Dan Margulis Applied Color Theory - 16-bit: Is It All Hype?

Date: Fri, 06 Dec 2002 12:30:35 -0000
From: Henry Domke
Subject: Does 16-bit help Up-sampling or USM?

I know that 16-bit files are supposed to be better than 8-bit files files if a lot color correction is needed.

However, is there any advantage to keeping them as high-bit files in Photoshop for Unsharp Mask or for up-sampling? Fred Miranda's website has Photoshop Actions specifically to do that. But is it really worth it?

http://www.fredmiranda.com/ActionprofilesPage/index.html

Dan Margulis says that despite the theoretical advantages of high-bit images there are no improvements in the printed image. Has anyone done tests side- by-side with USM or Upsampling?

My images are coming from a Nikon D1X camera and are being printed on an Epson 7600 printer on 24 x 30 inch matte paper.

Thanks!
Henry

---
Henry Domke
Cracked Pot Farm Photography
3914 Foxdale Road
New Bloomfield, MO 65063


Date: Fri, 6 Dec 2002 16:01:04 -0500
From: Stephen Jackson
Subject: Re: Does 16-bit help Up-sampling or USM?

>Dan Margulis says that despite the theoretical advantages of high-bit images
>there are no improvements in the printed image. Has anyone done tests side-
>by-side with USM or Upsampling?

I also use a D1x and have done extensive testing in the difference of 16 bit vs. 8 bit. IMHO editing files in any way in 16 bit mode tends to flatten the image in a very subtle way i.e there is less differentiation between adjacent tonal values. Where this may help in portraits of women or other fashion work, but personally I don't like the mushy look of the images. In terms of resizing, my work is primarily advertising and some of the images have gone up to Billboard and they looked just fine in 8 bit. I tend to concur with Dan that there is no reason to use 16 bit mode. It would appear that it is in most cases detrimental to the overall quality of the image.

Hopes this helps

BTW Capture 3.5 allows a 10MP Raw file which is even better than the original 6MP, again IMHO


Date: Fri, 6 Dec 2002 17:47:31 -0500
From: Michael Demyan
Subject: RE: Does 16-bit help Up-sampling or USM?

Henry:

You have the setup. Try it and let us know if there is any difference.

Mike Demyan


Date: Fri, 06 Dec 2002 16:03:31 -0700
From: Andrew Rodney
Subject: Re: Does 16-bit help Up-sampling or USM?

Henry Domke wrote:

> However, is there any advantage to keeping them as high-bit files in
> Photoshop for Unsharp Mask or for up-sampling? Fred Miranda's website has
> Photoshop Actions specifically to do that. But is it really worth it?

Yes! Fred1s actions are quite good and affordable.

> Dan Margulis says that despite the theoretical advantages of high-bit images
> there are no improvements in the printed image. Has anyone done tests side-
> by-side with USM or Upsampling?

Well when all you do is output to ink on paper with a halftone screen, your ability to detect banding and other issues are very hard to see!

Andrew Rodney


Date: Fri, 6 Dec 2002 20:18:49 EST
From: Dan Margulis
Subject: Re: Does 16-bit help Up-sampling or USM?

Henry writes,

>>I know that 16-bit files are supposed to be better than 8-bit files files if a lot color correction is needed.>>

How do you know that? Have you ever seen any side-by-side examples where the same corrections were done in 8-bit vs. 16-bit, and the results were substantially superior one way or another?

>>However, is there any advantage to keeping them as high-bit files in Photoshop for Unsharp Mask or for up-sampling?>>

For USM, no; for upsampling not as far as I can determine.

>> Has anyone done tests side-by-side with USM or Upsampling?>>

As for upsampling, I tested this fairly exhaustively in preparation for the latest version of my book, although I didn't publish the result. I tested a whole suite of images upsampling up to 500% in raw 8-bit, raw 16-bit, and 16-bit converted to 8-bit and reconverted back to 16-bit. My conclusion was that the 8-bit upsamples were consistently superior, although the difference was slight in the overall scheme of things. I had expected that the 16-bit might be better.

As for USM, when I first heard the idea a few years back of using 16-bit mode for it I rolled my eyes, since this actually works against what one is trying to accomplish with USM. I ran a few tests back then which confirmed the obvious, that 8-bit sharpening normally gives superior results. As is usually the case when comparing 16- to 8- bit corrections, though, the difference is too small to be of any practical importance.

