Why shoot RAW?

Started Jun 14, 2019 | Discussions thread
Thomas A Anderson Regular Member • Posts: 378
Myths persist around JPEG and RAW.
7

When this thread started I began making a list of all the inaccurate and incorrect claims/arguments being made about both JPEG and RAW files. Just recently I saw a video by Tony Northrup that casually stated one of these misleading claims. For the record, I don’t think he was intentionally misleading, but Tony tends to make some fairly generic, blanket statements that are very unhelpful only to come back later and claim “but this is what I really meant.” A great example is his “ISO is Fake” video https://www.youtube.com/watch?v=QVuI89YWAsw. He got a lot of responses, one of the best here https://www.youtube.com/watch?v=hwRsWomRzVQ. Then he came back with https://www.youtube.com/watch?v=dnB4NvIBlbQ. In the end, he tries to explain and justify why he made incorrect and misleading claims that were either completely useless or absolutely wrong.

The video that got me thinking about this is here https://www.youtube.com/watch?v=N9GaVlJ2KlE. Tony and Chelsea have some useful things to say about JPEG files. Unfortunately Tony makes the claim that “you can even actually recover shadows and highlights quite a bit. Not as much as RAW files, but you can still do a lot.” That’s around 20:13. If his definition of “recover” is to bring areas of no contrast or detail (highlights blown or shadows crushed) up or down in order to make contrast or detail more visible, then that is certainly possible: your monitor might not be able to show all the detail in highlights or shadows that may appear to be blown or crushed using the default processing. It may also be the case that an area that looks blown/crushed might not appear as such on a luminance histogram only because one channel is fine while one or two other channels (RGB) ARE in fact blown or crushed. Either way, that definition of “recover” isn’t really useful or even the most common usage. I would say that “recover” is used in reference to RAW files when comparing them to JPEG files: a RAW file contains data that has been compressed away or thrown out during the JPEG encoding process, and if a JPEG ACTUALLY has (not just appears to have based on the limitation of the monitor) a blown highlight or crushed shadow it is very possible that the RAW file still contains that lost information such that editing can bring those details back into the visible range for your monitor. Of course, RAW files can still suffer from the same phenomenon as JPEGs where the luminance histogram looks fine but individual channel histograms reveal one or two of the channels are actually blown/crushed. That doesn’t change fact that “recovery” of information implies that using one method (RAW editing) will allow you to preserve data and make use of that data which would otherwise be lost when using the other method (only having the JPEG straight out of the camera).

Over the years I’ve seen a lot of the same myths about both RAW and JPEG perpetuated all over the internet. I’ve never joined in the “debate” (there wasn’t much in the way of good faith debating going on) because people tend to take the topic very personally. In this case, it might not hurt to help new photographers shed their misconceptions and apprehension about both formats. If some of the posts in this thread are any indication, there aren’t any character limits so this will be as long as it has to be. I’ve been gathering my observations since this thread first started, and there is definitely some overlap in many of these common claims.

RAW Myths

RAW files require an expert to properly edit: Others have pointed this out, but I’ll repeat here for the sake of a complete list: when you open a RAW file in DPP, ACR, Capture One, or another common RAW post processing program it starts you off using all the JPEG image processing tags stored in the RAW file. From that point you have the ability experiment with the file as much as you want without making a single irreversible change: edits are stored in a sidecar file and nothing you ever do to a RAW file will ever be irreversible. Even spending a couple of minutes making universal changes can greatly improve an image. You can absolutely make local adjustments using masks or use the full Photoshop suite of tools. I don’t think images should be manipulated extensively since I’m prefer being as faithful to reality as possible, so if it takes me any more than ten minutes to edit a file I intend to print then I’m having a bad day.

If you want any help editing RAW files people in this forum and others will help you. Send me a private message and I’ll give you a tutorial that will have you creating images better than the JPEG in about thirty minutes.

RAW files have to be edited from a completely unprocessed file: As the previous point said, your starting point looks exactly like the original JPEG would look. If you shoot RAW+JPEG this is very easy to confirm.

