More on Canon Rebel XT noise at high ISO - 2 main new data points

Today Stan Birch commented courteously on the subject at hand

Thanks, Paul. What you talk about here is in a group of up
to 5 different processing parameters, two are fixed and
three I can vary. The settable functions are contrast,
sharpness, saturation, color, and something else, but I
get the idea. I've ran some very quickie tests varying
contrast and sharpness and my images look best at the
default for what is called Parameter 1, which is one step
more contrasty and one step more sharpness than "neutral".

These settings are not all that critical, since the very
same adjustments that can be done in the camera can all be
done with the post-processing software *IF* you shoot RAW.

Dunno what version of Digital Photo Professional came with
your camera, but the current (?) version 2.0.3 will cover
95% of your post-processing needs. If you don't have it,
download it from the Canon website. Aside from the
in-camera parameters DPP allows you to extend those numbers
way beyond the the in-camera numbers.

I started out using JPGs, since that had worked out very
well with my first point and shoot. But after trying
shooting RAW, I'd never go back to shooting JPGs. I tend to
do at least a little tweaking for anything that's going to
be printed; and if you don't want to do a whole lot of
fiddling, you can batch-process the entire bunch of photos
to JPGs; but most often with paramter changes will be
applied on a group by group basis. I rarely use Photoshop
for anything any more.

Haven't opened the CDs yet, never intend to. If PSP 9 can't do
it, I don't want it.

There is simply no way, no how, for no reason I will /ever/
shoot in RAW. You like it, I'm happy for you.

--
ATM, aka Jerry

"I disagree with what you say, but I will defend to the death
your right to say it." - Evelyn Beatrice Hall under the
pseudonym Stephen G. Tallentyre

Today Floyd Davidson commented courteously on the subject at
hand

Thank you.

All Things Mopar wrote:
I do /not/, not, not! expect perfect exposures! That just
is never gonna happen. What I /hope/ for is /reasonable/
exposures, within +/- 1 f/stop, and /reasonable/ exposure
consistency. What

If you can get /reasonable/, you can get perfect. First
you need to understand the use of your tools and how it
applies to the situation at hand though...

either the LCD and/or the histogram (yes, Ed, and others,
the LCD will show a small histogram, and yes, guys, I at
least know enough that what I want is a "mountain" and not
a sliver of light at one end or the other).

Use the histogram to get close. Then bracket your shots.
Your film is cheap, don't conserve it.

Iffn my 1st Programmed Auto or full manual exposure is
under by more than a stop or 2, I know in advance that
color balance will be poor tending toward yellow-red from
thee incandescent ambient light, and I know that the
background and car shadows will be noisy. So, with flash, I
will up the /flash/ EC (or EV if you prefer) by 1, 2, or 3
"EV" - i.e., stops - and take 1, 2, or 3 more shots of the
same view. Digital is free, I have plenty of memory and
plenty of battery.

Exactly. But don't bracket in 1 stop steps, get within 1
fstop and bracketed by 1/3 stop, or whatever you find is
close enough to call "perfect". (I'm not familiar with
your camera, so I don't know what can be done
automatically. You might consider a different camera if
yours cannot be set up to fire off a sequence of bracketed
shot at the necessary steps. And flash units with battery
packs that can recharge fast enough are also obviously
needed... if work rate is important.)

My goal is typically to get a couple hundred shots in a
couple of hours, and separate the really good ones from the
OK ones that'll need some work, from the unsavable ones.
And, my goal /always/ is to learn from what last worked and
what last failed, so that I can apply that new experience
next time and improve my ability to use judgment to
/predict/ how to set the camera and flash for best possible
results.

True.

But, the relatively trivial points I've made above aside, I
don't see any discussion in this article of the fix for
your problem. And frankly it seems *obvious* what it is!
Go back and read all of that discussion about over and
under exposure of variously unevenly light parts of a car.
Your problem is not *camera* adjustment, it's inadequate
lighting.