Dan Margulis


Date: Sat, 07 Dec 2002 03:13:56 -0000
From: Stephen Marsh
Subject: Re: Does 16-bit help Up-sampling or USM?

Dan writes:

> As for USM, when I first heard the idea a few years back of using 16-bit mode
> for it I rolled my eyes, since this actually works against what one is trying
> to accomplish with USM. I ran a few tests back then which confirmed the
> obvious, that 8-bit sharpening normally gives superior results.

Most like their raw high bit captures with no sharpening. But since sharpening can be performed in high bit, some users may like to restore the appearance of lost detail but perhaps not add the sort of sharpening that is required for halftone or stochastic output.

I would like to raise the stakes here...forget 16 bpc editing - why not 32 bpc edits on RGB data!<g>

http://www.applied-maths.com/paul/SharpControl.zip

http://www.applied-maths.com

PC only, freeware executable - virus check and use at with common caution, no warranty, usual disclaimers etc.

This app sharpens in the intensity channel only of HSI colour space, so it performs the mode switch from RGB > HSI > RGB in 32 bpc behind the scenes.

This uses a different algorithm than standard USM. Not very fast.

I don't know if this is a _valid_ contribution to the topic - it just seemed like a good time to post this new find. Some interesting features and results but nothing earth shattering.

Regards,

Stephen Marsh.


Date: Sun, 08 Dec 2002 02:12:54 -0000
From: "danzimmerman
Subject: Re: Does 16-bit help Up-sampling or USM?

Stephen wrote:

> This uses a different algorithm than standard USM. Not very fast.

Im fairly impressed with Sharp Control quality but have to admit that when you say not very fast, I think that should be underlined in bold letters. You can take a shower or eat dinner between start to finish for a big file.

Dan Zimmerman


Date: Sun, 08 Dec 2002 02:37:45 -0000
From: Stephen Marsh
Subject: Re: Does 16-bit help Up-sampling or USM?

> Im fairly impressed with Sharp Control quality but have to admit
> that when you say not very fast, I think that should be underlined
> in bold letters. You can take a shower or eat dinner between start
> to finish for a big file.

Hi Dan, well mileage will vary of course.<g>I was not putting true production files through it (being Mac based most of the time) but I was using my test images which contain combined crops from various images which all show different characteristics for evaluating sharpening, grain reduction, interpolation etc.

One reason that 16 bpc edits are not popular in high volume settings is due to the time penalty when working on typically high res files. Despite the power on the average workshop/studio desktop - many operations still take far too long (makes me wonder how we got by on Quadras or Centris back in the mid 90's<g>).

SharpControl converts 8 or 16 bpc source data to 32 bpc RGB data behind the scenes, then goes to HSI space and applies the effects to the intensity channel, then goes back to RGB (still in 32 bpc. All those operations are performed in 32 bpc RGB - even if only one channel of three is actually being processed for the sharpening.

This seems like a 'backyard' programmer, so perhaps a bigger budget would produce better optimizations. So perhaps the actual SharpControl algorithm may not be that much more processor intensive than more standard sharpening - I would attribute the lack of speed more to the 32 bpc editing.

Stephen Marsh.


Date: Tue, 07 Jan 2003 11:09:37 -0000
From: Henry Domke
Subject: Why is Adobe following 16-bit if it is all hype?

On this list and in his writing Dan Margulis has repeated questioned the benefit of 16-bit over 8-bit files. If he is correct then why would Adobe be adding more 16-bit features in the next version of Photoshop? (At least the rumors I read online say that they plan to do that).

Is it all hype? Is it too subtle to see? Some people have written online that Dan can't see the difference because he works for prepress and that if you use a higher quality output then the difference is obvious.

What is the truth here?
Is it just a marketing ploy by Adobe?


Date: Tue, 07 Jan 2003 11:56:33 -0000
From: Stephen Marsh
Subject: Re: Why is Adobe following 16-bit if it is all hype?

> On this list and in his writing Dan Margulis has repeated questioned the
> benifit of 16-bit over 8-bit files. If he is correct then why would Adobe be
> adding more 16-bit features in the next version of Photoshop? (At least the
> rumors I read online say that they plan to do that).

