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

If it were a viewable image you could have shown it to us.
It is a collection of data that are not viewable without jiggering them around.
Dcraw does that jiggering so you can see the pixels as a Bayer array. That I would call a viewable image. The raw data I would not.
I am not a software engineer, but if I were presented with 30 million 12-bit or 14-bit numbers I wouldn't store them as 30 million 16-bit numbers. I would pack them (not compression) so that the first 16 bit number is 12 or 14 bits of data followed by the 2 or 4 bits of the next number, then the remainder starts the second 16 bit number etc... (actually, today saving as 16 bit numbers might be possible, but these formats were invented a couple decades ago when memory wasn't as cheap as it is today).
That sort of format doesn't lend itself to viewing an "image", even a latent one, since computers are working with multiples of 8 bits.

In short, the raw data are not something you can view as an image on a normal computer without running it through software. The raw data contain an image, but it is latent, not viewable without some processing. Just like you couldn't see the negative film image without processing it first. Back then, film was described as having a "Latent Image". I see no difference now.

Ysarex Loc: St. Louis

Not using a web browser. Are you therefore saying the only files that are images are those viewable with a web browser? So all those Adobe PSD files aren't images as well. Show us how UHH and a web browser displays a PSD file.

Then there are no image files period. There is no image file type that can be viewed as an image on a normal computer without running it through software. JPEGs are not image files. TIFFs are not image files. PNGs are not image files.

How is a JPEG different? A JPEG goes through much more extensive processing to reach the point of being viewable on a computer than the raw files I displayed earlier. That's running the numbers in the file through software.


Good analogy in that a raw file is like a developed negative not a print. You can look at a developed negative and see the image there. Even though it's reversed tonally you can make out what it is. You can see the image the camera recorded and the same is true for raw files.

Joe

smf85 Loc: Freeport, IL

A TIF file can have any kind of data structure, that's the point - Tagged Image Format. By using the proper tags I can put any kind of imaging data into the file. Including the data elements coming from a sensor. As a matter of fact, the tags and subtags make that process easier as I can include under various tags all of the extra information that you'll need to process the raw file. Such as the EXIF, the bad pixel table, the white balance value or color temperature, the embedded JPEG image. Under one tag I can put the sensor output data, under a different tag I can put the CMKY The TIF file structure gives me a format to stick all this data in that is standardized. The standard also indicates that for the image tag types all the data needed to decode the file and present it should be within the file (more or less adhered to).



The photo sites are the pixels, or more accurately wells. Each photo site get converted to a digital value which undergoes quite a bit of processing to be ready for the raw file. The raw data is a tagged element within the file. The partial RGB array in the raw file is an image. If wanted to I could do the mosaic interpolation and generate full RGB values for each pixel and put them in the file without altering the sensor data.



I would say its more analogous to a developed negative - potentially greater than any print you can make from it. The chemists who developed the various developers worked out the best way to use each one of them and documented it. The engineers who develop the sensor packages also write down the information needed to successfully develop an image from the delivered data. Since each sensor implementation is unique, the development instructions are also specific to that sensor and fixed.

I think that a lot of confusion results from the fact that normally what you see in a TIF file is uncompressed RGB processed image data. There are many other types of image data in TIF files. The standard ones are:

bilevel
grayscale
palette-color
RGB full-color images
CMYK Images
YCbCr Images
HalftoneHints
Tiled Images
CIE L*a*b* Images

The first 4 are mandatory for baseline (all) TIF files the rest are for extended TIF files. Having multiple images, all of different types in the file is allowed. The most common of those is RBG CMKY dual images for commercial printing. If I want to put my own image format or compression format I can get a private tag from the administrating authority and use it. Standard readers wouldn't be able to read the image, but they would be able to read all the other tags in the file. Later I can get it published and added to the TIF standard.

Lastly, when you look at the NEF file specification the first thing it says is that its TIF format. Speaking of which, Nikon includes everything in the file that you need to render it - but they encrypted some of it to make reverse engineering of their files more challenging. This encryption scheme is also part of the reason why you can't modify a NEF file then save it as a NEF file.. So while I could take a Nikon raw file and edit it the file would have to be saved as a standard TIF file, not a NEF.
Ref: http://lclevy.free.fr/nef/

smf85 Loc: Freeport, IL

Actually they are stored as 16-bit elements in (usually) 32bit words. The data that comes off the sensor comes out as 'address, data' pairs in all kinds of interesting orders.

A raw file is just another image type that, like the JPG, needs processing to be viewed. The difference is that the viewer isn't built into the operating system. Some time ago for work I wrote a program that allowed the user to look at the hex data of a file directly [at the time it was a new thing, but now they're are commercial products that look far nicer]. But even then the data was processed - I had to convert the hex bytes into something that could be read by a human. So what you saw was my representation of the binary data, not the actual binary data. This is as close as it is possible to come to unprocessed data. [BTW, using it to look at image files was quite interesting].

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

It sounds to me as if you have a very broad definition of an image file while mine is narrower.

Yes, an image file is not viewable as an image in the sense that it consists of an array of numbers or triplets for color. The image is not viewable until it is displayed on a screen as pixels. At that point it's no longer a file.

The format that would take the least amount of processing to display the image on the screen would be a bmp. The jpg has to be decompressed to form a bmp on the way to the screen. The tif has to be sorted out to extract the image, but it's probably closer than the jpg. The png has to be decompressed to form a bmp but the compression is lossless unlike the jpg.

The Bayer array can form an image of a sort, but that is not the final image. You can produce something on the screen that is an array of pixels, but they are still monochrome pixels with different colors and require interpolation to get to the final image. I consider that "more processing" than the jpg/png/bmp/tif.

I do not know the format of a psd file. I would expect it to be a bunch of overlapping arrays (the layers) with 4 channels, R, G, B, and Alpha. Requires some processing to add all the pixels together to display on a screen. But that's purely a guess on my part as the the format.

So maybe we all have to find a good definition of "digital image".

Ysarex Loc: St. Louis

A raw file is not the final image. No argument. But you need the word "final" in there. A raw file using about the same software effort as your simplest BMP file displays as the image captured by the camera and is recognizable as such. That fits the definition of image and that was and remains my only point.

Joe

kymarto Loc: Portland OR and Milan Italy

I have three 8TB drives for my images, plus 3 more for backup.

smf85 Loc: Freeport, IL

The file formats are a combination address/length pointer and binary tag file. It’s about the oldest complex file structure there is. Basically it’s got the original image plus all the individual changes made to it in separate sections. There is also the final composite image. Plus other items that are needed, such as the modification history. Usually the elements are all compressed.

BMP files aren’t used in physical screen display any more (exception: specialized RGBY connected units). While there is much variation generally it’s now a Y-Cr format.

Marturo Loc: Western North Carolina

OP asked: one question.

Do you store your want to keep photos as Tiffs or Jpegs?

Do some of you know how to start a P-ing contest Post, by yourselves?

Architect1776 Loc: In my mind

Raw only.
Camera does it for me, download them, keep what I want and make JPEGS as needed for printing, sharing or viewing.

Marturo Loc: Western North Carolina

This continues to be the best answer. Why waste the step to create Tiffs, if RAW has more info
& can be converted to a Jpeg as needed.

I continue to see drops in Memory $ on drives & Thumbs with Tera Bytes all great for Photographers.

Architect1776 Loc: In my mind

Toment Loc: FL, IL

Save RAW