RAW vs JPEG

[DanK]

Alright, back to Richard's original posting, with a datum.

While on lunch break, I stepped out and took 4 quick shots. It's cloudy here today, so I deliberately set the WB to tungsten. I then took four shots of a corner of my yard. Two were raw, one with a whiBal card. Two were the largest JPEG (this was with a 5D3), set to standard picture mode, again one with a whiBal. Other than those two settings, the settings were identical.

I then imported them into LR. The originals, of course, were way too blue. In each pair, I did only one edit. I set the white balance in the images with the whiBal using the eyedropper tool. Then I copied that setting to the other image in the set.

Here is the one from the raw file. The colors are correct.



Here is the one from the jpeg. The colors are not correct, and the slightly overexposed yarrow is blown out, as is some of the greenery. I am guessing that this latter is from saturation and contrast adjustments made in the JPEG processing.



I first started shooting raw years ago after I captured a very good photo of my son with the wrong white balance, and I couldn't get it entirely fixed. A cousin who teaches photography at the university level told me that it would be a lot easier if I shot raw.

So, while it may not happen often, the fact is that baking in a set of edits by shooting JPEG does in fact sometimes mess things up.

now back to work...

[george013]

Images exist in multiple forms and are constantly being converted from one form to another as they are moved through a system. This happens even if no image is being displayed at all. For example, many folks think of a raw file as the data on the sensor. In fact, all manufacturers are processing as that image is moved into a memory buffer and on to a storage medium.

If we took a simple 3x3 sensor from a common manufacturer the camera maker can create it's own unique format for a data file and label it raw.

123
456
789

might be read by one manufacturer as 123456789 and by another as 147258369. The manufacturer then adds EXIF data and it's own proprietary data to the file.

The data gets changed again to be displayed on an EVF or LCD display. In fact those may well be two entirely different conversions. For example, my Nikon V2 has a 1,440,000 pixel EVF and a 921,000 pixel screen.

The image data is in whatever state it is at any one point in the processing cycle but in most instances that is unknown to the user. Most of that time, it is not one of the storage formats.

With all respect, I don't know what you are saying. I read it several times.

@Ted,
Don't get pissed. When I say I don't know much about colours, thats true. But also a pixel value is just a value, the colour comes from the screen or printer. I don't know what you try to proof with your example.

@Mike,

When the resulting JPEG is produced, it is exactly the same size in pixels as the RAW file produced by the camera. I have concluded that the File Format Conversion command used in this manner is only extracting the embedded JPEG, which in turn proves that the JPEG embedded by the camera is a full-size JPEG..

Maybe I don't read well, but to me this is complete nonsense. If you take out a embedded JPG out of the RAW, you get the same size as the RAW?????? Something A-B=A. The your RAW-file is a JPG.

Is there anything about the above that I am not properly understanding? The reason I am asking is that I am currently involved in an online discussion with people who are insisting that Nikon cameras do not embed a full-size JPEG in their raw files."

The original embedded JPG is full size. Only a less quality. When I'm on the other pc I will look for some examples.

George

[Mike Buckley]

@Mike,

Maybe I don't read well, but to me this is complete nonsense. If you take out a embedded JPG out of the RAW, you get the same size as the RAW?????? Something A-B=A.

You would be correct if we were referring to the size of the electronic file expressed in bytes. However, we're discussing the size only of the image expressed in pixels.

The original embedded JPG is full size. Only a less quality.

I didn't know a full-size JPEG embedded by the camera can be of lower quality other than the obvious limitations of the JPEG file format. I've never heard of that concept and don't understand it. If you can point me to something that explains it in layman's, non-technical terms, I would appreciate it.

[george013]

You would be correct if we were referring to the size of the electronic file expressed in bytes. However, we're discussing the size only of the image expressed in pixels.



I didn't know a full-size JPEG embedded by the camera can be of lower quality other than the obvious limitations of the JPEG file format. I've never heard of that concept and don't understand it. If you can point me to something that explains it in layman's, non-technical terms, I would appreciate it.

http://en.wikipedia.org/wiki/JPEG
Somewhere in the bottom there are some examples.

I just did a test.
Given a NEF original and a NEF edited. From both I extracted a JPG using IrfanView. Quality 100. The only thing I did was a small WB-correction to force a save.

NEF original 19.584.685 JPG original 2.617.273
NEF edited1 25.398.558 JPG edited1 6.758.155

Then I loade the NEF edited again in Capture and changed light with a factor 0.05 and extracted the JPG again using IrfanView.

NEF edited1 25.398.558 JPG edited1 6.758.155
NEF edited2 25.415.860 JPG edited2 6.774.813

You see, the first saving makes the difference.
George

[Mike Buckley]

George,

If you're referring to the images of the grave marker, no two of the images have the same number of pixels. To repeat: If the number of pixels is the same, I have never heard of any difference of quality. So, those examples don't help me understand.

Regarding the test, what camera model did you use?

[xpatUSA]

@Ted,
Don't get pissed. When I say I don't know much about colours, thats true. But also a pixel value is just a value, the colour comes from the screen or printer. I don't know what you try to proof with your example.

I am pissed off and who would not be? I asked you not to respond and, true to form, here you are responding with more garbage. Please STOP.

[george013]

George,

If you're referring to the images of the grave marker, no two of the images have the same number of pixels. To repeat: If the number of pixels is the same, I have never heard of any difference of quality. So, those examples don't help me understand.

Regarding the test, what camera model did you use?

If the image is the same in size, so the same amount of pixels. But a different amount of bytes. That's the result of the compression. Some pixels are gathered together and given the same value and stored in a special way in that JPG-file.
If there would be a 1 to 1 relation between pixels on the screen and in the file, than that file would be much larger.The image is 2832x4256, every pixel measures 3 byte, so in total 36.158.976 byte.


The camera is a D700.

George

[Mike Buckley]

If the image is the same in size, so the same amount of pixels. But a different amount of bytes. That's the result of the compression.

If I understand you correctly, the camera is using more compression when it embeds the JPEG than when Capture NX embeds it. Do I have that right? If so, I finally understand. Thanks!

[george013]

If I understand you correctly, the camera is using more compression when it embeds the JPEG than when Capture NX embeds it. Do I have that right? If so, I finally understand. Thanks!

That's it.

George

[Saorsa]

Wouldn't doing the compression on individual color channels be a better solution than working with whole pixels?

[dje]

I have to say that I wasn't aware that the jpeg embedded in the nef file was full raster size, so thanks Mike for pointing this out. I used free software called "jpeg stream extractor" to test this out. Co-incidentally this software is produced by the same crowd that produces Mike's "Photosupreme".

If anyone want's a bit more detail on embedded jpegs in nef files, see this article here. According to this author, there are two embedded jpegs in a nef file, one full raster size and a small one for preview purposes.

Dave

[dje]

Here is the one from the jpeg. The colors are not correct, and the slightly overexposed yarrow is blown out, as is some of the greenery. I am guessing that this latter is from saturation and contrast adjustments made in the JPEG processing.


I first started shooting raw years ago after I captured a very good photo of my son with the wrong white balance, and I couldn't get it entirely fixed. A cousin who teaches photography at the university level told me that it would be a lot easier if I shot raw.

So, while it may not happen often, the fact is that baking in a set of edits by shooting JPEG does in fact sometimes mess things up.

now back to work...

Yes Dan that's what I like about RAW. The ability to adjust WB as a "fresh start" and the extra margin in Highlight protection/Recovery. I also like to apply sharpening from scratch.

Dave

[Mike Buckley]

I have to say that I wasn't aware that the jpeg embedded in the nef file was full raster size

Prior to my helpful discussion with George, I was never aware that the quality of the full-size JPEG embedded by the camera is not the highest quality. I use it only to cull my images before doing any post-processing. Sometimes I display it at 100% but had never noticed an inferior quality. I assume I would notice a difference in quality if I compared the camera-embedded JPEG and the editing software-embedded JPEG side by side in the same window at 100%. Even so, I would have no practical reason to make that comparison.

According to this author, there are two embedded jpegs in a nef file, one full raster size and a small one for preview purposes.

That's true. Some software displays both of the embedded JPEGs.

[dje]

Prior to my helpful discussion with George, I was never aware that the quality of the full-size JPEG embedded by the camera is not the highest quality. I use it only to cull my images before doing any post-processing. Sometimes I display it at 100% but had never noticed an inferior quality. I assume I would notice a difference in quality if I compared the camera-embedded JPEG and the editing software-embedded JPEG side by side in the same window at 100%. Even so, I would have no practical reason to make that comparison.

Yes according to the author I quoted, the full size jpeg is "Basic Quality".

[Saorsa]

I just looked into the .NEF format. It does seem to be TIFF based and include reduced quality JPEG images. You can find it at http://lclevy.free.fr/nef/ .