You need at least one more flash (and maybe two or three
would be better). Set it up on a tripod with a remote
trigger (either wireless or optical). (I don't know what
kind of crowds you have to deal with, but this could end up
requiring one or more assistants to hold, or protect, you
lights.)

That is a start, but just as significant is that these
flash units need to be greatly diffused in some way.
You've mentioned a virtual lack of walls and ceilings in
all locations, so that cannot be part of the solution (and
since it would be inconsistent, it would be the least
desirable solution anyway).

You want to look at diffusion techniques for the flash
units. It sounds very much as if two or three flash units
mounted on tripods, with something like Stoffen diffusers
mounted on them and an umbrella or similar reflecting
device, would get predictable and repeatable lighting that
is appropriate.

--
Floyd L. Davidson
http://www.apaflo.com/floyd_davidson Ukpeagvik (Barrow,
Alaska)




--
ATM, aka Jerry

"I disagree with what you say, but I will defend to the death
your right to say it." - Evelyn Beatrice Hall under the
pseudonym Stephen G. Tallentyre

Chrlz wrote:
Jeremy..just blindly relying on
Jeremy auto-exposure.
Jeremy
Jeremy I think we know by now how seriously to take magazine
Jeremy "reviews"...

MoparHe wants it, he's gonna get it...

Nah, mopar, you don't have an attitude problem. No way.

MoparOK, numb nuts, why the hell do you think there is full auto
Mopareverything on a Rebel XT?

What a nice man...!

Call me numb nuts too - HERE'S WHY 'there is full auto everything':

- AUTO mode is for when you hand the camera to grandma or grandpa
(forgive the stereotype putdown, I'm a grandpa (just) myself!), or
to
someone who just wants a fair chance of getting a decent picture
from
an 'easy' scene.

- AUTO is also for more professional people, who will use it ONLY in
situations where they *know* that the conditions are not going to
exceed that particular camera's auto program 'vagaries' or
limitations. Every camera is a little different, and it takes your
average good-learner, I reckon, anywhere between a week and a few
months to get the hang of it. Once that time is up, they will know
if it has problems with highlights/backlighting/spotlighting, with
its meter's averaging methods, with its flash metering (which is
often quite different), and so on. At that time, the *good* camera
driver knows when to slip it into manual. Just like a good driver
will get the hang of his/her car, and know when to back off, when to
turn off the cruise control, when to override the Automatic gear
selection ....

And every other DSLR... not for me,either.

No, nothing ever will be, by the sounds of it. But most of us
somehow
manage to live reasonably happily within our equipment's somewhat
flawed, but (and here's where I *really* differ from you) *highly
predictable* behaviour. We still have never seen an example *of the
unpredictability* you claim, and if it's predictable and therefore
explainable, it's almost certainly solvable. I will continue to
maintain that a simple solution is to use manual flash using the
old-fashioned guide number method, and it truly isn't difficult.
(It
sounds as though you simply gave up on that after you did it
wrong..)
Given a bit of practice, you would almost learn the numbers off by
heart after a few shots.

And I will continue to maintain that NO automatic exposure system,
TTL
or otherwise, will reliably handle shiny cars in dark environments,
with flash as the main source of light.

amateurs who belive (sic) the bull**** that 8+ mega pixels is what
they need for 4 x 6 prints

Can you post a reference to who says that? I say 5 is *just*
enough.
Maybe 8 with a bit of cropping.. (O;

I /can/ look at the histogram, but why?

There's your problem illustrated again right there. It's to tell
you
when you have under- (or over-) exposed. It also helps identify
clipping, but it seems that is not a big problem.. yet. There are
myriads of references on the web on how to read/use histograms, eg:

http://www.luminous-landscape.com/tu...stograms.shtml

That's the first one I found and it's not that bad, even if it
doesn't
cover cars in museums...

I don't display "correct" histograms as wallpaper.

And it seems to be clear that you won't ever get properly-exposed
images to use as wallpaper either, except perhaps by dumb luck.