If the customer base requests it, Adobe would be wise to listen to their customers.

> Is it all hype?
> Is it too subtle to see?
> Some people have written online that Dan can't see the difference because
> he works for prepress and that if you use a higher quality output then the
> difference is obvious.

If the difference is obvious then there should be no debate - as the results should speak for themself. As there is some debate, perhaps the answer is not so obvious.

I consider this hypothetical question, for my work and workflow it is hard enough to get things done on 8 bpc data - let alone high bits. If there were no processing or workflow issues with high bit files and they were just as 'quick' and flexible as reg bit then I may use them. This is not the current situation for me (one off use for offset print).

> What is the truth here?
> Is it just a marketing ploy by Adobe?

I can't speak for Adobe or its marketing department, although I would guess that if they think there is a market for a feature, or if a market can be created by offering a new feature then that would be sufficient reason for inclusion in the feature set.

There is truth to the high bit workflow, it's just that it has not been shown to be a concern for many situations. There are many who endorse high bit workflows who do not seem to take into consideration that others may not have the luxury to follow these suggestions. Dan is the only 'name' that I know of who adds some balance based on production reality to an issue which usually only has one side presented.

Stephen Marsh


Date: Tue, 07 Jan 2003 06:00:34 -0600
From: Gary Roushkolb
Subject: Re: Why is Adobe following 16-bit if it is all hype?

Henry Domke wrote:

> On this list and in his writing Dan Margulis has repeated questioned the
> benifit of 16-bit over 8-bit files. If he is correct then why would Adobe be
> adding more 16-bit features in the next version of Photoshop?

I have been running a high end scanner for a number of years and have yet to
be asked for one from a customer. Adobe tends to put what they think people
may want if it1s cost effective maybe not what they are using. Good for
business sense "if it is there maybe they1ll use it".

Gary Roushkolb Donlevy Lithograph Wichita, KS


Date: Tue, 07 Jan 2003 06:55:57 -0700
From: Andrew Rodney
Subject: Re: Why is Adobe following 16-bit if it is all hype?

Read the following URL:

http://www.brucelindbloom.com/DanMargulis.html

> Is it all hype?

No.

> Is it too subtle to see?

It can be. Depends on the moves made on the image and where the image will
be output/

> Some people have written online that Dan can't see the difference because
> he works for prepress and that if you use a higher quality output then the
> difference is obvious.

Again, check out the URL above and check out the fellow (Bruce) who has vast experience doing scientific analysis of such issues. I think you1ll find it a VERY interesting read. Oh, the rest of this site if filled with really useful color theory information.

Andrew Rodney


Date: Tue, 07 Jan 2003 09:05:55 -0500
From: Jim Rich
Subject: More on the high bit vs 8-bit argument

Dan Margulis some months ago made a challenge to the 16-bit advocates to either put up or shut up. So far, I have not seen anyone show hard evidence to prove or disprove Dan's argument. So far no one that I have met has. Dan's argument was so good, I took the time to create a test to either nullify or validate some of his tests.

The high bit depth advocates make a good case in theory. The theory is that there are more bits in each pixel in a 16-bit image. So when you apply image processing in a program such as Photoshop, you will have less quantization errors and therefore less potential banding and overall image degradation issues. But in practice when you ask them to show you hard evidence, that is, to compare 16-bit prints to 8-bit prints with the same number of edits, they have some rational arguments such as reviewing histograms, or point out that an 8-bit workflow will ruin your images because there is not enough data so your images will be ruined or the rational discussion often stops and sometimes becomes a battle of, 3my technique is better than yours because I say so.2

One of the key end-user benefits of using 8-bit images over 16-bit images is that in the 8-bit modes, Photoshop has more tools available and therefore has more capabilities for adjusting and manipulating images. Other benefits are smaller file sizes. Using 8-bit images makes the workflow more efficient.

The Test
Last month (Dec 2002) at a color management conference, I setup and ran a test comparing 8-bit and 16-bit RGB images. For this test I printed 8-bit and 16-bit images via Photoshop (with profiles) to my Epson 5500 printer. Photoshop was set up to simulate an RGB workflow. Nothing was done purposely to bias the test. Images were scanned and brought into the Adobe 1998 working space. Some images in the test had 30+ edits as well as went from RGB to LAB and back a few times.

