FYI, the k20D sensor is not designed by PENTAX, it was conceived by an israeli company bought by Samsung sometime ago called originally transchip and now Samsung Semiconductor Israel R&D Center (SIRC).
Quit a feat, by the way, judging by its performance.
There are many aspects of raw that you have left out/misinterpreted.
Just to name a few: To begin, jpeg can never be better than raw, because in camera jpeg is created from raw. In camera jpeg *can* only be *better* than jpeg converted from a software converter.
Jpeg will allways be inferior, as it is only 8Bit and compressed.
But in camera jpegs are also limited to a certain color space (Mainly sRGB) and white balance. Alone for archiving, raw is better. In the future, new ways to develop bayer arrays or better noise removal softare will be found and displays and prints will have a better quality such as improved color gammut or dynamic rage. Other reasons to use raw have allready been pointed out by other users. If you want more reasons, look at other boards.
If you are a serious photographer, shoot Raw. Raw+jpeg is only meaningfull if you instantly need jpges, as full size jpegs are included in the raw files.
But 99% of the pictures made are rubbisch anyway. It's like playing guitar. If you are a bad musican, it doesn't really make a difference if you have a fender guitar or a 0815 cheapo guitar. The better you are, the sooner you can utilise better equipment.
When I was trying to make the change from film to digital I had a chance to use a top of the line point and shoot. The shutter delay made action shots impossible. Only the DSLR could take the picture almost as fast as the film camera. From the remarks made by friends who have new fixed lens cameras the problem is still here. At the time I made the move to digital, Canon was the best choice because of the lower noise in low light. Now there is little difference between the top models of each brand. The reason to consider a fixed lens is the cost of the DSLR lens. The small sensors demand the finest lens and that means $1000.00 and up, way up each. What do you get for that, outstanding prints. 35mm can not touch a 8 or 10 mp sensor. I used to be a pro and owned a studio. My 10mp Canon XTi has better image quality than my old Hasselblads. This will start a fight but I have tested it myself and that is my opinion. Remember it is the photographer that makes a great photo not the camera.
Article's title is about sensors, but it says lots about P&S vs DSLRs. There are some pieces missing - camera responsiveness and autofocus speed. Which have absolutely no relation to the sensor size but would be nice to mention.
I just don't understand the entire RAW vs JPG argument? If JPG is an end result of processed RAW image how can it possibly be better than RAW? RAW editor gives us the level of control unattainable from processed bitmaps. Even 40D makes mistakes and I have to correct white point or recover highlights. Then when I'm done, I save the result as JPG into my albums, but I keep few years worth of RAW images. Shooting JPG is for people who don't want to spend time to perfect their images, they want instant gratification, they shouldn't bother with DSLRs in the first place IMHO, a high-end point and shoot will be good enough for them.
Thanks for the great series of articles but I think you should reconsider your choice of example image.
The one you are using does test resolution but does not do a good job with detail in shadow area. In fact an overly contrasted image will do great in this test, appearing to have both high resolution and noise because there isn't much shading required in the test image. Most consumer cams, which are over sharpened, saturated, and contrasted out of the box will do great with this image but not with a most photos that involve more colors and shadow areas that actually require detail.
As mentioned above the main reason for RAW is being able to pull recover more information from over exposed or underexposed images if necessary than ever possible with a simple JPG. JPG should be only one solution not your goal. RAW removes all the mostly bad and inconsistent in camera processing as clearly illustrated by all these generic crops displayed. In shooting JPG only, you are stuck with those bad in camera decisions of how much noise to remove, how much tone balance, color saturation, etc. the camera post processes. The worst culprit which immediately deletes a great deal of your detail is the built in noise and moire removal filters on all JPG output in camera processed pictures. You are stuck with that. Why any self respecting photographer with a $1000 DSLR or higher would shoot RAW and count on built in JPG processing is beyond me. There are many nice point and shoot solutions for that. Just push a button and don't worry.
Any self respecting photographer will get a dual card DSLR and shoot RAW on one card, JPEG on second card. That way you can give some results to the client directly after the shoot (the card with the JPEGs) and they can choose what they like and what they don't like. After their selection you process the selected images from RAW with all the bells and whistles.
This way you save time and unnecesary work (processing all the RAW files). That's why you need good in-camera JPEG.
But for sub 1000$ DLSRs which will be bought by folks either commning from P&S or new to digital will almost always shoot in JPEG, check each shot on the LCD and reshoot if possible. So again good JPEG performance will sell the camera for them.
Which goup will turn more profit for the camera maker ? I bet the larger JPEG shooters group.
Point and shoots don't give the level of control even an entry level DSLR can achieve (not to speak about speed).
Right on. RAW images are 12 or 14-bit. Jpeg images are only 8-bit. Those extra bits of information which the camera "throws out" during the conversion process do make a HUGE difference in the quality of the final image. The obviously superior quality of images in RAW format has to be seen to be appreciated, especially in the highlight and shadow areas of a photograph. Even untrained people, not just pixel peepers, can discern the difference. Converting a RAW image to JPEG in order to demonstrate that there is little difference in the formats is self-fulfilling because you've taken a 12 or 14-bit image and downprocessed it to a 8-bit image in order to compare it to another 8-bit image. Naturally, you won't see much of a difference. An (imperfect) analogy would be using Windows XP 32-bit to test the difference between systems with 2 Gig and 8 Gig memory configurations. For obvious reasons, I don't think any self respecting tech geek would do that. If you want to do a fair image comparison, you have to post images in uncompressed 14-bit TIFF or RAW or even DNG formats. However, this would present a challenge for most websites because you'd be dealing with massive image files.
The article makes the point quite clearly IMO that RAW is a great medium for those that want to really tweak the images after shooting. While that's probably true of serious professionals, I can tell you that I have only once bothered to shoot in RAW and after the hassles of spending extra time converting the image for an article I never did that again.
Does that mean RAW is bad? Of course not, but the point is there are many (MANY!) people that just don't need/want that level of control. I personally am far more interested in getting a high-quality image quickly rather than an exceptional quality image after a lot of work.
We did look into trying to use/post the TIF images, but we ran into a variety of issues. File size was obviously a concern (up to 45MB files for TIF), and being able to show images in a browser meant we needed a web format. Our server also doesn't appear to allow direct downloading of TIF files, so we would need to put them in a ZIP wrapper. We may still add a link if there's a desire from enough people, but the main point is that while RAW is far more powerful in a variety of ways, a lot of that power is wasted on anyone short of the prosumer photographer.
Another point mentioned above is the choice of subject for the comparison photos. Remember, we need something that can be compared in future reviews. That pretty much means we want a room set where we can fully control lighting, the objects being photographed, tripod location, etc. Outdoors thus won't work - at least not for apples-to-apples comparisons. If you're a professional, I imagine you already have plenty of sources for camera information; if you're more of a hobbyist and someone looking for better than P&S, hopefully these articles are providing a lot of good information.
The goal was to produce a fair comparison and as staed in the article we do have the TIFF files available. The purpose was never to show there is minimal difference in RAW and JPEG. As we were preparing to post there were concerns that the TIFFs, at 45mb each, might cripple our server as a direct image view "click to see". That is the ONLY reason we converted to Maximum JPEG format directly from the RAW file.
Since there is some interest we will likely produce a ZIP of the TIFFS and create a download link on the RAW examples page.
Another option would be to post a ZIP of the RAW files. The latest Adobe Camera RAW can handle the PEF format in either Photoshop CS3 or Elements 5 and 6. OR we could save in Adobe DNG format and the DNG files could be read in almost any recent Photoshop or Elements - but not other programs. If either of these is a preferred option please let us know in these comments and we will go with what readers want.
A ZIP file of images in RAW or DNG format would be great. I don't think any other review site is doing this. Although I won't download them each time I read an article, I would absolutely look at them VERY carefully if I'm interested in a camera.
For your sample images, have you thought about using a high quality mannequin head (sounds creepy, I know), or silk flowers? Although not perfect, this would be much preferable to the product boxes you’ve been using. This will allow you to precisely control your test protocol while at the same time obtaining useful information about the imaging capabilities of cameras being tested).
Overall, your camera reviews are better than average and if you’re serious about it, why not make your testing as good as possible?
"Serious photo hobbyists will also be facing difficult decisions today and even more so in the near future. The cost of larger and larger sensors has been dropping rapidly; and CMOS sensor development from all the sensor manufacturers is also a factor in lowering costs and increasing resolution. Like it or not Canon and Nikon have already begun segregating their SLR line into full-frame and APS-C sensors. Those who wondered why Sony was introducing mainly full-frame lenses will finally get their answer later this year with Sony's 24.6MP full-frame flagship model."
This exactly makes me a happy film shooter :-) There's one disdvantage to full frame sensors however. They increase demand for full frame lenses and increase the prices for me :-(
Anyway good article. I'd have one comment and one request.
Comment: There was lots of heated discussion about your sample images (there is one again so far). Would be good if you could shoot manual with f/4.0 and whatever shutter reading for the selected ISO but same for all cameras to make the captured EV consistent. This should in theory lead to all images having same brightness. Of course that will vary by camera processing, but at least you get one more point you can compare from the same series of shots.
Request: Can you make an article on the processing path of different camera makes and sensors ? You covered the digital conversion so far, but the A/D part would be nice to have too. F.e how ISO is controlled (analog gain or digital interpolation) etc. Also explanation what a higher bit-depth sensor means (12 vs 16 bit sensor and A/D). I know these things are pretty basic, but this will create some common group which you can reference in the future and avoid stupid questions (well some of them at least).
Does anyone agree that a better object with clearly defined lines and colors should be used instead, for the comparison crops?
Most of the time these mass-produced product boxes have mediocre color matching and well-defined lines so it shouldn't be used as a 100% comparison crop.
The shots also seem to have a substantial amount of chromatic aberration, or maybe just due to pixel peeping, hmmm...
JPEG compression control would be nice on the p&s. The Canon SD1000 has compression artifacts all over, thus making the pictures it takes look worse than a $90 Nikon.
There are two key advantages of RAW that you missed and that could be useful to even people who are slighly interested in improving images.
First, RAW provides 12 or 14 bits/pixel while JPEG only provides 8 bits. This is extremely relevant if you edit at all in Photoshop/Lightroom because any kind of level adjustment (even auto) will cause more color degradation with an 8 bit image than with a 16 bit image (which is what the 12 bit RAW files are generally processed into).
Second, RAW provides a better chance of recovering improperly exposed images by allowing at least 1/2 stop of exposure correction without any penalty and often more without significant issues. With JPEG you simply won't get this.
You are absolutely correct that while JPEG is 24 bits per pixel it is only 8 bits per color channel (red, green, blue). The Pentax K20D is specified as a 14-bit A/D processor.
Photoshop and other processing software, as you point out, normally processes this in 16-bit. However, it really doesn't matter in the end how the program processed it or saved it, what matters is the input bit depth. Many are not aware of this, but Photoshop also processes JPEGS as 16-bit on more powerful computers, but this just speeds up processing because you start with 8-bit and save as 8-bit with JPEG. There is no real advantage processing 12-bit files as 16-bit except processing speed as you don't gain real resolution improvements.
I agree RAW SHOULD allow more dynamic range, but some of the newest models like the Nikon D300 can actually do just as well or better in dynamic range in JPEG. However, in general what you say is true although it is changing as the processing power in DSLRs is improving.
What we really need is a higher bit-width JPEG standard. Let's hope the JPEG standards committee is hard at work on just that.
The third page is there. We have been having server problems today and this article went to post twice when it was scheduled to go live on 5/20. You may have been caught in one of those accidental post cycles. We apologize for the confusion.
Yea. Right after I posted the comment I realized that the entire article was gone; so I figured the article must have not have been ready for prime time.
Very good read, a little of it is above/beyond me, but a good read none the less. Thanks.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
22 Comments
Back to Article
danette57 - Wednesday, June 4, 2008 - link
FYI, the k20D sensor is not designed by PENTAX, it was conceived by an israeli company bought by Samsung sometime ago called originally transchip and now Samsung Semiconductor Israel R&D Center (SIRC).Quit a feat, by the way, judging by its performance.
Daniel
Matt08 - Saturday, May 24, 2008 - link
There are many aspects of raw that you have left out/misinterpreted.Just to name a few: To begin, jpeg can never be better than raw, because in camera jpeg is created from raw. In camera jpeg *can* only be *better* than jpeg converted from a software converter.
Jpeg will allways be inferior, as it is only 8Bit and compressed.