RAW files have to be saved as another lossless format in order to save any post-processing edits: Depending on the file type you’re using, ACR stores changes either in a sidecar XMP file, a database file, or within the file itself. I usually use the XMP so that if I want to instantly undo all of my edits I can just delete that file. You also have the option of converting to DNG which stores edits within the file by default (can be changed to XMP sidecar) or choosing a TIFF or PSD based on your preferences. It doesn’t take long to get used to this and choose your preferences. It is absolutely not required to ever change the RAW format in order to retain edits.

RAW has to be saved as a JPEG for any normal person to be able to see the image: JPEG is not a self-extracting file as some believe or at least imply. JPEG was developed about 25 years ago around the time the internet started moving into the commercial sphere, and it became the primary file used for viewing and transmitting images because it was small and flexible. It retains the majority of image data that humans are capable of perceiving at its highest quality settings, and can be set to compress more to reduce file size if necessary but at the cost of increasing viewable artifacts. The RAW file is very large and because each camera has different parameters stored within the RAW file, each camera (even if it is a CRW, CR2, or CR3 file extension) requires a different codec to properly decode the image. Sending a JPEG to someone is easier than sending a RAW along with instructions on how to download a RAW codec if their computer doesn’t already have it. This is only because JPEG is a single, open standard that has been around since the dawn of the internet, but almost every camera maker provides codecs to the various operating systems to allow decoding of their files. But BMP, TIFF, PNG and various other formats are also very commonly used and universally supported. JPEG is small and convenient. Every computer will also open a BMP, TIFF, or PNG, but the files can get very large.

RAW should be used only by amateurs: Amateurs can’t adapt quickly and expose correctly so they need to be able to post process, right? Unfortunately amateurs are so busy trying to learn their cameras and photography in general asking them to learn to edit RAW files could be the straw that broke the camel’s back. It can be overwhelming for all but the most motivated and dedicated new photographers. Modern cameras all have a setting that solves this problem: RAW+JPEG. Also, pretending that even the most practiced professional can nail their exposure “correctly” every time is a flawed assumption on two levels: 1) Even professionals in a fast moving environment won’t be able to get every single setting exactly where they want it for that split-second shot; 2) The “correct” exposure assumes that the final creative intent of a shot is known before it has been taken, which might be true for a studio scene or landscape on a tripod but rarely occurs in real life.

RAW should be used only by professionals who know what they’re doing: Maybe when a 256MB CF card cost $200 and hard drives could only store 100GB if you had the money to afford such a luxury. But RAW editing doesn’t necessarily mean masking areas, applying spot edits, and everything you see experts doing in their Youtube videos. The tools available in RAW editors basically go in the order of most dramatic changes to finest adjustments: start with white balance, exposure, highlight recovery, shadow pushing/pulling, contrast, and the other basic image parameters and then move on to noise reduction, sharpening, lens distortion corrections, etc. Even using a color profiling tool like ColorChecker is pretty easy to do if you’ve got a calibrated monitor and really want extremely accurate colors and white balance.

RAW files should only be used when you know it’s going to be a major event or shot of a lifetime: I’ve been shooting snapshots when suddenly an amazing shot presents itself. I’ve been at family birthday parties and captured images that people wanted enlarged to poster size because an amazing moment was captured. You never know when the shot of your life will appear before your eyes. If you’re not going out to take important, once in a lifetime images then why are you taking pictures? Do you grab your camera and think “I’m really going to take a lot of boring images today!”? I know this sounds like the video posted by Jared Polin responding to Tony Northrup, but I’ve made the same observation when discussing this idea with amateurs and professionals.

You also don’t know what aspects of an amazing image might need some post processing. RAW has all the dynamic range the sensor is able to capture, and while the contrast curve and gamma correction applied by the camera can be very effective, since your monitor or even prints are only capable of showing a limited amount of that dynamic range you may want to choose how shadows, mid-tones, and highlights are balanced in the final image not to mention making changes to white balance, contrast, etc. without artifacts or reducing detail.

