DYNAMIC RANGE LOVES THE 40D!

Scott W wrote,on my timestamp of 17/09/2008 4:24 PM:

Yeah, I know:
http://wizofoz2k.deviantart.com/art/primeval-swamp-02-91534639
is an example. Once again, Superia 400. Not even Velvia
or Astia!

Or if you scan for shadows and correct curve for highlights
like I did in the above example, you end up with detail in all
of it. That's DR compression and is what negative film has
been doing for eons.
"compression", because most srgb monitors and printers have
difficulty showing more than about 5-6 EIs, even though
8-bit colour video cards can *theoretically* show 8.

Absolutely. Why do you think I have a D80 and film?
I *did* such comparisons regularly. And quite frankly,
there is simply no difference. With film, saturation
is easier to accent. With digital, you get less noise
problems. Overall, DR is the same in both. Medium
format is different, though. I still haven't worked
that one out, still trying to get it under control.

I reserve my opinion on this for raw files from the new
crop of dslrs, like the D700 and the 5D2: 14-bit colour
DR is some really serious stuff! If nothing else,
the resulting compression range will be amazing.

Actually, I disagree here. Negative films can do very well,
but need proper placement of exposure in their dynamic range.
Usually this means correct zone system placement, rather than
just the usual "open up 1 stop". Slide film will cover 5-6
EIs easily, which if exposed properly is *more than enough*
for the VAST majority of monitors and printers out there.
Although of course dynamic range compression is less there:
if the scene is more than 6 EIs, you gotta do some trickery
to get slides to cover it.

Exactly. In most cases it's not even high DR, it's just
different DR compression levels and ratios.
Most digital displays use 6-7 EIs at best and that's a physical
limit not easily overcome. Even less for most digital printers.
The workaround is to compress a higher DR into that range.
Which can be done with film or digital, it's just a means
to an end.

 

Gosh! You keep that up and you'll
get moiree, David!


David J. Littleboy wrote,on my timestamp of 17/09/2008 4:33 PM:

Exactly. I go for correct placement of tones in the
zone scale. Works every single time. When I get
the patience to do it, of course...

 

Scott W wrote,on my timestamp of 17/09/2008 4:11 PM:

It can compess 10 EIs into 6-7 in the latest emulsions - that
"5 stops" nonsense is just urban myth based on info gathered
8 years ago.
More than enough to do anything you might like with
a colour monitor or printer that can hardly do 6 EIs
in 99% of the cases.

 

Annika1980 wrote,on my timestamp of 18/09/2008 1:24 AM:

"of course much worse" is not a universal truth,
it's just parroting urban myths. I prefer facts.
You should try looking at those instead of believing
everything you find on the net.
Do you believe this guy:
http://fakechuckwestfall.wordpress.com/
as well?
I don't, but he's funny.

 

I prefer the real one.

http://www.insidedigitalphoto.com/radio-programs/idp-special-report-canon-eos-5d-mark-ii

 

My point was that anything under about 15,15,15 will print to black
and anything over about 240,240,240 will blow out to white on the
print.

BTW, the dynamic range of the digital greatly exceeds film.

 

Seeing that no post processing was applied, it's very impressive.
It's clear there is a greater range of tonal detail. Great example.
Helen

 

Can you give me the name of this film?

Scott

 

You could scan it at 8000 ppi and get 288MP, but you would not have
anymore detail.
You have 72MP of soft pixels, if that floats you boat go for it, me I
got 300MP of very sharp pixels,
I could have 1200MP of soft pixels, but what would the point be.

Scott

 

I got to side with Noons on this one, rare but it does happen. What
is the point in having a range of 0 to 255 if you don't use the whole
rangle? Seems to me if a print driver blows out anything pass
240,240,240 to pure white that is a problem with the print driver not
the image.

Scott

 

My own take on this is that 255,255,255 should always print out as
full white, anything less should not, but how far we can see into the
shadows is going to depend on the technology used to show the image.
So I guess I am saying that the white point show be fixed but the
black point is going to change from output to output.

Even with the same monitor I can see much further into the shadows
when viewing at night then during the day.

Scott

 

And when looking at a good print I can see much further into the
shadows when viewing the print under a very strong light. That's why
when I set my monitor to show me as close to the same as a print I
looked athe monitor in its ideal conditions, which was a dim shaded
room, and the print in its ideal conditions, which was outside in the
garden on a bright day.

 

The whole concept of dynamic range in the way Bret presents it is
extremely flawed.
HDR photos - not the ugly ones but ones with a high dynamic range
without going overboard... Seldom display a range from black to white.
In fact most of mine are in the lower numerical range but still needed
blending of two or more distinctly different tone ranges to produce a
high dynamic range from 40D images.