This was my third shot at creating a test and I am sure this is not the last one I will do since I only focused on RGB and not CMYK workflows.

For this test, I laid out the 28 prints on a table and let a group of imaging experts (pre-press types and photographers) inspect and review them. The test was not for color accuracy between prints. I gave any willing participant a form with yes and no response categories for yes, it is 8-bit or 16-bit or no, it is not.

The initial feedback from a group of experts who did not choose to fill out the form but who took a few minutes to compare the images was that they could not precisely see any visual differences between the 8-bit or 16-bit images and that any response they would have would be a guess. This result was verified again with approximately 20 test forms that were filled out. The overall outcome showed all participants were guessing (40% to 50% of the time) at which images were 8-bit or 16-bits.

Histograms Can Guide You But
The images I used were at risk to posterize due to the combination of their image content and the extraordinary image processing that was applied. Except for one image that had banding in the original, all images fared well and they did not degrade as one might expect.

In theory, one might have a bad looking histogram (with lots of gaps), making one believe that the image will posterize. In this test, all of the 8-bit histograms looked bad. However, the number of visible gaps in these histograms was not the defining factor, it was how the final prints looked. Any other problems, such as banding or image degradation related to too many image-processing edits, were indistinguishable.

Are more than 8-bits necessary?
The test I have described covers a majority of images used in most RGB workflow situations. But, there are some exceptions. So as you look at and consider the implications of this information, take care. Some situations require using 16-bit images in Photoshop.

1. One reason to work with 16-bit images in Photoshop is if your original image (film) has a harsh break or banding in the way the image was originally photographed and/or in the way the film processed. Once you scan that problematic image into your system, using 16 bits will allow you to process the file without much degradation to the image area with the harsh break. If you apply the same image processing to an 8-bit image, the file will degrade and fall apart. One solution is to only scan those types of images as 16- bit. Then work in the 16-bit mode as long as possible. Since banding problems show up in a small percent (2 percent or less) of all images that means 98% of all images should be scanned into an 8-bit workflow.

2. What I have commented on is related to working in Photoshop. It does not mean you should only use 8-bits for image capture, that is, for scanning and digital cameras. Digital cameras and scanners often need more than 8-bits per pixel to capture a good range of tones so you can get good shadow detail and color saturation. An input device that can capture over 8-bits per pixel is a positive attribute to an imaging system. After the input device does the capture, say in 16-bits, and if the image does not have any banding created in the original photograph, then switch to 8-bits and begin your working session in Photoshop.

3. In the early days of desktop scanning, and digital cameras the development of sensors, and optical technology and poor implementation of scanning and digital camera technology caused lots of banding problems. One of the technical reasons for banding was poorly written scanner drivers. The problem typically was created by inadequate or incorrect math when the driver was developed and written. At that time (actually just a five or six years ago), this problem made some experts recommend working in 16-bits. At that time, it was probably the correct thing to do. In the last few years, scanning and digital camera technology has matured and become much improved. Recent tests, including mine, validates that there are very few reasons to create a high bit depth workflow as least for RGB workflow.

If you are skeptical about working in 8-bits then consider this strategy. Use the scanner or digital camera to capture the best 16-bit data. Make a copy and work on that. Use the original as a backup.

From an end-users perspective, the reality is, if you have a scanner that introduces banding with either 8-bit or 16-bit images, fix the problem either by acquiring a better scanner driver or by getting a new scanner.

In the initial stages of these tests, I actually had a 16-bit image with banding from an older 16-bit scanner. This actually fooled me until I rescanned the image on our newer and improved 8-bit and 16-bit scanner. The problem went away.

In another part of this testing, I took a group of these test images to a photo finisher and output some of them to a Lambda print device and a to a LVT Transparency (Light Value Technology) device. Again, on photo quality prints and transparencies, the 8-bit and 16-bit results were indistinguishable.