That 'dumb luck' comment was the only insult contained in this post.
Go on, check. Yes, I've been mildly sarcastic and holier-than-thou
(as usual), but try arguing the *points* instead of attacking the
person. If you can't dispute what I say in a logical fashion, then
you lose. And as for this lame 'I won't post examples except on
groups' rubbish.. give me a break. I'll happily click on a link,
but
I'm not wandering off to other binary groups to wade through images
there.


What a pity: if you were to go to alt.binaries.pictures.autos and look
at the thread "Midget racer in speed shop", you'd see the next step,
which I expect will explode on this scene very soon.

That's it from me, unless we get some meaningful examples posted
somewhere reasonable.


Horses, water, etc....

Today Frank ess commented courteously on the subject at hand

Chrlz wrote:
Jeremy..just blindly relying on
Jeremy auto-exposure.
Jeremy
Jeremy I think we know by now how seriously to take
magazine Jeremy "reviews"...

MoparHe wants it, he's gonna get it...

Nah, mopar, you don't have an attitude problem. No way.

MoparOK, numb nuts, why the hell do you think there is
full auto Mopareverything on a Rebel XT?

What a nice man...!

Call me numb nuts too - HERE'S WHY 'there is full auto
everything':

- AUTO mode is for when you hand the camera to grandma or
grandpa (forgive the stereotype putdown, I'm a grandpa
(just) myself!), or to someone who just wants a fair
chance of getting a decent picture from an 'easy' scene.

- AUTO is also for more professional people, who will use
it ONLY in situations where they *know* that the
conditions are not going to exceed that particular
camera's auto program 'vagaries' or limitations. Every
camera is a little different, and it takes your average
good-learner, I reckon, anywhere between a week and a few
months to get the hang of it. Once that time is up, they
will know if it has problems with
highlights/backlighting/spotlighting, with its meter's
averaging methods, with its flash metering (which is often
quite different), and so on. At that time, the *good*
camera driver knows when to slip it into manual. Just
like a good driver will get the hang of his/her car, and
know when to back off, when to turn off the cruise
control, when to override the Automatic gear selection
....

And every other DSLR... not for me,either.

No, nothing ever will be, by the sounds of it. But most
of us somehow manage to live reasonably happily within
our equipment's somewhat flawed, but (and here's where I
*really* differ from you) *highly predictable* behaviour.
We still have never seen an example *of the
unpredictability* you claim, and if it's predictable and
therefore explainable, it's almost certainly solvable. I
will continue to maintain that a simple solution is to use
manual flash using the old-fashioned guide number method,
and it truly isn't difficult. (It
sounds as though you simply gave up on that after you did
it wrong..) Given a bit of practice, you would almost
learn the numbers off by heart after a few shots.

And I will continue to maintain that NO automatic exposure
system, TTL
or otherwise, will reliably handle shiny cars in dark
environments, with flash as the main source of light.

amateurs who belive (sic) the bull**** that 8+ mega
pixels is what they need for 4 x 6 prints

Can you post a reference to who says that? I say 5 is
*just* enough. Maybe 8 with a bit of cropping.. (O;

I /can/ look at the histogram, but why?

There's your problem illustrated again right there. It's
to tell you when you have under- (or over-) exposed. It
also helps identify clipping, but it seems that is not a
big problem.. yet. There are myriads of references on the
web on how to read/use histograms, eg:

http://www.luminous-landscape.com/tu...nderstanding-s
eries/understanding-histograms.shtml

That's the first one I found and it's not that bad, even
if it doesn't cover cars in museums...

I don't display "correct" histograms as wallpaper.

And it seems to be clear that you won't ever get
properly-exposed images to use as wallpaper either, except
perhaps by dumb luck.