All Bret's image shows is a high contrast, not a high dynamic range.
http://www.d-mac.info/examples/HDRatdawn.htm is just one example of
how narrow DR images have to be blended together to match what
transparency film is able to get with just a 1/3rd stop of exposure
correction.

After many years making contrast masks for "Cibachrome" printing and
making wide format prints from digital images, I know full well that
digital cameras have some uncanny quirks that will produce seemingly
wide DR shots at close range but magical cannot come close to film
when the distance increases.

 

Conside also Scott, that photo paper itself is not 255, 255, 255. Most
(if not all of it) is a paler shade of beige. Infact much of the "fine
art" paper on the market is quite soft in it's whiteness. This in
itself will make producing a colour at 250, 250, 250 impossible which
is why many of us see "blown highlights" when in reality they are not
blown but are the colour of the paper we use.

 

Noons, 9/19/2008 11:49 PM:

No sir. Detail is detail, sharpness is sharpness. You are mistaking
sharpness for detail. Sharp edges do not detail make. Push that image
up a bit - there is no detail whatsoever in the roofs of the buildings
behind the trees, no hint of corrugations or tiles, or bricks in the
buildng beyond - in fact, almost all surfaces are artificially smooth,
only tonal boundaries are at all sharp. In other words, the MTF sucks.

Given your claimed 'knowledge' of things photographic, I am surprised -
or maybe not.

Colin D.

 

Colin.D wrote,on my timestamp of 21/09/2008 7:05 PM:

are you totally deranged? WTF are you "pushing up" a
downressed jpg for? Do you *seriously* expect it to show
more detail? It's a 1280X850 downress, dickhead!
If you want to see MORE detail, just ask which area and
I'll post a crop of the original FULL SIZE image. Can you
even comprehend what the term "downress" means?
Did you even REMOTELY read what was written
there, you moron?

Dear me, you just showed again what a stupid idiot
you really are, Colin. Must be that high wind in NZ:
straight between the ears, eh?

I'd STFU if I were you: it's really embarrassing
to hear you clop all over your own d**k...

 

David J. Littleboy wrote,on my timestamp of 20/09/2008 9:11 AM:
Not quite correct. And anyways, you only got 8 bits to play
with in that printer driver, that's 8 stops at best *if* the
printer can handle the subtleties in tonality: why shoot
yourself in the foot by creating images with even LESS dr
to start with?

Folks, the REAL problem is that everyone here is talking
different units of measurement without defining what each means!

Printers can *easily* produce images which originally had more than
8 EIs *relative* DR even though most cannot do more than about 6
stops *absolute* range. All that is needed is to compress the
additional DR into what the printer can do!

This is what colour negative films have done for years by compressing
10 stops into 7, while modern slide does 9 into 7 and b&w does 10
into 8. That's the function of the "knee" and "elbow" in the response
curves you see in the technical data. Yes, there are films with
flat curves, but for now just stay with me here: I'm talking the
*general* case!

The *real* trick is to make sure the *final* output media is capable
of displaying the expanded differences in tonality range.

To make it easier to understand, an analogy: make the printer
*separate* EI 5 and 6 in a form that is *clearly* visible in the
final print. It doesn't have to be a *full* 1 EI desnity difference,
it just must be *clearly* visible.
Not easy, and few printers can do it. The cost also rises
proportionally... Requires also all kinds of curve adjustment
and fitting in the drivers and that also costs.

That is also why some new cameras process 14-bit colour even though
we can easily fit 10 EIs into 12-bits: to give them the ability to
slightly manipulate the curves so the different tones can be discerned
in the final jpg output at 8 EIs. Assuming of course a good
quality final output screen, properly adjusted.

That's been the problem all along. Most folks who use film are not
aware of these subtleties of digital image rendering and D-A
conversion. That is why they end up with overly saturated images
with lots of exaggerated noise. Once that learning curve is
overcome there is no difference between what one can achieve
with film or digital sensors, if the output media is kept the
same for both.

Counter case: without any adjustment, try to print with an
enlarger a dslr image and print with a digital printer a
negative image, and THEN compare both side by side! See what
I mean?


Horses for courses, is what is needed. Fit the process
to the target you got, rather than fit the target to
your process.


Witness this for an example of bad digital processing:
http://members.iinet.net.au/~nsouto/photos/latest/5d2 cap.jpg
above is a 100% size 300X200 crop of a 5D2 image available from here:
http://www.dpreview.com/gallery/canoneos5dmkii_preview/
it's this one:
http://a.img-dpreview.com/gallery/canoneos5dmkii_preview/originals/img_0660.jpg
if you care to download and verify by yourself.
It's one of the sample images of the new 5D2.

No: I'm *not* picking on Canon, it's just an example:
I could do the same with a Nikon, Sony, whatever.