The Bottom Line.
The conclusions of this post are based on hard evidence and years of practical experience. If you are skeptical, my advice is to consider this: If you have a good input device with well-integrated technology, it should offer high bit depth capabilities. Do some test scans or digital capture in both 8-bit and 16-bit modes. If the 8-bit mode has any problems you don1t like, then use 16-bits as a capture option. Then in Photoshop convert to 8-bits and start your working session. The only exception would be images that have banding built into them then you would want to work in 16-bits as long as possible.

Jim Rich


Date: Tue, 07 Jan 2003 14:24:58 -0000
From: Stephen Marsh
Subject: Re: Why is Adobe following 16-bit if it is all hype?

Andrew Rodney wrote:

> Read the following URL:
> http://www.brucelindbloom.com/DanMargulis.html

I think Dan will post on this, but I suspect I know his answer since I a have questioned him about this part of Bruce's site last year.

I have great respect for both Dan and Bruce, but in this case I think Bruce still does not follow the entire testing procedure correctly.

I have taken the time to go back through the list archives in date order and I came to the conclusion that Dan is indeed performing all three tests and I know this as well from writing to Dan about this last year.

Even if Dan was _only_ performing the test on reg bpc data and false high bit data (which he is not, he tests the true high bit too) - that is not to say that others can't compare thing's themself and make their own decision, that is what counts in the end. I have done tests with true high bit data, reg bit data and the false high bit data so I do not think the process is flawed, each has a purpose and point.

The 'condition #5' that Bruce points (false high bit) is only one of three tests - high bit, false high bit and reg bit. It is my understanding that three and not two tests are performed for each image.

Stephen Marsh.


Date: Tue, 07 Jan 2003 14:58:41 -0500
From: Ric Cohn
Subject: Re: More on the high bit vs 8-bit argument

From: Jim Rich

> The conclusions of this post are based on hard evidence and years of
> practical experience. If you are skeptical, my advice is to consider this:
> If you have a good input device with well-integrated technology, it should
> offer high bit depth capabilities. Do some test scans or digital capture in
> both 8-bit and 16-bit modes. If the 8-bit mode has any problems you don1t
> like, then use 16-bits as a capture option. Then in Photoshop convert to
> 8-bits and start your working session. The only exception would be images
> that have banding built into them then you would want to work in 16-bits as
> long as possible.

Jim-

Your tests and conclusions basically agree with mine. I originally set out to try and prove, not disprove, a 16 bit advantage. I'm proud to say some of my testing is represented in Dan's excellent new book, including some 8 bit and 16 bit image details on the CD. I've personally seen examples of, and have heard of real world tests like yours, which fail to prove a 16 bit workflow advantage.

Unfortunately, the 16 bit proponents are treating it as an article of faith and I've never seen any purported examples of a file that benefited from being worked on in 16 bits so I could agree or disagree with their methods. Yet practically every month I see the 16 bit advantage quoted as a matter of irrefutable fact. As a former science major I find this "religion based" attitude infuriating. I too would love to find a magic way to improve my images, I just see no evidence that this is it.

Ric Cohn
http://www.riccohn.com
212.924.4450


Date: Tue, 07 Jan 2003 13:38:26 -0500
From: Jerry Lodriguss
Subject: Re: Why is Adobe following 16-bit if it is all hype?

Photoshop does not even really work on 16-bit data.

It really works on 15-bits.

This has been confirmed by an Adobe staff member (Thomas Ruark) on the Photoshop SDK mailing list.

It is apparently done for performance reasons.

Jerry


Date: Tue, 07 Jan 2003 22:21:04 -0600
From: David Riecks
Subject: Re: Why is Adobe following 16-bit if it is all hype?

Jerry:

In order for that to really matter, we'd also need input devices that are gathering 16-bits per channel.

To the best of my knowledge, I don't know any that do. Most scanners are either 10, 12, or 14 bit, most digital cameras tend to be in the 10 to 12 bit per channel range.

I usually just call it "high-bit" data, and bypass all the numbers. ;-)

David

David Riecks
http://www.riecks.com, Chicago Midwest ASMP member


Date: Wed, 08 Jan 2003 04:39:09 -0000
From: Stephen Marsh
Subject: Re: Why is Adobe following 16-bit if it is all hype?

David Riecks wrote:

> To the best of my knowledge, I don't know any that do. Most scanners are
> either 10, 12, or 14 bit, most digital cameras tend to be in the 10 to 12
> bit per channel range.