RAW is only used by those who don’t want to get exposure right or take care in setting their in-camera processing parameters: I mentioned the myth of the “correct” exposure earlier. One should also know that the histogram on your camera displays the JPEG data only. I don’t know why that’s the case, but it’s true. And JPEG converts the RAW data into a gamma 2.2 corrected sRGB or AdobeRGB colorspace file. That means half of the image data which is contained in the top one stop of dynamic range has to be greatly reduced to make room for shadow detail that humans are much more sensitive to. The next few brightest stops are also reduced and the extra room is taken up by shadows and mid-tones. JPEG only uses 8 bits for each channel (red, green, and blue) compared to 14 bits in each RAW channel, so each bit has to be used to its maximum potential (256 possible values in an 8 bit JPEG color channel, 16,384 values per channel in RAW). The JPEG standard was designed around human perception and efficient use of data in the most compact file possible. And the claim is not that RAW is 64 times better than JPEG (16,384/256=64) or even 262,144 better (using total bit depth for each, 16,384^3/256^3), only that RAW has a lot more useful data.

RAW is almost exclusively used by lazy people to fix errors: If you’re painting with twenty various brushes and fifty colors of oil paints you can do more than using one brush and five colors. And that includes fixing things. The ability to more easily fix things does not imply that now the only reason to use that tool is exclusively to fix things. This idea is counterintuitive but still appeals to our desire appear highly skilled and beyond the need for help: RAW has so much data that fixing small errors becomes much easier, so despite the same feature making creative control much more extensive, the ability to also fix things makes you lazy and a bad photographer for shooting RAW. Creative control and needing to fix your incompetent mistakes have been equated there, and they are not the same thing.

RAW is an ancient format that hasn’t changed: Canon has gone from CRW and CR2 to CR3 and new features have been added. Various RAW files with reduced pixel dimensions, lossy compression in CR3 for smaller files, etc. This is mentioned in the same argument as JPEG files having become greatly improved over the years. One has nothing to do with the other, and neither is actually true.

RAW files won’t be supported in the future: If you look at Capture One or Adobe Camera Raw, you’ll find almost every camera ever made with RAW available listed. I believe there have been some bodies that the manufacturer refused to allow third party support for that eventually were no longer even supported by the manufacturer. I’d say this has happened maybe three times in the last 20 years although the number is probably one.

RAW should always be used no matter what: If you take one image every minute and also own a 512GB memory card, then you can choose to use RAW exclusively. That means you also want to edit pictures in front of a computer, and not all people who enjoy photography also enjoy editing. They are two very different experiences. Like many others, I shoot RAW+JPEG but I also own expensive cameras and have the money for hard drives and memory cards. I also enjoy editing and printing very large photographs. I think editing is very fun, but I don’t edit every single image I take. I will recommend people give it a try, but if they don’t like it then it’s absolutely not a requirement to be a great photographer.

I once knew a professional photographer who shot RAW. He was a paste-eating moron (all due respect), which means RAW is bad: Anyone who makes this argument isn’t being helpful. There are amazing, smart, talented professionals who do dumb things. There are amateurs that do very smart things. If a professional who shoots RAW is amazing at his or her art, does that also mean everyone should shoot RAW? Those people do exist, but not everybody has the same needs or interests.

I once knew an amateur who shot RAW exclusively and was amazed to see how easy it was to edit JPEG files: Amateurs are inexperienced and usually very impressionable. They may shoot RAW exclusively because someone told them to despite having no idea what to do with the files in post. Stories like this don’t have any relevance to what someone else wants or needs.

An unscrupulous liar edited stuff out of his RAW images once, so RAW is only for fraudsters: But you just said that JPEG can be easily edited, too. I’m not sure that someone using a BMW 328i to rob a bank will make me believe that those cars are only good for bank robbing. Again, this is not an argument. As with so many personal accounts, this is an appeal to emotion. Other arguments citing personal accounts appeal to purity or to authority or even fear. When choosing the right tool for my ethical activities I don’t worry about what tool an unethical person used for their nefarious purposes.

JPEG Myths

JPEG images have just as much dynamic range as RAW files: The histogram you see on your camera screen is that of a fully processed JPEG file. The histogram represents all of the data available in the JPEG file. If you take a few shots and blow some highlights using RAW+JPEG you can take a look at the RAW histogram in a program like RawDigger (free trial version if you don’t want to pay) and see the data that the JPEG disposed of during encoding. JPEG is more compact exactly because highlights use a large proportion of the available data but humans are more sensitive to darker tones, so the upper stops of dynamic range are cropped and compressed. Even in the most recent DPR video from Chris and Jordan (https://www.dpreview.com/videos/5816213935/dpreview-tv-photo-lingo-101-a-guide-to-common-photographic-terms starting at 15:27) say about shooting RAW “And if that sounds complicated or you just don’t give a <beep> about dynamic range well then you should probably shoot JPEG.” The JPEG histogram represent all the data that exists in that file and there is no hidden data to recover. Data is recovered in the RAW file because it still contains all of the original data (also discussed below).