Now, I ask you: is that 21MP resolution? Or even good colour
saturation in the full size jpg? Of course not! I really hope
Canon will fix this problem in the Digic IV. Notice how the
"easy to process" edge between the cap and the background water is
quite sharp, yet *on the same plane* the letters in the back of the
cap are completely messed up!
If that is not a *bug* in the Digic IV software demosaic and sharpen
algorithms, I don't know what is!


I can do MUCH better rez than that with film, and I have.
Here is the demonstration:
http://wizofoz2k.deviantart.com/art/colourful-rescue-98461919
same image type, same sunny conditions, look at the colour
saturation DIFFERENCE! And no, good rez is not synonymous
with bad saturation:
http://members.iinet.net.au/~nsouto/photos/latest/astia full def02.jpg
that is also a 100% size 300X200 of the original 5600X3700
scan. I didn't particularly sharpen this one, although of course
I could if I wanted to: that is not the purpose of this.

But even without it, - unfair to film given the Digic IV is clearly
over-processing *selective* areas of the jpg - I can still resolve
text at much SMALLER size than the one in the cap in the 5D2 crop!

Same size, same 100%, same original image dimensions, same
21MP theoretical resolution, same jpg compression in both cases.
Film shows very small detail, the 5D2 does not: simple as that!

The 5D2 is also *clearly* much *worse* in colour saturation - both
photos were taken in full sun - in resolution and even in sharpness
of an already ultra-processed final jpg:
http://members.iinet.net.au/~nsouto/photos/latest/5d2 oof.jpg
(Ye gawds! *Who* let this one out?!...)

I hope someone alerts Canon to the obvious problem they have with
this incarnation of the Digic IV: it's obvious it's got its knickers
in a twist in the colour curve processing and demosaic and sharpen
software.

Not to say they won't fix it: I fully expect in a year's time we'll
be seeing full rez, fully saturated colour photos from the 5D2 with
reasonable sharpness even at full size crops.

It will never be totally sharp, that's also why Bret's 100% original
crop was so soft. That is NOT a problem of the software or lens:
it's just the way demosaic works in Bayer sensors. Or ANY d/a
sensor for that matter, including scanners!

But for now it remains a perfect example of the problems of
fitting correct DR into a limited media and then "re-fitting"
it all over again in the final output media. A common problem
with ANY digitized image, no matter what sensor was used to
make it. Which is the point I wanted to make.

Now, to bed: tomorrow is another work week!

 

Colin.D wrote,on my timestamp of 21/09/2008 7:21 PM:

No, I *really* needed you to alert me to that...

Try Acros at 50ASA with 8 minutes versus Acros at 100ASA and
9 minutes, both in DDX, then come back to me with that
"modern films are not usable in the zone system" and "everything
has to be half-way up the curve"! And BTW: the *whole* point
of the zone system development control is to *move* the curve,
not the exposure!

Cripes, Colin: you really don't know what you're talking
about, do you? Spend some time in apug and learn, man.
Instead of just spurting the usual dslr marketing moronic
utter *crap*!

Really? I'd like you to explain that to all the apug
members. Let me see: they are all ignorant idiots while
you are the only "genius". Right?

Yes, it does. That is however not the case with you...

 

Noons, 9/22/2008 12:15 AM:

Reading between the inevitable ad hominem remarks - your trademark - it
was you who posted that image saying:

Try this for REAL detail:
http://wizofoz2k.deviantart.com/journal/20548136/
LOL!

Why would you post that boast about film and provide the link if you
were going to show an image that was not capable of substantiating your
claim?

and, 1280x850 *is* capable of showing more than that image does, so, as
I said, that image sucks. As do you.

The day you're embarrassed by anything will be the day, Noons. You're a
bullshit artist of the first order, a champion of the art. To
paraphrase the old saying, you have no science so you try to baffle with
bullshit. Try you might, but few here would be baffled by you. and yes,
in case you missed it, that was an ad hominem attack, just so you might
recognise another one when it comes along.

Colin D.

 

Noons, 9/22/2008 1:27 AM:

Ah, yes, I was a bit slack there; till now the conversation has been
around 35mm film machine processed, as in color negative film, and I
wasn't considering home processing.

However, unless you are processing sheet film individually, exposure by
exposure, my comments still apply. You cannot fully use the zone system
if you are shooting roll film, unless you have a magic way of varying
the development frame by frame, which I doubt even you could perform.

Not to forget that the original zone system strictly applies to
monochrome film. Very limited development variation can be applied to
color film without color shifts.

Unless Apug members are shooting cut or sheet film one at a time, as did
Ansel Adams, then they cannot adjust individual process times either.

I have reason to doubt your reasoning powers, Noons. FYI, knowledge is
a continuum, there is a huge gulf between an ignorant idiot and a
genius, and most of us are somewhere between the two, including you and
me, and everybody else you take issue with. Only difference is, you
appear to be closer to the former limit.

See my para. above ...

Colin D.