From: C J Campbell on
On 2008-08-30 20:40:55 -0700, "Bob Donahue" <bobmgtd(a)comcast.net> said:

> I take a lot of pictures at car shows. The digital cameras I've had to date,
> have trouble rendering white cars in direct sunlight. You can't see the
> curves of the body panels, they come out pure white with no shading! Cars
> that are not white come out beautifully. I'm in the market for a new "point
> and shoot" camera. I've narrowed my choice down to the Fujifilm FinePix
> F100fd or the Nikon CoolPix 610. Both of these models are advertised as
> having special "dynamic range" modes. Which one would be my best bet for
> avoiding washed out highlights?

I agree with those who say you do not have a dynamic range problem. You
have an overexposure problem. Stop it down 1/3 stop and see if that
does not improve things.
--
Waddling Eagle
World Famous Flight Instructor

From: Olin K. McDaniel on
On Mon, 01 Sep 2008 05:40:18 GMT, "David J Taylor"
<david-taylor(a)blueyonder.neither-this-bit.nor-this-part.co.uk> wrote:

>Alan Browne wrote:
>[]
>> JPG's 8 bits/color is compressed DR, not more DR. The 'loss' is in
>> graduation 'tween colors. JPG cannot contain an expression of more
>> information than the original raw, compressed or otherwise.
>>
>> The key point is that in-camera JPG leaves you with much less in terms
>> of options than post-processed raw.
>
>With an 8-bit linear coding, the ratio between maximum and minimum signal
>level is 255:1.
>
>The typical RAW data is 12-bit or 14-1bit, having a ratio of max/min of
>4095:1 or 16383:1.
>
>With JPEG, taking 2.2 as the typical gamma correction, the ratio is
>255^2.2:1, or about 200,000:1.

David, I don't quarrel with your math, but could you help me
understand the Logic applied here? I accept the 8 bit yielding 255
levels, and same for 12 bit being 4095 levels and 14 bit being 16383.


But what is the principle being applied here to raise the JPEG number
to the exponent of the Gamma? Excuse my ignorance, if that's the
problem, but seems to me some sort of exponential treatment (Gamma or
other) might also be applied to the RAW data. Help please.

Olin McDaniel


>Agreed that JPEG compromises on the accuracy with which any particular
>brightness can be represented, but the range of values which can be
>represented is greater with JPEG than RAW.
>
>Cheers,
>David
>
>

To reply by email, please remove "abcd" from Return address
-----------------------------------------------------
"Ignorance is treatable, Stupidity is incurable. Sometimes
the difference is hardly distinguishable, however."
From: SMS on
Bob Donahue wrote:
> I take a lot of pictures at car shows. The digital cameras I've had to date,
> have trouble rendering white cars in direct sunlight. You can't see the
> curves of the body panels, they come out pure white with no shading! Cars
> that are not white come out beautifully. I'm in the market for a new "point
> and shoot" camera. I've narrowed my choice down to the Fujifilm FinePix
> F100fd or the Nikon CoolPix 610. Both of these models are advertised as
> having special "dynamic range" modes. Which one would be my best bet for
> avoiding washed out highlights?

Your first clue sould be to avoid any camera that offers a very high ISO
setting, like the F100fd, which also somehow forgot to include aperture
priority and shutter priority modes.

You should try using aperture priority mode on your existing camera, and
using a smaller aperture (larger number).

If you've narrowed your choices down to those two abysmal models, then
you should just give up. Seriously.

As someone else suggested, try under-exposing a little.

My suggestion would be a Canon A570 IS, and if you can't find one, get
the newer A590 IS, but the dynamic range will be slightly poorer than
the A570 IS. Of course dynamic range isn't your problem here, so it's
really immaterial.

The 570 IS and 590 IS are good bargains as well. Everyone seems to be
buying the super compact cameras these days and overlooking the more
capable compacts.

The older 570 IS has a better movie mode than the newer 590 IS, if
that's important to you.
From: Eric Stevens on
On Tue, 02 Sep 2008 06:08:23 GMT, "David J Taylor"
<david-taylor(a)blueyonder.neither-this-bit.nor-this-part.co.uk> wrote:

>ray wrote:
>[]
>> OK, so you've attempted to show why (theoretically) I was wrong. That
>> still does not prove that the jpeg has more information than the RAW
>> data from which it was derived. Theoretical limitations are one thing
>> - practical applications are another.
>
>Ray,
>
>I'm not saying that JPEG has any more information, simply that because of
>the gamma-correction which is typically used in JPEGs the dynamic range of
>the image is not restricted compared to RAW. The precision is less,
>though, and the RAW data will be best for post-processing.
>
>However, the Fuji F100fd does not offer RAW as a file saving format as far
>as I can see.
>
> http://www.fujifilm.co.uk/consumer/digital/digital-cameras/advanced-compact/finepix-f100fd-119591/Specs
>
As other people have already pointed out, the real problem is over
exposure. I have encountered this problem and have dealt with it by
going to manual settings of the camera and using the histogram feature
of my display to confirm I was not driving too many pixels up to the
high end. Once I had that bit sorted out I beneficially used a flash
for fill-flash, in brilliant day light.



Eric Stevens
From: Dave Martindale on
mcdanielo.abcd(a)bellsouth.net writes:

>>With JPEG, taking 2.2 as the typical gamma correction, the ratio is
>>255^2.2:1, or about 200,000:1.

>David, I don't quarrel with your math, but could you help me
>understand the Logic applied here? I accept the 8 bit yielding 255
>levels, and same for 12 bit being 4095 levels and 14 bit being 16383.

>But what is the principle being applied here to raise the JPEG number
>to the exponent of the Gamma? Excuse my ignorance, if that's the
>problem, but seems to me some sort of exponential treatment (Gamma or
>other) might also be applied to the RAW data. Help please.

The point of RAW is to preserve the unmodified data out of the camera's
A/D converter. You could apply a non-linear transform to it, but
(a) if you kept the same number of bits as the A/D, there is no point in
doing the nonlinear transform
(b) if you store the result in fewer bits, some codes that are different
in the A/D output will become the same in the file, so you're losing
information, which is violating the spirit of raw files.

On the other hand, the point of JPEG is to store something that is not
identical to the original, but looks "pretty good", in much less space.
Thus there is a large incentive to use the power-law encoding, which
gives an intensity range greater than you can see under ordinary
conditions with intensity steps too small to see under most conditions,
assuming that the image is in substantially final form and you aren't
going to alter its tonal scale significantly.

RAW is designed to carry all that the camera originally captured, to
permit as much post-editing as possible - limited only by the physical
capabilities of the sensor and analog electronics. JPEG is a good
space-saving "final output" form, as long as most of the necessary
processing has happened before the conversion to JPEG.

Dave