JPEG files have information that can be recovered: As I stated previously, the histogram you see on your camera screen is generated based on the JPEG file. If you have the RGB histogram displayed you are looking at 100% of the data available to edit. Any highlights that are almost blown could be pulled back so that detail is more visible. Any shadows close to being completely crushed can be brightened up to reveal some hidden detail. And don’t forget that the luminosity histogram does not indicate if a specific color channel is blows. Luminosity could look fine, but you might still have a color channel oversaturated (common when shooting scenes made of one or very few colors).

JPEG files are better today than they were 10 years ago because of advanced imaging processors: Image processors are much faster today and there’s no doubt they are better at sharpening, noise reduction, and other processor-intensive tasks. However, the biggest reason your JPEG images look better today than they did 15 years ago is because imaging sensors have improved by leaps and bounds. When I first moved to digital there were still discussions about whether film was better than digital. When was the last time you saw one of those discussions anywhere? And think about what your Digic processor does today that it didn’t ten years ago: nearly full-sensor DPAF, subject tracking across an entire sensor of DPAF pixels while shooting images or video, eye detection, 4K video, much higher frame rates of continuous shooting, flicker detection, and probably ten other things I can’t think of right now. It’s easier to reduce noise when a sensor is producing far less noise. A good contrast curve can really make an image pop when it has 13 stops of dynamic range instead of 8.

This claim also implies that at some point imaging processors were so bad that it made sense to shoot RAW and post process using much more powerful computers. Most people who make this claim also say they’ve never shot RAW, or at least very rarely. That means that all of these old JPEG files were of compromised quality and they didn’t care enough to do something about it. Also, I have 15 year old RAW files. I can prove conclusively that this claim is false. If quality improvements were primarily due to processors and not sensors I could take a 15 year old RAW file and create a JPEG with similarly low noise and high dynamic range (or at least apparent dynamic range since I would have to move shadows and highlights around in order to make them visible in the limited dynamic range available within JPEG 8 bit files). It’s not possible. Largely sensor improvements, but a good processor helps.

JPEG files can’t be edited at all without causing artifacts: There’s plenty of editing that can be done to a JPEG to make it look even better without risking sharpening halos, posterization, and other visual anomalies or artifacts. Unfortunately if you want to sharpen, you have to apply that to an image that has already been sharpened. If you shot the JPEG with sharpening set as low as it will go, you’ll have greater sharpening flexibility in post. However, when you’re trying to reduce noise, sharpen, correct color and white balance, or adjust exposure you have less latitude than RAW.

JPEG files have all the information a GOOD photographer would ever require: A good photographer always nails exposure, right? A good photographer can set in-camera processing parameters before taking the shot so that no post-processing is necessary, right? The idea of a “right” exposure is the first stumbling block here. Is the exposure “right” if the histogram shows no blown highlights but is as far to the right as possible (Expose to the Right)? If ETTR is your thing then the histogram will guide your JPEG exposure very well, but your RAW file has another two stops further to the right than the histogram is showing. That not only means that you’re saving more of your highlight data, it also means you’ve brought your shadows another two stops to the right. That means shadows the JPEG exposure would have crushed now have an additional two stops of detail in a good RAW exposure. Noise that would have been obvious has an additional two stops of signal on top of it. That significantly increases the signal to noise ratio in areas where detail would have otherwise been destroyed by noise. If you want detail in your shadows and you don’t like noise, RAW can be a big help.

And here’s the real point: excellent, professional photographers who do any work at all outside of a studio or away from static landscapes with very slowly changing light constantly deal with quickly changing light or subjects wearing clothing of different colors. Even if they believe they nail the exposure at the time, when they review images later they may see something that makes them wish they’d exposed differently. Exposure during fast moving situations where the subject is in motion and the light is changing quickly, especially if it is a high dynamic range situation with a lot of deep shadows and very bright highlights, is extremely difficult. In the end there are many situations where the more flexibility you have in post the better your final image will be. Whether your exposure is perfect or off by a half or whole stop doesn’t matter. Creative control is what matters.