I think that's a valid point David, not to mention that even with 16 bpc CCD cooled devices like a EverSmart Supreme, there is still going to be at least 1-2 useless bits of data representing sensor noise or other issues.

As for the Adobe marketing speak - I'm not sure how you can say it is 16 when it is 15, but I am not in marketing.<g>I do not recall Adobe admitting the fact (at least in mainstream press), but it has been noted by other sources.

Stephen Marsh


Date: Tue, 07 Jan 2003 21:49:43 -0700
From: Ron Kelly
Subject: Re: More on the high bit vs 8-bit argument

Ric Cohn wrote:

> Unfortunately, the 16 bit proponents are treating it as an article of faith
> and I've never seen any purported examples of a file that benefited from
> being worked on in 16 bits so I could agree or disagree with their methods.

Yes, it is indeed curious. But then again, a lot of people seem to see Elvis regularly, and a guy named "BigFoot". No way you can change their point-of-view either.

Cheers,
Ron Kelly


Date: Wed, 08 Jan 2003 06:48:26 -0700
From: Andrew Rodney
Subject: Re: Why is Adobe following 16-bit if it is all hype?

David Riecks wrote:

> In order for that to really matter, we'd also need input devices that are
> gathering 16-bits per channel.

I'm told that the Imacon scanners and cameras do capture full 16 bits of data.

> To the best of my knowledge, I don't know any that do. Most scanners are
> either 10, 12, or 14 bit, most digital cameras tend to be in the 10 to 12
> bit per channel range.

Even if that is the case, we still get the benefit of high bit data. So in my mind, a 14 bit scanner isn1t necessarily better or worse than a 12 bit scanner. In fact, I1d prefer dynamic range over higher bit depth any day.

> I usually just call it "high-bit" data, and bypass all the numbers. ;-)

Me too. I think since Photoshop considers any file with more than 8 bits per color a 16 bit file, calling what is really a 12 or 14 bit file 3High Bit2 is more accurate.

Andrew Rodney


Date: Wed, 08 Jan 2003 06:53:18 -0700
From: Andrew Rodney
Subject: Re: Re: Why is Adobe following 16-bit if it is all hype?

Stephen Marsh wrote:

> As for the Adobe marketing speak - I'm not sure how you can say it is
> 16 when it is 15, but I am not in marketing.<g>I do not recall
> Adobe admitting the fact (at least in mainstream press), but it has
> been noted by other sources.

Adobe can do two things.

1. It can operate as it does now and call all high bit files 316 bit.2 I guess we could argue it should be called 3High Bit2 (I1d vote for that). 2. It can have a huge Mode with 10 bit-12bit-14bit-16bit and then have to figure out (if it even can) what the 3true2 bit depth of the file is (do we count a bit or two of noise?). Then we1d need a very complex set of info palettes to provide the 3true2 numbers in these files (the size of the info palette would have to be greatly enlarged to accommodate this). Then we1d need a Histogram the size of a small wall to show each of the 640000 odd values in just the 16 bit file.

I think what Adobe is doing now (with perhaps a change in the menu name as indicated in #1 above) makes the most sense.

I don1t think marketing has a clue with high bit data is. The Alpha and beta testers as well as the engineers certainly do as do many of the users of the product.

Andrew Rodney


Date: Wed, 8 Jan 2003 12:20:58 -0800
From: Jan Steinman
Subject: Re: Why is Adobe following 16-bit if it is all hype?

From: Henry Domke

>On this list and in his writing Dan Margulis has repeated questioned the
>benefit of 16-bit over 8-bit files.

Not to put words in Dan's mouth, but I think this is an argument over the degree to which 16 bit files are useful, rather than an absolutist "16 bit files are never useful."

>If he is correct then why would Adobe be
>adding more 16-bit features in the next version of Photoshop?

I would welcome certain new 16 bit features. For example, 16 bit grayscale masks, a la "CMD-OPT-~". Sometimes on very poor quality or difficult material, I go through contortions bring out shadow detail without blowing out highlights.

16 bits are most useful when working with poor quality material, where one may have to perform gross corrections on portions of the image that otherwise might lead to posterization.

On material that is already pretty "normalized" in tonal range, I agree that 16 bits is of little use. It's when you need to make gross adjustments that it is handy -- and from my experience, visible.

--
: Jan Steinman -- nature Transography(TM): http://www.Bytesmiths.com>


Date: Wed, 08 Jan 2003 13:35:16 -0800
From: Dennis Dunbar
Subject: Re: Why is Adobe following 16-bit if it is all hype?

As a digital artist I've worked on lots of images that were tweaked to extremes. And worked on lots of bad images, (due to bad scanning or bad photography). In the movie biz you don't have the greatest choices for your original pieces and you spend a lot of time making silk purses, (BG).

There are times when I have seen high bit workflows avoid problems that the reg bit work is prone to. If you ever had a designer open a Hue Sat adj and click on "Colorize", you'll know that the reds going into the shadows are gonna band all over the place. Finding a way to do this in high bit helps to avoid that problem. Painting is another area where high bit depth is demonstrably better, you get much smoother transitions, especially going into the shadows. The best area where high bit depth helps is in making gradations, obviously with more steps available you'll get smoother transitions and less banding.

The problem is that PhotoShop doesn't let you do most of that work. Live Picture did, and that was one of my biggest reasons for keeping it around. Many times I made a gradation in LP and then brought it into PS for compositing. When I created the art for the Blair Witch Project poster I had to paint the shadows in LP because no matter what I tried the transitions were not smooth enough in PhotoShop.

But I almost never get great quality, scanned photos that need lots of whacking around via curves or whatever. If I have an image that really needs extreme fixing, (as opposed to artistic manipulation) it's 'cause it a bad scan, (i.e. PhotoCD etc.). So in my work I'd love to see PS give us more tools in high bit, but not for any of the reasons that have been previously discussed.

Dennis Dunba


Date: Wed, 08 Jan 2003 16:11:31 -0700
From: Ron Kelly
Subject: Re: Re: Why is Adobe following 16-bit if it is all hype?

Dennis Dunbar wrote:

> So in my work I'd love to see PS
> give us more tools in high bit, but not for any of the reasons that have
> been previously discussed.

At last, a reasonable explanation of why/where high bit is useful. Thanks,

Ron Kelly


Date: Wed, 8 Jan 2003 22:15:29 -0800
From: "Darren Bernaerdt"
Subject: RE: Why is Adobe following 16-bit if it is all hype?

Andrew,

Before Dalsa started manufacturing the ubiquitous 6MP CCD that Imacon and many others use, Philips stated their 6MP CCD (FTF3020) produced 13 bits of data per channel. I doubt Dalsa has changed the specs. AFAIK, there aren't any area array digital backs out there that are capturing 16 bits of data. I recall that the new 11MP CCD has a lower dynamic range since the pixels are smaller. Extra resolution, but less dynamic range.

Darren Bernaerdt


Date: Thu, 09 Jan 2003 09:17:30 -0600
From: Lori Sabo
Subject: Re: Why is Adobe following 16-bit if it is all hype?

Most scanners or cameras don't give us true 16-bit anyway. I think the Imacon Flextight Photo for instance is 12 bits. Which is 16 times the gradation info of 8-bit. (I might be remembering wrong, but I know it is less than 16 bits)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Lori Sabo


Date: Thu, 9 Jan 2003 10:53:20 -0800
From: Jan Steinman
Subject: Re: Why is Adobe following 16-bit if it is all hype?

> From: Dennis Dunbar
>
>Live Picture did, and that was one of my biggest reasons for keeping it
>around...

Sheesh. Why doesn't someone buy LP and port it to MacOS X? It did things that no other application could at the time, and probably things that no current application does.

I miss working on 200MB images efficiently with only 32MB of RAM... I'm tired of banging my head against unexpected and undocumented limitations in Photoshop...

--
: Jan Steinman -- nature Transography(TM): http://www.Bytesmiths.com>
: Bytesmiths -- artists' services: http://www.Bytesmiths.com/Services>


Date: Thu, 09 Jan 2003 14:01:50 -0700
From: Andrew Rodney
Subject: Re: Why is Adobe following 16-bit if it is all hype?

Darren Bernaerdt wrote:

> Before Dalsa started manufacturing the ubiquitous 6MP CCD that Imacon and
> many others use, Philips stated their 6MP CCD (FTF3020) produced 13 bits of
> data per channel. I doubt Dalsa has changed the specs. AFAIK, there aren't
> any area array digital backs out there that are capturing 16 bits of data. I
> recall that the new 11MP CCD has a lower dynamic range since the pixels are
> smaller. Extra resolution, but less dynamic range.

Could just be the scanner which of course uses a different CCD.

Andrew Rodney


Date: Thu, 09 Jan 2003 14:14:40 -0700
From: Andrew Rodney
Subject: Re: Why is Adobe following 16-bit if it is all hype?

http://www.imacon.dk/sw172.asp

They do say 3True2 16 bit...

If we could get Bruce Lindbloom to once again come back and help us out, I have little doubt he1d be able to look at a high bit Imacon scan (I1d be happy to provide him) and let us know the 3true2 bit depth. That should be much easier to figure out than dynamic range (talk about measuring with a rubber ruler...).

Andrew


Date: Fri, 10 Jan 2003 14:23:46 -0800
From: Jan Steinman
Subject: RE: Why is Adobe following 16-bit if it is all hype?

From: Darren Bernaerdt
>
>Before Dalsa started manufacturing the ubiquitous 6MP CCD that Imacon and
>many others use, Philips stated their 6MP CCD (FTF3020) produced 13 bits of
>data per channel. I doubt Dalsa has changed the specs.

CCDs are analog devices. They don't have *any* bits. You can digitize the output to any number of bits you wish.

Of course, they have a noise floor, and beyond a certain point, it makes no sense to make up more bits, but there are things like active cooling that can improve that a bit or two.

--
: Jan Steinman -- nature Transography(TM): http://www.Bytesmiths.com>
: Bytesmiths -- artists' services: http://www.Bytesmiths.com/Services>


Date: Fri, 10 Jan 2003 17:18:34 -0600
From: N9VJG
Subject: Re: Why is Adobe following 16-bit if it is all hype?

>16 bits are most useful when working with poor quality material,
>where one may have to perform gross corrections on portions of the
>image that otherwise might lead to posterization.
...
So, if we use your suggestion, does we are assuming the "poor quality material" is being scanned in 16 bit mode, right?

--
Eric Curtis Bond
Photo Grafix
http://www.abetterreality.net
(847) 673-7043


Date: Fri, 10 Jan 2003 20:39:33 EST
From: Dan Margulis
Subject: Re: Why is Adobe following 16-bit if it is all hype?

Jim Rich writes,

>> The images I used were at risk to posterize due to the combination of their image content and the extraordinary image processing that was applied. Except for one image that had banding in the original, all images fared well and they did not degrade as one might expect. In theory, one might have a bad looking histogram (with lots of gaps), making one believe that the image will posterize. In this test, all of the 8-bit histograms looked bad. However, the number of visible gaps in these histograms was not the defining factor, it was how the final prints looked. Any other problems, such as banding or image degradation related to too many image-processing edits, were indistinguishable.>>

I have nothing in particular to add to this excellent post except to confirm that I have seen Jim's test and he is describing it accurately both in terms of what it was and what the results were. He obviously put a great deal of time and effort into it, and deserves thanks from those interested in the topic. Some of the original images were of poor quality and needed extensive editing. He output at quite large sizes. If anything was ever going to show a 16-bit advantage, these factors ought to have done it.

Stephen writes,

>> I think Dan will post on this, but I suspect I know his answer since I a have questioned him about this part of Bruces site last year.I have great respect for both Dan and Bruce, but in this case I think Bruce still does not follow the entire testing procedure correctly. I have taken the time to go back through the list archives in date order and I came to the conclusion that Dan is indeed performing all three tests and I know this as well from writing to Dan about this last year.>>

Those who insist that 16-bit edits produce a "night and day" difference, or that they're "critical" to quality reproduction, or that anybody who doesn't edit in 16-bit is an amateur, have painted themselves into a considerable corner. If there were really any advantage, let alone a big one, it would be pretty easy to produce real images that illustrate it in side-by-side tests such as the ones Jim ran or the ones found in the latest edition of my book.

As they can't produce any such images, their only alternative is to go ad hominem, as Bruce and others unfortunately have. The site speaks for itself.

Dan Margulis


Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.