That 'dumb luck' comment was the only insult contained in
this post. Go on, check. Yes, I've been mildly sarcastic
and holier-than-thou (as usual), but try arguing the
*points* instead of attacking the person. If you can't
dispute what I say in a logical fashion, then you lose.
And as for this lame 'I won't post examples except on
groups' rubbish.. give me a break. I'll happily click on
a link, but I'm not wandering off to other binary groups
to wade through images there.


What a pity: if you were to go to
alt.binaries.pictures.autos and look at the thread "Midget
racer in speed shop", you'd see the next step, which I
expect will explode on this scene very soon.

That's it from me, unless we get some meaningful examples
posted somewhere reasonable.

**** Off, Frank Ass! You've contributed no-thing, and I mean
NO-THING, in /ANY/ news group you've ever visited except
smart-ass crap, like here. Yes, look in Disneyland, asshole,
and keep looking, you'll see something when you ain't looking
that'll curl the pixels on your monitor. You should stick to
your buds in the kooks NG, they're more your speed.

--
ATM, aka Jerry

"I disagree with what you say, but I will defend to the death
your right to say it." - Evelyn Beatrice Hall under the
pseudonym Stephen G. Tallentyre

All Things Mopar wrote:

There is simply no way, no how, for no reason I will /ever/
shoot in RAW. You like it, I'm happy for you.

Then why come to a newsgroup like this one and ask for help?

If you won't do what works, why pester people with questions?

--
Floyd L. Davidson http://www.apaflo.com/floyd_davidson
Ukpeagvik (Barrow, Alaska)

Today Floyd Davidson commented courteously on the subject at
hand

There is simply no way, no how, for no reason I will /ever/
shoot in RAW. You like it, I'm happy for you.

Then why come to a newsgroup like this one and ask for
help?

If you won't do what works, why pester people with
questions?

Floyd, please enlighten me and the rest of the world - how does
whether I choose to shoot in RAW or not have anything whatsoever
with a request for help? What annoys me the most about people
like you, is that they are hell bent for leather to show off
/their/ knowledge, and not really help people. Again, if you
like creating 150MB files, then do whatever floats your boat.
Or, just go away until you learn how to play well with the rest
of the boys and girls.

--
ATM, aka Jerry

"I disagree with what you say, but I will defend to the death
your right to say it." - Evelyn Beatrice Hall under the
pseudonym Stephen G. Tallentyre

OK well we at least determined your noise isn't all that bad. Noise
looks worse on a monitor than in a print. This was not as bad as noise
can be at ISO 1600 on a D70. Yes it was available (sodium? bleck)
streetlights, handheld at 1/3 second, Auto white balance & I cooled that
down in the raw converted version. I also reduced contrast, applied
noise reduction removed sharpening till after & even then used less
sharpening, reduced brightness & only a bit on the shadows slider. And
the photo was shot at +1EC to overcome the camera meter's reaction to
those hot highlights.

PS I'm not going to any binaries newsgroup to see pictures, get a web
site if you want to share. It probably would be interesting to plenty of
car buffs.

PPS Oh & I used a wide angle lens to exaggerate the proposterous
proportions of that old Caddie. Not a great pic, but relevant to the
discussion.

All Things Mopar wrote:

Today Paul Furman commented courteously on the subject at
hand


Paul Furman wrote:

ATM, check this out:
http://www.edgehill.net/1/?SC=go.php...c/photography/
car-iso-1600 Is that similar to what you are getting?

Well?
I'll post some pics for you if it helps.


Paul, I thought I'd replied to you last night. You've been a
refreshing voice of reason in a world gone wild with my simple
questions and assertions.

I did look at this link and it isn't what I see. Similar in many
ways, but not the same. Not for you, but fortunately for me, the
night scene in this link is far worse than the more minor noise,
underexposure bad color balance, and noise. Here's a copy-paste
of what I said in reply to you last night:

The image in your link, Paul, appears to have been shot
available light. I am certainly not criticizing to insult you,
but it is soft, looks like it was taken with daylight WB, and is
quite noisy.

My ISO 800 and 1600 noise, fortunately, isn't this bad. However,
if /I/ manage to create and underexposure (it was pointed out to
me that it is me, and not the camera's fault grn), then I do
get really bad noise.

Somebody yesterday said "99% of all noise is caused by
underexposure". I would modify that quote this way "Severe
underexposure will cause some degree of noise 99% of the time",
which isn't at all the same thing.

Let me describe a typical shooting situation of a typical car in
a typical museum:

The car is behind short barriers and there are other cars all
around, which limits how far back I can stand and often limits
the views of the car I can reasonably get. There is /never/ a
low ceiling that I could bounce flash off. There is seldom even
one nearby wall for bounce, much less two walls. And, the
background is usually more exhibits with the closes wall 20, 30,
50 feet away. In the foreground are typically light colored
barriers and one or more signs.

Now, by whatever means, suppose I am able to get a "correct"
exposure of the car. First, if the I am shooting the car
anything other than straight-on, I know that lighting will vary
from too bright nearest the camera, somewhere about right in the
middle, and took dark/underexposed near the rear of the car,
because light from a flash falls off as the /square/ of
increased distance. So, it is not uncommon for the fender to be
2 f/stops brighter than the far end of the rear quarter panel.

Now, some parts of the car, maybe the fender closes to me, will
overexpose and blow out the highlights, there may be flash glare
in the door glass, and on the lower parts of the car, shadows
may form creating noise in them. And, across a shot taken at a
45 degree angle, lighting will be very uneven.

Now, if there's a big sign of some small exhibits in front of
the car, they'll be overexposed, sometimes with highlights blown
out to white. And, the entire background, what can be seen of
the high ceiling, exhibits to the left and right, and the
nearest wall behind the car, will all be underexposed by 4-6
f/stops. And, all this dark stuff will be too yellow/red because
the high color temp flash pulse doesn't reach that far.

So, I strive to expose for the bulk of the key part(s) of the
subject, and just let the highlights, shadows, and background do
what they will.

Available light is a different challenge. However much
"inherent" noise anyone wants to believe that a Canon Rebel XT
with kit lens will produce at ISO 1600, "correct" exposure /is/
possible, but hardly across the entire scene, including car,
foreground, and background. First, ambient light is very low and
very uneven. Typically, there are high spot lights shining on
parts of the car and other exhibits in the area. Maybe even some
fluorescents to mess up the color and maybe some windows to
provide our fav problem - backlighting. The overhead spot lights
will produce streaks, blobs, and bright oval white spots all
over the dark paint of the car. And, the best one can expect to
achieve is a relatively even exposure, prefably one stop over,
and no more than 1 or 2 stops under throughout the entire scene.

Not.

I'm going back to the WPC Museum this morning for yet another
try with my current treasure trove of new knowledge and give it
a go. I'll try "normal" flash issues with my newer Canon 430 EX
external flash which has enough range to cover anything in that
museum. And, I'll try hand-held available light /with WB set to
Tungsteen/ at both ISO 800 and 1600.

I do /not/, not, not! expect perfect exposures! That just is
never gonna happen. What I /hope/ for is /reasonable/ exposures,
within +/- 1 f/stop, and /reasonable/ exposure consistency. What
I mean by that last, is if I take 6 shots of one car such as
front 3/4, side, rear 3/4, full front, full rear, and a detail
shot of the front alloy wheel, I'd like to not have to
individually tweak every blinking shot and experiment to get
either the LCD and/or the histogram (yes, Ed, and others, the
LCD will show a small histogram, and yes, guys, I at least know
enough that what I want is a "mountain" and not a sliver of
light at one end or the other).

Iffn my 1st Programmed Auto or full manual exposure is under by
more than a stop or 2, I know in advance that color balance will
be poor tending toward yellow-red from thee incandescent ambient
light, and I know that the background and car shadows will be
noisy. So, with flash, I will up the /flash/ EC (or EV if you
prefer) by 1, 2, or 3 "EV" - i.e., stops - and take 1, 2, or 3
more shots of the same view. Digital is free, I have plenty of
memory and plenty of battery.

My goal is typically to get a couple hundred shots in a couple
of hours, and separate the really good ones from the OK ones
that'll need some work, from the unsavable ones. And, my goal
/always/ is to learn from what last worked and what last failed,
so that I can apply that new experience next time and improve my
ability to use judgment to /predict/ how to set the camera and
flash for best possible results.

I will be posting finished pictures from The Henry Ford Musuem
and The Walter P. Chrysler Museum to alt.binaries.pictures.autos
within a month. You can easily tell which are mine from my All
Things Mopar handle. And, if people are truly interested, I can
post a /sampling/ of the unedited JPEGs to any NG of choice that
I can get with GigaNews.

Paul, and others, my belief is that learning is a life-long
process, so I never stop trying to improve my techniques for
daylight, available light, or flash photography. The reason I
recoil against Photog 101 lessons is that while I /understand/
the basics, putting theory into practice in an actual challengin
g museum shoot takes more than just RTFM and more than reading
magazines and more than looking at other people's successes.

So, for you Paul, and for any others who want to approach these
issues sanely, rationally, factually, and specifically to my
camera and to my peculiar type of museum photography, I would
highly encourage continued discussion under a new thread.

Perhaps later this week or next week, I'll attempt to start a
new thread on one, two, or three particular situations I found
myself in, what I tried, what worked and what didn't, and ask
for opinions as to how I could have done better. If we /all/
park our egos at the door, things will go well. If we /all/
assume that the other guy is sincere, honest, and reasonably
bright, things will go quite well. And, if we stay on-topic and
avoid picking spots off the backs of gnats, we'll all get along
well, and thus we'll all learn from each other.


--
Paul Furman
http://www.edgehill.net/1
Bay Natives
http://www.baynatives.com

All Things Mopar wrote:

Today Alan Browne commented courteously on the subject at
hand

Wherever, just say where and put a flag in the subject such
as [ATM350] so we can find them. I suggest that you post
the photos you were having trouble with as well as the new
and improved. OTOH, if you believe you're out of the woods
on this issue, don't go to any particular trouble. [We
reopen tomorrow, so I'll not have as much time for this...]


Alan, anybody who wants to can simply filter me out, do a simple
plonk, or just ignore anything with "ATM" on it. It ain't like
that is difficult to comprehend and do.

No, I don't think I'll post the total crap, the later crap, the
almost crap, the new & improved crap, and what I think is OK
crap. I bought the camera for /me/, not you. Now, you want to
see what it can do? Wait about a month, then take a look at
alt.binaries.pictures.autos for my car posts. It should give all
you guys major grins to laught silently or out loud on what a
flaming, incompetant asshole I really am. So what?


I was really interested in the problem you and/or your camera is having.
I might even be able to drop a word of advice. I might even learn
something. I've done nothing to merit your tone above and I resent it
very much.


--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.

On Mon, 02 Jan 2006 23:32:27 GMT, wrote:

In message ,
Rich wrote:

Popular Photography magazine characterized the Rebel XT noise
level at 1600 ISO as unacceptable.

What did they do; shoot a black frame and then use histogram
equalization?

That's very funny.
But they seem pretty picky when it comes to noise.
They characterized the Panasonic P&S cams as too
noisy past 80 ISO.
-Rich

On Mon, 02 Jan 2006 23:38:56 -0000, Jeremy Nixon
wrote:

wrote:

Popular Photography magazine characterized the Rebel XT noise
level at 1600 ISO as unacceptable.

What did they do; shoot a black frame and then use histogram
equalization?

They probably shot JPEGs using the LCD review to judge "proper" exposure.
Or even just blindly relying on auto-exposure.

I think we know by now how seriously to take magazine "reviews"...

But honestly, what exactly do you test for when analyzing noise?
Most scenes have a wide latitude of colours and illumination levels,
some of which will likely show some noise.
-Rich