, please refer to information provided by others. I'm reading contradictory comments, including "raw contains a jpg, so yes the file sizes will be different," to "all my raw files are exactly the same size."

John_F Loc: Minneapolis, MN

I sense aneed to take a contrary view. The camera sensor is an interconnected transister network of charge traps. These are organized in pixel groups of a certain bit depth. My Sony a6300 sensor has some 24,000,000 pixels of 14 bits depth. Each bit creates a 0 or 1. A photo of a green wall will have all pixels with identical values. As you look at a pixel you will see a string of 0s or 1s. These are coefficients of powers of 2: 2^13 down to 2^0. These may be converted to a decimal number or a number in any other base. Hexadecimal is common, base 16. The binary sensor image data will be a constant, but an image file also will contain file header data which is necessary for outside-the-camera devices to read and display the image.

Gene51 wrote:

Content - a shot of a clear sky sunset will have considerably less information than a busy landscape or city street.

DirtFarmer Loc: Escaped from the NYC area, back to MA

John_F wrote:

.... A photo of a green wall will have all pixels with identical values...

You are neglecting noise. There is always noise. Statistically the variability of the signal will be the square root of the number of electrons in each pixel (Gaussian statistics, for very small signals you have to use Poisson statistics). So a photo of a uniform color wall will not have all pixels at the same value.

jeep_daddy Loc: Prescott AZ

Linda From Maine wrote:

File size is dependent on how much "stuff" is in your image: number of colors and details. For example, a photo that has a lot of blue sky will have a smaller file size (MB) than a garden full of flowers.

(dang, 46 seconds too late to be first

)

warrenvon Loc: Ellicott City, MD

DirtFarmer wrote:

You are neglecting noise. There is always noise. Statistically the variability of the signal will be the square root of the number of electrons in each pixel (Gaussian statistics, for very small signals you have to use Poisson statistics). So a photo of a uniform color wall will not have all pixels at the same value.

Yes . . there is noise as a variable. But isn't noise still reduced to a pixel site number representing it and as such is seen as occupying the very same bit count as any other viable image pixel?
It begins to sounds like we may be correct in our responses. the mapped data from each uncompressed pixel remains the same; while there may well be header data that most likely remains the same for each image unless it is variable in nature. Even if the header data varied, it would most likely a very small impact on file size. That leaves the camera data generated for each image to represent the .JPG representative for that image and that will vary as a function of image content. Not knowing how that data is constructed, I would guess that is where the files difference would be located.

As noted all of this conversation is purely academic when the cost of storage is such a minor factor in our photography expense.

Dikdik Loc: Winnipeg, Canada

Linda From Maine wrote:

According to my Olympus manual, it's a loss-less compression orf (raw) file

If it's lossless compression, file sizes can be different. I was looking it up, and the header construction, is fixed, and should be the same for all files, and if the number if pixels is the same, then the file size should be the same. As one poster noted, it's just data. If any compression is used, then even if lossless, the file sizes will likely change.

Dik

Kuzano

rb61 wrote:

I didn't expect this on RAW files. Copied directly from camera, did not open in an application. In camera processing for RAW?
OMD 10ii

A rare find on my part. Of two files taken at the same setting and no changes, otherwise. The file that is best focused has a higher byte size count. Of course there are many other factors, but focus affects file size. I have tested this to my satisfaction.

f8lee Loc: New Mexico

John_F wrote:

I sense aneed to take a contrary view. The camera ... (show quote)

Well, since only 50% of the photo sites on the sensor are covered with green filters (the rest are red and blue) then half of the photo sites will register the green while the other half will pick up (ideally) no photon activity (except infrared photons from heat - otherwise known as noise). Only after the raw file is demosaic'd will it "decide" that each pixel in an image file will be green (say, RGB 0:256:0)

f8lee Loc: New Mexico

rplain1 wrote:

You must have magic cameras.

Nope - just an observant eye. Now I shoot raw only, no JPEG in camera, but that should not make a difference.

In truth, the raw files might differ by all of 1 or 2 MB but that could be due to the way they are stored on the sectors on the disk.

rb61 Loc: Maple Grove, MN

Linda From Maine wrote:

This is interesting to know, and I guess in Gene's haste to be first, he didn't read the opening carefully

A quick check in my file manager:
Olympus EM10: orf raw files from 13.2 - 16.9 MB
Panasonic G7: rw2 from 16.9 - 17.2 MB
Canon T3i: cr2 from 21.9 - 30.6 MB

The last group of images that I shot ranged from 15.8 to 18.0 MB
I will need to look for information regarding whether or not my camera (E-M10ii) does any compression with RAW files. I have not run across anything like this in the menu options.

Linda From Maine Loc: Yakima, Washington

rb61 wrote:

The last group of images that I shot ranged from 15.8 to 18.0 MB
I will need to look for information regarding whether or not my camera (E-M10ii) does any compression with RAW files. I have not run across anything like this in the menu options.

I got my info from the chart on page 131 of the user manual. I inferred there is no way to change the compression of raw, but I've been known to be wrong

aubreybogle Loc: Albuquerque, NM

DirtFarmer wrote:

You are neglecting noise. There is always noise. Statistically the variability of the signal will be the square root of the number of electrons in each pixel (Gaussian statistics, for very small signals you have to use Poisson statistics). So a photo of a uniform color wall will not have all pixels at the same value.

Agree.

rplain1 Loc: Dayton, Oh.

I don't know about any of this stuff. I just look at my RAW files in Bridge. I have used a Canon 5D, 5DII and 5DSR. Every single file is a different size. And when I look at the pictures, it is exactly and Linda said two pages ago. The cleaner pictures are smaller files, the busier pictures are larger files. Even when I compare a two-shot burst taken on a tripod, there is still a difference - sometimes up to 1mb. All the explanation is not going to change what my computer tells me. Or maybe it's just my cameras - someone else said his are always the same. And it is not Bridge - I can just go to DOS and look at the file size. Still different.

xt2 Loc: British Columbia, Canada