[george013]

Prior to my helpful discussion with George, I was never aware that the quality of the full-size JPEG embedded by the camera is not the highest quality. I use it only to cull my images before doing any post-processing. Sometimes I display it at 100% but had never noticed an inferior quality. I assume I would notice a difference in quality if I compared the camera-embedded JPEG and the editing software-embedded JPEG side by side in the same window at 100%. Even so, I would have no practical reason to make that comparison.



That's true. Some software displays both of the embedded JPEGs.

If you print small prints, you probably won't see a difference. But big size prints you will. And with editting.

The small sized embedded jpg is meant for thumbnails, by example in the windows explorer.

George

[Manfred M]

Asking whether we can tell whose images were shot as raw or JPEG is asking the wrong question, IMHO.

Dan - I was just trying to get people to think rather than repeat the mantra of one or the other formats being "better". As I've said before, I know several commercial photographers that shoot virtually 100% JPEG, simply because it results in a faster workflow for them. I'm simply not good enough to rely on getting my JPEG shots to that quality level all the time.

Personally, I usually shoot raw + JPEG, although I've shot just raw or just JPEG; the quality of the jpegs will vary by what I am shooting and the final purpose of my shots. For certain types of shooting (strange lighting conditions, panos, etc.) I shoot RAW only. When I am on the road and am blogging, I will shoot the lowest quality jpegs and post straight out of the camera. Some of these shots have definitely made it to CiC. I will do custom white balance in the field to get the "right" white balance and find these are as good as I can get with raw.

I tend to take the "pragmatic" route; i.e. whatever works with the least amount of effort. That also means I do use raw at times it is not strictly necessary, just to keep my RAW import skills up.

I clearly remember only one instance where I had to rely on my raw file to rescue a badly overexposed shot. For me, it can be a bit like insurance; I shoot raw hoping that I won't need to use it.

[DanK]

Dan - I was just trying to get people to think rather than repeat the mantra of one or the other formats being "better". As I've said before, I know several commercial photographers that shoot virtually 100% JPEG, simply because it results in a faster workflow for them. I'm simply not good enough to rely on getting my JPEG shots to that quality level all the time.

Personally, I usually shoot raw + JPEG, although I've shot just raw or just JPEG; the quality of the jpegs will vary by what I am shooting and the final purpose of my shots. For certain types of shooting (strange lighting conditions, panos, etc.) I shoot RAW only. When I am on the road and am blogging, I will shoot the lowest quality jpegs and post straight out of the camera. Some of these shots have definitely made it to CiC. I will do custom white balance in the field to get the "right" white balance and find these are as good as I can get with raw.

I tend to take the "pragmatic" route; i.e. whatever works with the least amount of effort. That also means I do use raw at times it is not strictly necessary, just to keep my RAW import skills up.

I clearly remember only one instance where I had to rely on my raw file to rescue a badly overexposed shot. For me, it can be a bit like insurance; I shoot raw hoping that I won't need to use it.

Manfred,

I'm definitely not good enough to rely on JPEGs.

If I were doing something like blogging from the road, I would probably do as you do. One of my friends is out in the southwest right now doing that, and he is shooting JPEG for that reason.

Personally, I just don't want the loss of control. After a few weeks, I found dealing with raw files simple enough, and I just can't see the advantage--for me--of shooting JPEG most of the time. Also, I am trying to push myself to get more flexible in my postprocessing--for example, doing more of it in selections--and shooting JPEG just reduces flexibility.

Dan

[Brev00]

I shoot raw only but the arguments presented in the article, while clear and factual, are coming from a pro raw shooter. The advantages listed and explained are only important if they have value to the photographer. The values of shooting jpeg are not listed or explained but, I think, operate at a different plane than the values of shooting raw. For just one example, a person may not like working on a computer and would rather produce the desired result in the field. The optimal 14 bit photo file is discarded in order to spend more time shooting. All effort is made to get the desired result in the moment. Is it the ideal maximum realization of the initial raw potential? No. But, that photographer is not playing that game. Different strokes.

[george013]

I just looked into the .NEF format. It does seem to be TIFF based and include reduced quality JPEG images. You can find it at http://lclevy.free.fr/nef/ .

Be carefully with this statement. It looks like one says that if several books have a table of content on page 3, they have something in common.
When I remember well, Nikon has 3 main filestructures, all called NEF.
George