JPEG files aren’t flat and soft like RAW images are: JPEG files certainly are usually colorful, bright, with plenty of pop if you’ve set your camera up to process images that way. Some people also set the profile to more neutral color and less contrast in order to create a different look or maybe they want to do some post-processing and know it’s easier with a flatter image. Almost every single method for opening a RAW file will put an image on the screen that looks exactly as the JPEG would have. You have to go out of your way to open the image data exactly as the sensor recorded it. I’ll mention RawDigger here again, but there are other RAW viewers that ignore all the processing data. This myth demonstrates a misunderstanding of RAW decoding.

JPEG should only be used for capturing large numbers of frames at high rates: Most cameras can shoot bursts of many more JPEG files than RAW files before the buffer gets full. Often the JPEG buffer will never get full, but there will be a wall for RAW. The really expensive, professional cameras shoot very fast for a very long until the memory card is full or the battery dies. This is another reason JPEG exists: RAW files are large and any time you have to move large files very quickly it costs more money. However, if someone doesn’t want to post process RAW files, they don’t have to because JPEG files are great. Fine artists don’t have to care about post processing, enthusiasts may not want to sit at a computer and tweak fifty settings in front of a calibrated monitor (recommended but not required) to get an image juuuuust right, and amateurs taking snapshots may just want to post to Instabook and Facegram or send them via email to old people who know social media is responsible for all the ills in our modern world.

JPEG files are images and RAW files are not: The JPEG file is thought of as a “real” image now because it is almost universally used and every computer on the planet can open it (because they have codecs that will open them, not because a JPEG file opens and displays itself).

I’ve seen experts refer to RAW as not a real image file. Their argument is that the RAW file contains light readings for each of millions of pixels. Also, because processing and gamma conversion hasn’t been applied to that RAW data the pure, unprocessed RAW file looks drab and flat.

RAW files are certainly much more complex than JPEG files. A RAW file isn’t simply decoded and displayed in a single way: the data gathered by the image sensor is stored separately from even the most basic processing. So, if you choose to open a RAW file with a program that does not apply any processing by default like RawDigger, you can see exactly how the sensor captured data: the entire right half of the histogram is a single stop of data, which is to say a full half of the data is contained in the brightest stop of highlights (that is not how the human eye responds to light, so the RAW data is converted to standard color space which uses a gamma 2.2 correction, giving more room on the histogram to the shadows and removing highlight data to make that possible). You can use Microsoft codecs that you have to download yourself or open a RAW file in ACR, DPP, Capture One, and many other programs if you want to view the image as processed according to the stored parameter.

A JPEG file is no more an image file than any RAW file. Just because the RAW file stores processing parameters separate from the image data means that decoding the image is not already predetermined. That’s why JPEG is more universally supported: the image is encoded AFTER colorspace, sharpening, contrast, etc. are applied. A RAW file is encoded BEFORE those settings are applied, but the image is still the vast majority of that data. There’s an image, a full-size JPEG used for in-camera viewing, and all processing parameters required to recreate that JPEG precisely (as a starting point) when opening the RAW file. There’s a word for a bunch of data describing a matrix of pixels: image. Both files are 1’s and 0’s — there is no image impressed upon your hard drive.

JPEG files should only be used by experts because they can get exposure and in-camera processing parameters perfect no matter what the situation: This goes back to the argument about RAW being used by the lazy or only to fix errors. It also creates a kind of worshipful mindset of what must be a professional’s almost superheroesque superpowers. That mindset can intimidate new photographers, and it isn’t a reasonable goal. It’s not a valid argument.

Many publications only want their professional photographers to submit JPEG files: Are you an employee of this publication? Many publications have also cleared out their offices of full time staff photographers in favor of reporters using their smartphones. Publications have a lot of concerns these days, and the internet giants lying to advertisers in order to put print publications out of business is a much greater concern than whether they have two dozen professionals to cover all the beats. The universe of publications covers a lot of ground and it is irrelevant what each one requires unless you are working for them.

Post (hide subjects) Posted by
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow