Deal 5: Natural Light: Our best-selling ebook
A Guest post by Steve Berardi from PhotoNaturalist
In digital photography, there’s a lot of talk about bits:
But, what does it all mean? Although more bits generally means better quality processing, it’s not always that simple. Sometimes you really need to know the full story before making a conclusion.
Before we jump into these specific examples that are common in digital photography, it’s important to understand what a bit actually is.
The word “bit” comes from two words: binary and digit. Each bit has two possible values: 1 or 0. Sometimes it’s also helpful to think of a bit as either “on” (1) or “off” (0).
Although one bit can store two possible values, a sequence of two bits can store four possible values: 00, 01, 10, and 11. With a sequence of bits, order matters, so “01” is very different from “10.”
Each time you add a bit to a sequence, you double the number of possible values, so if you go from two bits to three bits, you go from four possible values to eight possible values.
As an example, let’s say you were interested in storing something that has 16 possible values. You would need 4 bits in this case (2 x 2 x 2 x 2 = 16).
Every digital camera has some kind of analog to digital converter (ADC) that converts the analog signal captured by the sensor into a digital signal that produces your image. Most DSLRs these days either have a 12-bit or 14-bit ADC.
The bits in this case refer to the number of possible tonal values that your camera’s sensor can capture. For example, a 12-bit ADC can capture 4,096 possible tonal values for each pixel. On the other hand, the 14-bit ADC can capture 16,384 tonal values for each pixel.
At first, it may seem like the 14-bit ADC is a clear winner: it can capture 12,000 more tonal values! But, these two extra bits are not increasing the dynamic range of your camera, they’re only adding more steps within that range. It’s kind of like if you took a loaf of bread and cut each slice in half to make even smaller slices. You might have more slices of bread now, but the size of the loaf is the same!
So, you’ll experience slightly better image quality with the 14-bit ADC, but it’s really just noticeable in the deep shadows and smooth gradients (like those you find in a photo of a sunset).
For more details on 14-bit vs 12-bit ADCs, check out these helpful articles:
In most post-processing software, you have the option between 8-bit color and 16-bit color. The bits in this case refer to the number of possible tonal values available to each color channel (red, green, and blue) of each pixel.
With 8-bit images, you have 256 possible values for the red channel, 256 values for the green channel, and 256 values for the blue channel. And with 16-bit images, you have 65,536 possible values for each color channel.
Using 16-bit color will result in some pretty large filesizes, but it’s worth the extra size because with 16-bit processing you’ll significantly reduce your chances of posterization (as seen in the photo below). It’s a good idea to use 16-bit color even if you originally shot in JPEG (which is 8-bit), because the extra bits will help reduce rounding errors when performing common post-processing tasks like Curves or Levels.
For more information on 8-bit vs 16-bit color, check out these helpful articles:
Some post-processing applications, like Adobe Photoshop, offer 32-bit versions and 64-bit versions. The bits in this case refer to the number of possible memory addresses. With 32-bits, you can use up to 4GB of physical memory, but with 64-bits you can theoretically use up to 17.2 billion GB of memory (although this amount is usually severely limited by the operating system).
There’s a common misconception that the 64-bit version of Photoshop is always faster, but in reality, to take advantage of the speedup (which is minimal), three things need to happen:
You might think that 800 MB is larger than you’ll ever work with, but filesizes can get big pretty quick if you’re building a panoramic image, or working with multiple layers of images to blend exposures.
Even if you do meet all three conditions above, there’s still a chance you won’t notice any speed-up at all, depending on the operations you’re performing on the image. Also, keep in mind that many third-party plugins do not work with the 64-bit version of Photoshop.
For more details on 32-bit Vs. 64-bit Photoshop, check out these informative articles:
The key thing to take away from this post is that not all bits are created equally. Just because something has twice as many bits that doesn’t mean it’s automatically twice as fast or two times better quality. Before making any kind of conclusion about 16-bit vs 32-bit / etc, you really have to understand the story of how those bits are being used.
About the Author: Steve Berardi is a naturalist, photographer, and computer scientist.
Become a Contributor: Check out Write for DPS page for details about how YOU can share your photography tips with the DPS community.
December 22, 2010 07:39 am
I have 32 or 64 bit options to use in photoshop. Is the printing photo quality better when using a higher bit?
September 1, 2010 10:24 am
@Dean: I think you are referring to page file. While this considerably slows down systems when trying to access the bits that were temporarily stored on the HD. I am confident this wouldn't effect the quality. Bits are Bits, and they are in an organized structure when being stored. The bits are kept in tact, but it just takes much longer to retrieve once the program decides it needs them again (to store in the registers on the processor).
-Senior in Computer Science.
August 30, 2010 09:44 pm
I have had good success in reducing the posterized effect by selectively blurring it when it occurs. Sometimes as much as 130 pixels in Elements. Great when you have to use JPEG
August 30, 2010 02:27 pm
That is an amazing shot. I want CS5. Lightroom, 64 bit quad core 2.5 ghz 4g ram computer right now. Oh, and all your talent too.
August 30, 2010 03:41 am
Question. Explain the use of HD memory usage as virtual RAM when needed during processing over and beyond the RAM limit. I know this can affect speed but what about any quality issues??[eimg url='http://deanhuggins.smugmug.com/Landscapes/Landscape-and-Scenics-You-Will/Pal-falls-evening-v1/569829822_uL3pr-M-4.jpg' title='569829822_uL3pr-M-4.jpg']
Using CS5, LR2, 4G RAM, ESTAT SSHD, GT9600 512Mb video, Win7 64 bit, QUAD core 2.5 ghz
August 27, 2010 01:33 pm
thanks for informative and comment producing article.
I run two systems one older pc with 32 bit with cs2 etc and one brand newquad core pc set up as 64 bit and with stacks n stacks of memory, the lastest vid cards and I have noticed the difference in photoshop, bridge and premier in hadling bigger files - THEY FLY!!
I use both jpgs and raw and love the fast pace the new machine with windows 7 and cs5 etc
best investment ever.
Both computers were hand built by computer nerds who understand what i needed to do and fingers crossed no problems or issues.
fast turnaround = happy customers = keeping Sam in business.
But in saying that my old machine can still crunch when it has to, it just means i get an extra cuppa whilst waiting.
It is about knowledge, demand and finances at the time..... just enjoy and upgrade if and when you can...and learn from great articles like this one - although the maths part leaves me for dead - too creative me thinks...
August 26, 2010 08:54 am
Excellent. Thanks for a clear "exposure" :)
It answered a number of doubts I had on the subjet of 64-bit operating system.
August 25, 2010 05:32 am
Love the "Loaf of bread" analogy. Perfect.
August 24, 2010 01:24 pm
I'm not sure why I never find anyone explaining what is really going on. Regardless of the number of bits you're using, you're never perfect. In other words, you quantize the color level to be some number between 0 and (2^n)-1 (where n is the number of bits), but you're always off by something between (max level/(2*((2^n)-1))). That is somewhere between 0 and half the amount between each quantization level.
Say, for example you have a 2 bit system. That means you have 4 levels, numbered 0 to 3. Say the real value of the color level is 2.145678762..., but your camera rounds down to 2.0. That means you are off by 0.145678762... If you had some way to tell the actual (i.e., real) value for all the pixels, you would be able to tell that the amount of this extra bit is perfectly random. This can be proven mathematically.
In other words, what you get is extra noise in the raw image. That's all. Extra noise. More bits = less noise.
Of course, if you're not using the RAW image, it's being processed so that it probably no longer looks like perfectly random noise.
So, fewer bits is very similar to using a higher ISO.
August 24, 2010 11:18 am
Very nice!! Thanks a lot!!
August 24, 2010 07:55 am
I agree with everyone else. This is a super informative article. Now, please write one about dpi vs ppi and printing sizes. That is something I've never really understood.
August 24, 2010 05:07 am
Thanks all for your nice comments!
@speedy - Great point! Running a ton of other applications while you're editing photos can defintely eat up a lot of memory. I should have mentioned that point #2 assumes you don't have any other apps open :)
@karen - Try saving the image as a PNG, which uses a compression algorithm that's better suited for the sharp transitions you often find in logos. JPEG was developed specifically for photographs, where you have smooth transitions between edges.
@JP - Ahh, thanks for pointing that out about 10.6!
@fortunato_uno - You're right that 11111111111111 = 16383, but computers start counting at 0, so 14 bits can actually represent a total of 16384 different values (from 0 to 16383).
August 24, 2010 03:31 am
I especially appreciated the facts about what it takes to get the benefit of 64-bit Photoshop.
August 24, 2010 01:08 am
I think the toast example is misleading as most people will think that thinner toast is bad. I suggest using posterization in a greyscale image as an example since you go on to mention highlights and shadows. That would also provide continuity with your colour example.
I've turned 'off' my 64 bit Aperture: slower, don't have that much RAM, and only Viveza and NN plug-ins are 64bit.
PS bread is just raw toast ( sorry, due to poor memory, I can't attribute this to the comedian who said it)
August 24, 2010 12:33 am
also (I glossed over your point on this the first time I read it):
It's possible your books are saying the values are 0-15 or 0-16383. Note that 0-15 is 16 different values, just as 0-16383 is 16384 different values.
August 24, 2010 12:17 am
@fortunato_uno, that's a common misconception. 64-bit data buses have been around for many years. When people talk about 32-bit vs 64-bit OS's, it's the memory address bus they're talking about. The article is correct (which isn't surprising considering it was written by a computer scientist).
August 24, 2010 12:01 am
16383 is correct - and counting zero as a representable number - 14 bits are capable of allowing 16384 distinct possibilities.
32 vs 64 bit issue. It is both processing, addressing, data transfer, and memory. My AMD 64 bit processor has a 52 bit limit on physical memory and only supports a 48-bit virtual address space. So, the software has to be ported, and the hardware has to be compatible to work right.
64-bit architecture is a major change from 32-bit.
August 23, 2010 11:19 pm
I hate to say it but, I thought the 32 vs 64 had to do with the proccesing. not the memory. if my elementry electronics is correct. it's a packet (or file of) of 32 bits being transfered vs 64 bit's being transfered. i also have to think my books must be wrong on the 14 bit value. if 00000000000001 = 1 and 0000000001000 = 8 and 00000000001111 = 15 then 11111111111111 should = 16383. i could be wrong, i don't have my calculator with me. but i am rather confident about the transfer thing. i'd be willing to bet my certificate on it.
August 23, 2010 09:35 pm
Great info and easily understood by this totally non-techie who just likes for her photographs to look nice. I always wondered what the "bits" meant. I still don't quite understand, but at least next time I can reference this article to help me out. Very well-written...thanks!
August 23, 2010 07:55 pm
Excellent and informative. Please write more articles on DPS.
August 23, 2010 05:00 pm
Let me add a little more to 32bit vs 64bit discussion. You do not have to have more than 4GB RAM installed to see benefits of 64bit OS. In 32bit OS we have 4GB address space for all application and HARDWARE and peripherals included. Today's graphic cards alone easily come in 1GB memory (graphic and gaming desktops very often are configured with dual cards), this is common reason for question "Why I can see only 3 GB of memory even though I have 4GB installed.
To cut the long story short, if you have 4GB RAM and video card (one or two) with 1GB video memory, you will need 64bit OS to be in able to take advantage of your hardware.
August 23, 2010 03:17 pm
great article. one of the more informative ones I've seen on DPS... since they are usually really dumb nowadays.
fyi, 10.6 doesn't natively boot up in 64 bit. it boots in 32bit, you have to hold 6 and 4 in the boot sequence. I've done it with my 2009 MBP and half the programs don't work properly or don't even work at all so don't bother considering using 64 bit mode for your mac. On the other hand, my Win7 in 64 bit (driver support is fantastic as well) is a dream to work with especially with 16GB's of ram. Even ubuntu in 10.04 is pretty good with 64bit although I've found that networking some of my printers were troublesome and one didn't even network at all.
August 23, 2010 12:19 pm
thanks so much for this very helpful post. i have a question about saving images for viewing on the web. i seem to have a lot of posterization on jpeg images with my logo. is there a best way to save for web that would reduce this?
August 23, 2010 10:38 am
To clarify a little on the 32 bit vs. 64 bit software applications, what the author (I believe) meant to say is that to be able to exploit the extra address space provided by the 64 bits, i.e. to take advantage of the ability to address more than 4 GB of memory, you have to actually have more than 4GB of physical memory installed in your computer (point 1), you must be running software (both operating system and applications) that can take advantage of 64 bit addresses (point 3), and you must be performing tasks that actually require more than 4 GB of memory usage. The last point was the intended meaning of point 2, I think, but actually the file size is fairly irrelevant. If you are running your image editing software, and are simultaneously running your web browser (maybe you're following an online image editing tutorial), your email program, listening to online radio and downloading a movie for later tonight, then memory usage may well exceed 4 GB even if the image you are editing is very small. This is especially so if you're using a dumpy operating system like Vista, which needs nearly 1 GB just to be alive, without any applications running at all. Good post though. Thanks!
August 23, 2010 08:46 am
Well written and thorough enough to be informative and helpful, while still light enough for the non-techie to understand. Thanks, Steve!
Receive a FREE SAMPLE of our Portrait Photography Ebook
Receive a FREE SAMPLE of our Portrait Photography Ebook
Receive a FREE SAMPLE of our Portrait Photography Ebook
Sign up to the free DPS PHOTOGRAPHY COURSE
GET DAILY free tips, news and reviews via our RSS Feed