But in camera jpegs are also limited to a certain color space (Mainly sRGB) and white balance. Alone for archiving, raw is better. In the future, new ways to develop bayer arrays or better noise removal softare will be found and displays and prints will have a better quality such as improved color gammut or dynamic rage. Other reasons to use raw have allready been pointed out by other users. If you want more reasons, look at other boards.
If you are a serious photographer, shoot Raw. Raw+jpeg is only meaningfull if you instantly need jpges, as full size jpegs are included in the raw files.
But 99% of the pictures made are rubbisch anyway. It's like playing guitar. If you are a bad musican, it doesn't really make a difference if you have a fender guitar or a 0815 cheapo guitar. The better you are, the sooner you can utilise better equipment.
Regards, Matthias
BillWilliam - Saturday, May 24, 2008 - link
When I was trying to make the change from film to digital I had a chance to use a top of the line point and shoot. The shutter delay made action shots impossible. Only the DSLR could take the picture almost as fast as the film camera. From the remarks made by friends who have new fixed lens cameras the problem is still here. At the time I made the move to digital, Canon was the best choice because of the lower noise in low light. Now there is little difference between the top models of each brand. The reason to consider a fixed lens is the cost of the DSLR lens. The small sensors demand the finest lens and that means $1000.00 and up, way up each. What do you get for that, outstanding prints. 35mm can not touch a 8 or 10 mp sensor. I used to be a pro and owned a studio. My 10mp Canon XTi has better image quality than my old Hasselblads. This will start a fight but I have tested it myself and that is my opinion. Remember it is the photographer that makes a great photo not the camera.tehcook - Friday, May 23, 2008 - link
Article's title is about sensors, but it says lots about P&S vs DSLRs. There are some pieces missing - camera responsiveness and autofocus speed. Which have absolutely no relation to the sensor size but would be nice to mention.Zak - Thursday, May 22, 2008 - link
I just don't understand the entire RAW vs JPG argument? If JPG is an end result of processed RAW image how can it possibly be better than RAW? RAW editor gives us the level of control unattainable from processed bitmaps. Even 40D makes mistakes and I have to correct white point or recover highlights. Then when I'm done, I save the result as JPG into my albums, but I keep few years worth of RAW images. Shooting JPG is for people who don't want to spend time to perfect their images, they want instant gratification, they shouldn't bother with DSLRs in the first place IMHO, a high-end point and shoot will be good enough for them.Z.
Hulk - Tuesday, May 20, 2008 - link
Thanks for the great series of articles but I think you should reconsider your choice of example image.The one you are using does test resolution but does not do a good job with detail in shadow area. In fact an overly contrasted image will do great in this test, appearing to have both high resolution and noise because there isn't much shading required in the test image. Most consumer cams, which are over sharpened, saturated, and contrasted out of the box will do great with this image but not with a most photos that involve more colors and shadow areas that actually require detail.
warezme - Tuesday, May 20, 2008 - link
As mentioned above the main reason for RAW is being able to pull recover more information from over exposed or underexposed images if necessary than ever possible with a simple JPG. JPG should be only one solution not your goal. RAW removes all the mostly bad and inconsistent in camera processing as clearly illustrated by all these generic crops displayed. In shooting JPG only, you are stuck with those bad in camera decisions of how much noise to remove, how much tone balance, color saturation, etc. the camera post processes. The worst culprit which immediately deletes a great deal of your detail is the built in noise and moire removal filters on all JPG output in camera processed pictures. You are stuck with that. Why any self respecting photographer with a $1000 DSLR or higher would shoot RAW and count on built in JPG processing is beyond me. There are many nice point and shoot solutions for that. Just push a button and don't worry.haplo602 - Tuesday, May 20, 2008 - link
Any self respecting photographer will get a dual card DSLR and shoot RAW on one card, JPEG on second card. That way you can give some results to the client directly after the shoot (the card with the JPEGs) and they can choose what they like and what they don't like. After their selection you process the selected images from RAW with all the bells and whistles.This way you save time and unnecesary work (processing all the RAW files). That's why you need good in-camera JPEG.
But for sub 1000$ DLSRs which will be bought by folks either commning from P&S or new to digital will almost always shoot in JPEG, check each shot on the LCD and reshoot if possible. So again good JPEG performance will sell the camera for them.
Which goup will turn more profit for the camera maker ? I bet the larger JPEG shooters group.
Point and shoots don't give the level of control even an entry level DSLR can achieve (not to speak about speed).
pinto4402 - Tuesday, May 20, 2008 - link
Right on. RAW images are 12 or 14-bit. Jpeg images are only 8-bit. Those extra bits of information which the camera "throws out" during the conversion process do make a HUGE difference in the quality of the final image. The obviously superior quality of images in RAW format has to be seen to be appreciated, especially in the highlight and shadow areas of a photograph. Even untrained people, not just pixel peepers, can discern the difference. Converting a RAW image to JPEG in order to demonstrate that there is little difference in the formats is self-fulfilling because you've taken a 12 or 14-bit image and downprocessed it to a 8-bit image in order to compare it to another 8-bit image. Naturally, you won't see much of a difference. An (imperfect) analogy would be using Windows XP 32-bit to test the difference between systems with 2 Gig and 8 Gig memory configurations. For obvious reasons, I don't think any self respecting tech geek would do that. If you want to do a fair image comparison, you have to post images in uncompressed 14-bit TIFF or RAW or even DNG formats. However, this would present a challenge for most websites because you'd be dealing with massive image files.JarredWalton - Tuesday, May 20, 2008 - link
The article makes the point quite clearly IMO that RAW is a great medium for those that want to really tweak the images after shooting. While that's probably true of serious professionals, I can tell you that I have only once bothered to shoot in RAW and after the hassles of spending extra time converting the image for an article I never did that again.Does that mean RAW is bad? Of course not, but the point is there are many (MANY!) people that just don't need/want that level of control. I personally am far more interested in getting a high-quality image quickly rather than an exceptional quality image after a lot of work.
We did look into trying to use/post the TIF images, but we ran into a variety of issues. File size was obviously a concern (up to 45MB files for TIF), and being able to show images in a browser meant we needed a web format. Our server also doesn't appear to allow direct downloading of TIF files, so we would need to put them in a ZIP wrapper. We may still add a link if there's a desire from enough people, but the main point is that while RAW is far more powerful in a variety of ways, a lot of that power is wasted on anyone short of the prosumer photographer.
Another point mentioned above is the choice of subject for the comparison photos. Remember, we need something that can be compared in future reviews. That pretty much means we want a room set where we can fully control lighting, the objects being photographed, tripod location, etc. Outdoors thus won't work - at least not for apples-to-apples comparisons. If you're a professional, I imagine you already have plenty of sources for camera information; if you're more of a hobbyist and someone looking for better than P&S, hopefully these articles are providing a lot of good information.
Wesley Fink - Tuesday, May 20, 2008 - link
The goal was to produce a fair comparison and as staed in the article we do have the TIFF files available. The purpose was never to show there is minimal difference in RAW and JPEG. As we were preparing to post there were concerns that the TIFFs, at 45mb each, might cripple our server as a direct image view "click to see". That is the ONLY reason we converted to Maximum JPEG format directly from the RAW file.Since there is some interest we will likely produce a ZIP of the TIFFS and create a download link on the RAW examples page.
Wesley Fink - Tuesday, May 20, 2008 - link
Another option would be to post a ZIP of the RAW files. The latest Adobe Camera RAW can handle the PEF format in either Photoshop CS3 or Elements 5 and 6. OR we could save in Adobe DNG format and the DNG files could be read in almost any recent Photoshop or Elements - but not other programs. If either of these is a preferred option please let us know in these comments and we will go with what readers want.pinto4402 - Tuesday, May 20, 2008 - link
A ZIP file of images in RAW or DNG format would be great. I don't think any other review site is doing this. Although I won't download them each time I read an article, I would absolutely look at them VERY carefully if I'm interested in a camera.pinto4402 - Tuesday, May 20, 2008 - link
For your sample images, have you thought about using a high quality mannequin head (sounds creepy, I know), or silk flowers? Although not perfect, this would be much preferable to the product boxes you’ve been using. This will allow you to precisely control your test protocol while at the same time obtaining useful information about the imaging capabilities of cameras being tested).Overall, your camera reviews are better than average and if you’re serious about it, why not make your testing as good as possible?
haplo602 - Tuesday, May 20, 2008 - link
"Serious photo hobbyists will also be facing difficult decisions today and even more so in the near future. The cost of larger and larger sensors has been dropping rapidly; and CMOS sensor development from all the sensor manufacturers is also a factor in lowering costs and increasing resolution. Like it or not Canon and Nikon have already begun segregating their SLR line into full-frame and APS-C sensors. Those who wondered why Sony was introducing mainly full-frame lenses will finally get their answer later this year with Sony's 24.6MP full-frame flagship model."This exactly makes me a happy film shooter :-) There's one disdvantage to full frame sensors however. They increase demand for full frame lenses and increase the prices for me :-(
Anyway good article. I'd have one comment and one request.
Comment: There was lots of heated discussion about your sample images (there is one again so far). Would be good if you could shoot manual with f/4.0 and whatever shutter reading for the selected ISO but same for all cameras to make the captured EV consistent. This should in theory lead to all images having same brightness. Of course that will vary by camera processing, but at least you get one more point you can compare from the same series of shots.
Request: Can you make an article on the processing path of different camera makes and sensors ? You covered the digital conversion so far, but the A/D part would be nice to have too. F.e how ISO is controlled (analog gain or digital interpolation) etc. Also explanation what a higher bit-depth sensor means (12 vs 16 bit sensor and A/D). I know these things are pretty basic, but this will create some common group which you can reference in the future and avoid stupid questions (well some of them at least).
7thSerapHim - Tuesday, May 20, 2008 - link
Does anyone agree that a better object with clearly defined lines and colors should be used instead, for the comparison crops?Most of the time these mass-produced product boxes have mediocre color matching and well-defined lines so it shouldn't be used as a 100% comparison crop.
The shots also seem to have a substantial amount of chromatic aberration, or maybe just due to pixel peeping, hmmm...
sprockkets - Tuesday, May 20, 2008 - link
JPEG compression control would be nice on the p&s. The Canon SD1000 has compression artifacts all over, thus making the pictures it takes look worse than a $90 Nikon.dblevitan - Tuesday, May 20, 2008 - link
There are two key advantages of RAW that you missed and that could be useful to even people who are slighly interested in improving images.First, RAW provides 12 or 14 bits/pixel while JPEG only provides 8 bits. This is extremely relevant if you edit at all in Photoshop/Lightroom because any kind of level adjustment (even auto) will cause more color degradation with an 8 bit image than with a 16 bit image (which is what the 12 bit RAW files are generally processed into).
Second, RAW provides a better chance of recovering improperly exposed images by allowing at least 1/2 stop of exposure correction without any penalty and often more without significant issues. With JPEG you simply won't get this.
Wesley Fink - Tuesday, May 20, 2008 - link
You are absolutely correct that while JPEG is 24 bits per pixel it is only 8 bits per color channel (red, green, blue). The Pentax K20D is specified as a 14-bit A/D processor.Photoshop and other processing software, as you point out, normally processes this in 16-bit. However, it really doesn't matter in the end how the program processed it or saved it, what matters is the input bit depth. Many are not aware of this, but Photoshop also processes JPEGS as 16-bit on more powerful computers, but this just speeds up processing because you start with 8-bit and save as 8-bit with JPEG. There is no real advantage processing 12-bit files as 16-bit except processing speed as you don't gain real resolution improvements.
I agree RAW SHOULD allow more dynamic range, but some of the newest models like the Nikon D300 can actually do just as well or better in dynamic range in JPEG. However, in general what you say is true although it is changing as the processing power in DSLRs is improving.
What we really need is a higher bit-width JPEG standard. Let's hope the JPEG standards committee is hard at work on just that.
Bull Dog - Monday, May 19, 2008 - link
Is it just me or is the third page missing?Wesley Fink - Tuesday, May 20, 2008 - link
The third page is there. We have been having server problems today and this article went to post twice when it was scheduled to go live on 5/20. You may have been caught in one of those accidental post cycles. We apologize for the confusion.Bull Dog - Tuesday, May 20, 2008 - link
Yea. Right after I posted the comment I realized that the entire article was gone; so I figured the article must have not have been ready for prime time.Very good read, a little of it is above/beyond me, but a good read none the less. Thanks.