 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
| PhotoKB HomeDiscussion Groups Digital Photography Digital PhotoDSLR CamerasZLR CamerasPoint & Shoot CamerasFilm Photography 35 mmLarge FormatMedium formatDarkroomFilm and LabsOther EquipmentPhoto Technique Nature PhotographyPeople PhotographyTechnique GeneralGeneral Photo Topics General TopicsAustralian PhotographyUK PhotographyDirectoryPhoto Clubs |
Photo Forum / Digital Photography / DSLR Cameras / July 2005Tell me again, what's the big deal about RAW images?
Okay, I've finally installed Photoshop CS2 and now have a program that can manipulate the RAW image format from my Nikon D70. In a completely uncontrolled test (I'll do a formal shoot-off later) I took a few pictures of flowers in the garden using the RAW format for recording and imported them into Photoshop. At least with the Mark One Eyeball Sensor, I didn't see anything different than I would have expected to see if I had used the JPEG-Fine mode I've been using, other than files that are twice as big (6MB vs 3MB). Is there something I'm supposed to be able to do with the RAW format images that I can't do with the JPEGs? Is the resolution better? What? TIA Norm > Is there something I'm supposed to be able to do with the RAW format > images > that I can't do with the JPEGs? Easier to fix white balance and you avoid in-camera algorithms that might not always be as good as what you can do in post-processing. > Is the resolution better? Dynamic range is better. Check this: http://www.digitalphotobook.net/tutorials/camera_raw/cameraraw_1.html RAW takes time and it might not be worth it in your case. It offers no huge advantages for most shots. >> Is there something I'm supposed to be able to do with the RAW format >> images [quoted text clipped - 12 lines] > RAW takes time and it might not be worth it in your case. It offers no > huge advantages for most shots. Under "difficult" conditions you should probably use it though. > > Is there something I'm supposed to be able to do with the RAW format > > images > > that I can't do with the JPEGs? You're saying that I can manipulate the color balance better in the raw format than I can by using levels and curves on the JPEG in Photoshop, right? > Easier to fix white balance and you avoid in-camera algorithms that might > not always be as good as what you can do in post-processing. > > > Is the resolution better? > > Dynamic range is better. Since the number of bits per pixel is the same in both, there's got to be something else going on then, right? I'm guessing but I think what you're implying is that the camera, to produce the JPEG, assigns binary values to the various brightness levels in a non-linear and non-uniform way, right? If that's the case, then the RAW format more closely resembles the range of brightness in the original image but since the number of bits per pixel is the same (at least think it is), then the brightness seen on the screen has to have the same range, doesn't it? Thanks Norm >> > Is there something I'm supposed to be able to do with the RAW format >> > images [quoted text clipped - 28 lines] > Thanks > Norm Actually the number of bits per pixel is NOT the same. RAW = 16 bits, JPEG = 8 bits. JPEG compresses the dynamic range dramatically. [..] > >> Dynamic range is better. > > [quoted text clipped - 12 lines] > Actually the number of bits per pixel is NOT the same. RAW = 16 bits, > JPEG = 8 bits. JPEG compresses the dynamic range dramatically. That's not exactly true. Most RAW file formats are 12 bits per channel (bpc), and typically what you'd do is import that into Photoshop as 16 bpc, and then save as a TIFF or JPEG or whatever, at however many bpc you need. The advantage is that the difference between 12 bpc and 8 bpc isn't automatically lost in JPEG compression, and if you end up making a JPEG, you can correct for loss of dynamic range and control the compression yourself. Some folks might not cherish the idea of going through this process in order to tweak the hell out of the image this way, and for those people the JPEG fine setting is just that: fine. [] > Actually the number of bits per pixel is NOT the same. RAW = 16 > bits, JPEG = 8 bits. JPEG compresses the dynamic range dramatically. Not quite true - the JPEG is gamma corrected, meaning that the dynamic range is similar to that of the sensor, but the luminance steps in the JPEG are non-linear. David In message <PkMze.66070$G8.63663@text.news.blueyonder.co.uk>, "David J Taylor" <david-taylor@blueyonder.co.not-this-bit.nor-this-part.uk.invalid> wrote: >[] >> Actually the number of bits per pixel is NOT the same. RAW = 16 [quoted text clipped - 3 lines] >range is similar to that of the sensor, but the luminance steps in the >JPEG are non-linear. In-camera JPEGs typically either clip away a good deal of highlight range from the RAW, or may compress part of the range so that they are good as-is in the output, but are not particularly editable in that range due to posterization.  Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< >Actually the number of bits per pixel is NOT the same. RAW = 16 bits, JPEG >= 8 bits. JPEG compresses the dynamic range dramatically. I *wish* RAW were 16 bits. It is generally 12 bits. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< > >Actually the number of bits per pixel is NOT the same. RAW = 16 bits, JPEG > >= 8 bits. JPEG compresses the dynamic range dramatically. > > I *wish* RAW were 16 bits. It is generally 12 bits. But 12 bits gives you 16 times the range that 8 bits does. That's a range of 4096:1 which covers a range of 12 f-stops (or 12 EV brightness) which is, if not huge, then certainly enough for almost all but the most demanding work. How many color (print) films can match that? I can tell you that at least for prints in the 4x6" through 8x10" range that even the 8-bits you get with a JPEG format is enough to garner "WOW" from most people if applied properly. Museum quality? No way! But then I'd doubt that even the 20-odd megapixel backs for Hasselblads can do that yet. Right now, if I think I'm going to want to produce a "keeper" print of some subject, I'll switch from the D70 to the N90 which is usually loaded with Kodachrome these days. Norm >> I *wish* RAW were 16 bits. It is generally 12 bits. > > But 12 bits gives you 16 times the range that 8 bits does. That's a range > of 4096:1 which covers a range of 12 f-stops (or 12 EV brightness) which is, > if not huge, then certainly enough for almost all but the most demanding > work. How many color (print) films can match that? That's not the point -- the point is that each successively darker stop of range uses half the number of levels, so those lowest (darkest) stops in the range lose most of their detail to posterization. So you really want several more stops of range than you actually intend to use, so that you can clip off the dark end of the range without losing anything you actually wanted to keep. SignatureJeremy | jeremy@exit109.com >>> I *wish* RAW were 16 bits. It is generally 12 bits. >> [quoted text clipped - 11 lines] > can clip off the dark end of the range without losing anything you > actually wanted to keep. I thought it was more that you gain precision in the colors, not more dynamic range in f-stops. The reason the in camera jpegs have lower dynamic range is because the contrast level is fixed and they have to fit the shot into the set contrast range. I've not noticed any dynamic range clipping in images -developed- from RAW files as 8 bit rather than 16 bit, but have seen some posterization when editing the color of 8 bit images, especially if using some of the larger color spaces like aRGB or proRGB. SignatureStacey > I thought it was more that you gain precision in the colors, not more > dynamic range in f-stops. The two go hand-in-hand. Because each successively darker stop has half the tonal values available to it, there comes a point where there is so little precision that the recorded data is not very useful, despite the fact that it is there. When an entire stop of brightness has only 4 different values to represent it, you're not going to get much out of it, even if it is clean and noise-free. Getting more dynamic range is also a benefit, and is not the same thing as the range in the gamma-corrected 8-bit final output since you could process the 16-bit original to use the greater range in different ways. > I've not noticed any dynamic range clipping in images -developed- from > RAW files as 8 bit rather than 16 bit, but have seen some posterization > when editing the color of 8 bit images, There is no problem with 8-bit for final output. Editing color in 8-bit, however, can cause significant loss. But, if you could capture (say) 15 stops of dynamic range in your original RAW file, you'd be at an advantage. You know from the start that you can't actually represent them all accurately (without compression) in an 8-bit final output, but that's not the point -- you may not need *all* of them, but you may need (for example) two ranges that are 10 stops apart in brightness, and be willing to compress the midtones because they aren't as important to your image. Or you may have vast areas of shadow detail that you need recorded perfectly, but not need much "special" treatment in your highlights. So, there would be real benefits all around in having true 16-bit capture of a wide dynamic range. SignatureJeremy | jeremy@exit109.com >I thought it was more that you gain precision in the colors, not more >dynamic range in f-stops. Even if you don't gain any dynamic range at the highlight end (in a particular number system), you do at the shadow end, ignoring noise and amplification and digitization quality, with a higher bit depth. Of course, in the real world, there are limitations. Most DSLRs are limited in dynamic range at ISO 100 by the 12-bit digitization; not the sensor noise. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< >>> I *wish* RAW were 16 bits. It is generally 12 bits. >> [quoted text clipped - 8 lines] >more stops of range than you actually intend to use, so that you can clip off >the dark end of the range without losing anything you actually wanted to keep. Or just keep it dark. Clipping always happens at the high end, but clipping at the bottom is kind of optional. It's not like east/west, or top/bottom. You can always have more illumination; you can not have less than zero, unless you are running a photon deficit! Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< > Clipping always happens at the high end, but clipping at the bottom is > kind of optional. But that is exactly what they do with the in camera jpegs, they clip the top and bottom to fit within the contrast range set. IMHO the sensors aren't good enough right now to worry too much about 12bit being an issue. SignatureStacey >> Clipping always happens at the high end, but clipping at the bottom is >> kind of optional. > > But that is exactly what they do with the in camera jpegs, they clip the top > and bottom to fit within the contrast range set. Right. Now, what if you don't want it to clip at the top, but don't care as much about the bottom? Or vice-versa? Or what if you really do care about both, but aren't concerned about the midtones? The "default" algorithm is a general case and can't take into account the needs of a particular image. > IMHO the sensors aren't good enough right now to worry too much about > 12bit being an issue. On the contrary, their capabilities (at least at low ISO) seem to be limited by the 12-bit data representation rather than the sensor itself. I say that not knowing the first thing about A/D converter design, though, so it's really easy for me to say "use a 16 bit conversion" when I have no idea what technical barriers prevent that from happening. SignatureJeremy | jeremy@exit109.com >On the contrary, their capabilities (at least at low ISO) seem to be >limited by the 12-bit data representation rather than the sensor itself. >I say that not knowing the first thing about A/D converter design, >though, so it's really easy for me to say "use a 16 bit conversion" >when I have no idea what technical barriers prevent that from happening. The answer is not in a black box; just compare a high-ISO shot with shadow details to the same f-stop and shutter speed at the lowest ISO. If you can see anything extra in the high-ISO shot, the sensor is being under-digitized at ISO 100. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< >>> Clipping always happens at the high end, but clipping at the bottom is >>> kind of optional. [quoted text clipped - 9 lines] > a general case and can't take into account the needs of a particular > image. Which is why we shoot RAW! :-) But you already knew that part. SignatureStacey >> Clipping always happens at the high end, but clipping at the bottom is >> kind of optional. > >But that is exactly what they do with the in camera jpegs, they clip the top >and bottom to fit within the contrast range set. IMHO the sensors aren't >good enough right now to worry too much about 12bit being an issue. It is important when scanning film. >> Clipping always happens at the high end, but clipping at the bottom is >> kind of optional. >But that is exactly what they do with the in camera jpegs, they clip the top >and bottom to fit within the contrast range set. They may clip the bottom, but it is not necessary, to achieve the mapping. The deepest shadows are often clipped or sharply compressed to hide noise in the shadows, not to make it fit, per se. The histogram is not symmetrical; There is nothing to the left of zero; there are potential highlights to the right of the histogram's highest level. >IMHO the sensors aren't >good enough right now to worry too much about 12bit being an issue. Nope. ISO 100 dynamic range in DSLRs is limited by the 12-bit digitization, not the noise. I estimate about 2 or 3 more highly-usable stops with finer digitization. This is easily proven by taking a shot at the camera's highest ISO, a shot that has shadow detail, and then leave the shutter speed and the f-stop right where they are, and taking the same shot with the ISO reduced to the camera's lowest. Bring them both to the same output tonal levels in the RAW conversion, and see how much more detail the high-ISO image has. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< > But 12 bits gives you 16 times the range that 8 bits does. That's a range > of 4096:1 which covers a range of 12 f-stops (or 12 EV brightness) which > is, if not huge, then certainly enough for almost all but the most > demanding > work. How many color (print) films can match that? ???? The dynamic range lost isn't due to 8bit vs 16 bit. It's because the in camera jpeg's dynamic range has been compressed to fit within the contrast setting you have selected. * bits will nromally work fine is used in a smaller color space like sRGB, when you go to the larger color spaces, the gaps between the different colors becomes larger and you might begin to see posterization when you start editing the color balance etc. SignatureStacey In message <9IWze.1110650$w62.292513@bgtnsc05-news.ops.worldnet.att.net>, >> I *wish* RAW were 16 bits. It is generally 12 bits. >But 12 bits gives you 16 times the range that 8 bits does. The number of bits have more to do with precision than range. 12 bits of linear precision, as used in RAW, is actually less precise in the deep shadows than 8 bits of gamma-adjusted data, as you have in typical 8-bit graphics files, in the shadows. >That's a range >of 4096:1 which covers a range of 12 f-stops (or 12 EV brightness) which is, >if not huge, then certainly enough for almost all but the most demanding >work. It's not that simple; first of all, in a typical exposure, not all 4096 values are even used. On Canon 10D, for instance, the numbers 125 to about 1200 are all that are used for the 0 to 255 in the red channel of a JPEG, with normal contrast and daylight white balance. Secondly, 11 stops of digital numbers does not mean that each stop has the same level of usability, even if noise is ignored. You need several different linear levels, at least, to have a reasonable signal in the shadows. The sensors on most current DSLRs have more dynamic range than can be rendered in 12 bits, and you can see this in the shadows of ISO 100 images; they are clearly posterized, and if you boost the shadows, they are ugly as hell, and are inferior to shooting the image with the same absolute exposure at a higher ISO. If ISO 100 had 16 bits of high-quality digitization, it would revolutionize the way most people use exposure and ISO; you could leave the camera set to "sunny f/16" or f/11 or whatever, in manual mode, and still get a decent image if you pointed the camera into the shade. Right now, you'd boost it in software and it would get ugly, because it doesn't have enough bit depth. >How many color (print) films can match that? I can tell you that at >least for prints in the 4x6" through 8x10" range that even the 8-bits you [quoted text clipped - 3 lines] >going to want to produce a "keeper" print of some subject, I'll switch from >the D70 to the N90 which is usually loaded with Kodachrome these days. It is time to go to sleep, but I hope you see from the rest of my post that your last paragraph here is not relevant to the issue, IMO. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< >I'm guessing but I think what you're implying is that the camera, to produce >the JPEG, assigns binary values to the various brightness levels in a [quoted text clipped - 4 lines] >the same (at least think it is), then the brightness seen on the screen has >to have the same range, doesn't it? If you're comfortable with the processor in your camera making all the decisions on saturation, exposure, white balance, sharpening, luminance smoothing, and noise reduction then you should probably shoot jpegs and not worry about it. If you really want to learn more about the advantages of shooting raw you should probably download free from Adobe: http://www.adobe.com/digitalimag/pdfs/understanding_digitalrawcapture.pdf http://www.adobe.com/digitalimag/pdfs/ps_workflow_sec1.pdf There are 4 parts to the workflow article, just change the section numbers to get all of them. > >I'm guessing but I think what you're implying is that the camera, to produce > >the JPEG, assigns binary values to the various brightness levels in a [quoted text clipped - 18 lines] > There are 4 parts to the workflow article, just change the section > numbers to get all of them. These look like some very interesting documents and I'll study them before I ask any more questions. Thanks Norm Less than 2 minutes got me very good results in curves only. Yes, I worked on the jpeg and compression hasn't helped much, as the histogram shows. Without see the cup itself it's hard to know what looks best, but I'm not so sure RAW is a big deal for colour casts. http://www.theimageplane.net/postimages/cup.jpg >Okay, I've finally installed Photoshop CS2 and now have a program that can >manipulate the RAW image format from my Nikon D70. [quoted text clipped - 8 lines] >Is there something I'm supposed to be able to do with the RAW format images >that I can't do with the JPEGs? I suggest you get and study: _Real World Camera Raw with Adobe Photoshop CS2_ by Bruce Fraser. You might want to read this also: http://www.informit.com/articles/article.asp?p=388270&rl=1 It is the only way to start to work with your digital images if you are serious about quality. ********************************************************* "I have been a witness, and these pictures are my testimony. The events I have recorded should not be forgotten and must not be repeated." -James Nachtwey- http://www.jamesnachtwey.com/ If you like to spend lots of time to get your image perfect in your eyes, RAW is for you. If you just want excellent images, high quality JPEGS are fine. If you showed an expert 2 different images, he would be guessing to say which was RAW, Don > If you like to spend lots of time to get your image perfect in your > eyes, RAW is for you. If you just want excellent images, high quality > JPEGS are fine. > If you showed an expert 2 different images, he would be guessing to say > which was RAW, I usually don't spend much more than 30 seconds (if that long) importing a RAW image into PhotoShop. Considering the control I have, the results are worth it. >If you showed an expert 2 different images, he would be guessing to say >which was RAW, Not if there were very bright highlights clipped away by the RAW. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< >If you like to spend lots of time to get your image perfect in your >eyes, RAW is for you. If you just want excellent images, high quality >JPEGS are fine. >If you showed an expert 2 different images, he would be guessing to say >which was RAW, That's only true if the JPG was shot with the optimal settings for (foremost among other things) contrast and white balance. With JPG it's vital you choose the right white balance setting at shooting, since there is little you can do to correct later without skewing the pic horribly. Similarly, if you don't get the contrast right, you can end up with shot-out highlights and no detail in the dark bits. I find that trying to keep all these variables in mind while shooting makes JPG a true pain-in-the-arse. I already have enough to think about with composition, f/shutter and zoom/framing. If you shoot raw, it's an advantage to have a reference white balance shot, but it isn't necessary. The contrast setting also doesn't matter a bit, since you can change all those things after the fact for best performance. It frees up your mind and time during shooting to capture the subject at the right instant. That's the primary reason to choose RAW, IMO. SignatureKen Tough >Okay, I've finally installed Photoshop CS2 and now have a program that can >manipulate the RAW image format from my Nikon D70. [quoted text clipped - 5 lines] >I had used the JPEG-Fine mode I've been using, other than files that are >twice as big (6MB vs 3MB). In other words, they both look as average as each other. >Is there something I'm supposed to be able to do with the RAW format images >that I can't do with the JPEGs? Yes, correct / improve a number of aspects about them during the import process. >Is the resolution better? No. >What? So, you are telling me that: The exposure was spot on, and you didn't need to increase/decrease the exposure or shadows slider to give a more pleasing/stunning image. The color balance was perfect, and it didn't look any better slightly cooler or warmer than the camera's choice. The lens amazingly showed no vignetting that you wanted to eliminate or even enhance for artistic reasons. The saturation was spot on. The sharpness, luminance smoothing and color noise reduction was exactly correct. ....because, I don't believe you ;-) Go back, twiddle the sliders and if you still think the JPEG looks as good or better, switch to shooting JPEG. Also, be certain that you are *not* using the Nikon RAW importer inside Photoshop, it's junk. SignatureOwamanga! http://www.pbase.com/owamanga [..] > >Is the resolution better? > > No. The resolution of the RAW file itself isn't any better or worse in terms of pixel count, but you can interpolate a larger image very easily during the import process. This enlargement will look much, much, (that is: *MUCH*) better than an enlargment of a JPEG. Lossy compression is the enemy. I wish camera manufacturers would start allowing their cameras to output PNG. In fact, I wish I could run linux on my DSLR. O-Well. :-) [] > Lossy compression is the enemy. I wish camera manufacturers would > start allowing their cameras to output PNG. In fact, I wish I could > run linux on my DSLR. O-Well. :-) Perhaps you could run Linux if your camera was built into a PDA. I'm sure it would be much better than any DSLR! <G> David > Is there something I'm supposed to be able to do with the RAW format > images that I can't do with the JPEGs? If your picture is exactly as you want it, then jpeg is fine. But the jpeg is 'frozen'. The 12-bit values are white balanced, exposure adjusted, sharpened, saturated, etc. and then whacked down to 8-bits. Any changes you attempt to make to it will lower the quality of the result. The NEF lets you make much more dramatic changes to the picture without losing quality. If you accidentally screw up a great shot and have no easy way to go back and take the shot again the NEF gives you much more leeway in saving it. It's not a panacea, and the resulting picture won't look as good as if you'd taken the picture properly, but it'll certainly look better than what you could get out of the jpeg. The D-Lighting/boost shadow stuff is especially effective. Since storage is cheap now and the D70 doesn't really take any longer to process NEFs than JPGs I shoot everything in NEF and occasionally it saves me. YMMV! > Okay, I've finally installed Photoshop CS2 and now have a program that can > manipulate the RAW image format from my Nikon D70. > In a completely uncontrolled test (I'll do a formal shoot-off later) > I took a few pictures of flowers in the garden using the RAW format > for recording and imported them into Photoshop. At least with the > Mark One Eyeball Sensor, I didn't see anything different than I > would have expected to see if I had used the JPEG-Fine mode I've > been using, other than files that are twice as big (6MB vs 3MB). > Is there something I'm supposed to be able to do with the RAW format > images that I can't do with the JPEGs? The colour gamut of your camera (and your Mark One Eyeball) is far greater than that of your PC's monitor or your printer. So, when displaying a digital image it's necessary to map a large colour space onto a smaller colour space. This process is known as colour rendering". (The colour spaces used in camera are usually sRGB or Adobe RGB. Adobe RGB is a larger space so it can record more colours, but it's still a lot smaller than that of the camera.) If you use JPEG, the colour rendering is done by your camera at the point of taking the photograph, without any control from you. The result of this can be a photograph of oranges and tomatoes that appear to be the same colour, because both are outside the colour gamut used in the JPEG. If you had used RAW, the oranges and tomatoes would have been recorded as being different colours, and you could have adjusted the image later to make them look distinct. If you use JPEG there's nothing you can do: the different colours have been lost. It might well be that you got lucky with the flowers in your garden -- perhaps all of them were within the colour gamut of the colour space used in your JPEGs. But that's pretty unlikely really, as flowers are wildly colourful, especially in direct sunlight. Andrew. >Is there something I'm supposed to be able to do with the RAW format images >that I can't do with the JPEGs? Only if you have to change white balance, apply curves, get more highlights or shadows, etc. >Is the resolution better? Technically, no, but you can get more detail sometimes if you process correctly. >What? RAW makes it possible to make adjustments in one step (doing them twice or more increases posterization), and get more dynamic range out of the shot. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2009 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.