Why bother changing base ISO?

Started Jul 18, 2007 | Discussions
Julia Borg Veteran Member • Posts: 7,280
Re: uniWB histrogram all the time

Pierre Sottas wrote:

Julia Borg wrote:

only as an option. in many cases it is impossible to held highlights
and to get decent clean image in midtones, much less in shadows.

it should be ok for scenes with less than 10 stops of DR.

given most cameras have 3 stops useful DR at low light? oh I doubt your application is typical.

-- hide signature --

Julia

Julia Borg Veteran Member • Posts: 7,280
Re: Dear Jeff

Raul wrote:

I find that that uniWB histogram suits me better than the others. I
use RawMagick and I believe that I am not loosing any highlights
there.

it is another histogram limitation. it does not show clipping correctly. at least 2/3 stops after hitting the right wall can be recovered without any damage. means one can expose to the right and then add 2/3eV overexposure

-- hide signature --

Julia

Julia Borg Veteran Member • Posts: 7,280
Re: uniWB histrogram all the time

Derouyag wrote:

Could this be done with a menu option and an upgrade to the firmware
(rather than a new camera)?

for current cameras this can't be done through firmware upgrade.

-- hide signature --

Julia

Raul Veteran Member • Posts: 8,434
And if I am

I will sell all the stupid equip (including the color meter) and go to work with Le cirque du soleil.

Regards

Raul

 Raul's gear list:Raul's gear list
Nikon Df Nikon D5 Nikon D850
Derouyag
Derouyag Contributing Member • Posts: 824
Re: uniWB histrogram all the time

I think I would like it in all modes. I mostly use spot metering, rather than matrix.

I would just like to have an option in the menu or during playback to show what we currently have for histograms and what UniWB would have.

Now that I think about this a little further, the histogram is probably based solely on the picture being displayed. The cameras computer probably calculates the picture information and displays the histogram based on the WB (as shot). That would mean, in order for it to display the UniWB equivalent, the in camera computer would need to reverse the WB calculations used to create the picture in the first place.

It sounds like this would require knowing more about the cameras workings/processing to determine if it were even possible from a "switch in a menu."

It would be nice... but I don't know if it is as simple as it sounds.

Pierre Sottas wrote:

Derouyag wrote:

Could this be done with a menu option and an upgrade to the firmware
(rather than a new camera)?

this should be very easy to implement since this is a very special
case of the full matrix metering mode.

Pierre

 Derouyag's gear list:Derouyag's gear list
Nikon D200 Nikon D810 Nikon AF-S Nikkor 24-70mm f/2.8G ED Nikon AF Nikkor 50mm f/1.4D Sigma 70-200mm F2.8 EX DG Macro HSM II +7 more
Raul Veteran Member • Posts: 8,434
Dear Ms Borg

so: do I or do not go to Le Cirque?

Very kind regards

Raul

Julia Borg wrote:

Raul wrote:

I find that that uniWB histogram suits me better than the others. I
use RawMagick and I believe that I am not loosing any highlights
there.

it is another histogram limitation. it does not show clipping
correctly. at least 2/3 stops after hitting the right wall can be
recovered without any damage. means one can expose to the right and
then add 2/3eV overexposure

 Raul's gear list:Raul's gear list
Nikon Df Nikon D5 Nikon D850
Julia Borg Veteran Member • Posts: 7,280
Re: uniWB histrogram all the time

I mostly use spot metering

then you do not need that mode at all. just meter from important highlights you want to keep and dial 2 to 3eV plus.

Now that I think about this a little further, the histogram is
probably based solely on the picture being displayed.

currently, yes.

in order for
it to display the UniWB equivalent, the in camera computer would need
to reverse the WB calculations used to create the picture in the
first place.

no. histogram should be of raw data, before creating jpgs.

-- hide signature --

Julia

Fotomat Senior Member • Posts: 1,220
Re: There's no red data in the sky

D-200
typical shot

I want to try with black sky (different subject/scene)

shot with fuji S3 but typical shot..

Thanks!!
--
Fotomat

Jeff Folkins
Jeff Folkins Contributing Member • Posts: 853
Max uniWB value?

Julia

I am confused. Are you saying that although the UniWB equalizes the channels that a value to the far right doesn't correspond to the maximum exposure value? Why not?

-- hide signature --
Jeff Folkins
Jeff Folkins Contributing Member • Posts: 853
Re: And if I am

Raul
What will you do with the circus? high wire? lion tamer?

-- hide signature --
Julia Borg Veteran Member • Posts: 7,280
Re: Max uniWB value?

Are you saying that although the UniWB equalizes the
channels that a value to the far right doesn't correspond to the
maximum exposure value?

yes.

Why not?

because of the tone curve Nikon uses in cameras, and simplified demosaicing used to produce jpgs.

-- hide signature --

Julia

Tom Christiansen Senior Member • Posts: 2,239
Re: There's no red data in the sky

kevm14 wrote:

Tom Christiansen wrote:

Even a "clear blue California sky" is going to have a red-channel
multiplier. And just how much red signal can there be in a blue
sky anyway? It's the same problem as blue channel in incandescent
light. You have very little signal to start with, boost it 3 stops to go
from ISO 100 to pseudo-ISO 800 equivalent, then further add
distortion by a white-balance coefficient. GIGO, magnified.

But that was a daylight shot, so working from a daylight WB
perspective, the red channel wouldn't have had to be pushed to obtain
the correct WB.

Sure it would.

http://www.pochtar.com/NikonWhiteBalanceCoeffs.htm

If daylight is 5200K, then your red channel gets multiplied by 1.8x
and your blue channel by 1.4x on a D200. That on top of the 3
stops of push from the ISO.

So you have a big 1.8 push on the red channel already (daylight
is mostly green, then blue; incandescent is mostly red, then green),
which you further distort with 3 stops of ISO pushing, and you do
this in an area where there is virtually no red signal at all. That
doesn't seem a good thing.

I'd region-select the sky, then noise-reduce the red channel, maybe
blur it.

-- hide signature --

tom

Julia Borg Veteran Member • Posts: 7,280
Re: And if I am

What will you do with the circus? high wire? lion tamer?

but they have nice free tarts for personnel, that is the main reason I would like to work there.

-- hide signature --

Julia

Raul Veteran Member • Posts: 8,434
Jeff

That was precisely my problem (I am not particularly corageous). But Ms. Borg has helped a lot and I shall be "Raul The Tart Eater"

Regards

Raul

Jeff Folkins wrote:

Raul
What will you do with the circus? high wire? lion tamer?

 Raul's gear list:Raul's gear list
Nikon Df Nikon D5 Nikon D850
cycle61 Senior Member • Posts: 1,550
Re: There's no red data in the sky

Julia Borg wrote:

Red channel definitely suffered the most in this shot, here's the red
channel only for the ISO 100 and 800 crops:

please e-mail me the nefs

On the way. May take 2 or 4 emails.
--
-Nick Davis
Please feel free to critique anything I post. I'm here to learn.

My galleries, such as they may be...
http://www.pbase.com/cycle61

cycle61 Senior Member • Posts: 1,550
Re: There's no red data in the sky

Tom Christiansen wrote:

If daylight is 5200K, then your red channel gets multiplied by 1.8x
and your blue channel by 1.4x on a D200. That on top of the 3
stops of push from the ISO.

So you have a big 1.8 push on the red channel already (daylight
is mostly green, then blue; incandescent is mostly red, then green),
which you further distort with 3 stops of ISO pushing, and you do
this in an area where there is virtually no red signal at all. That
doesn't seem a good thing.

I'd region-select the sky, then noise-reduce the red channel, maybe
blur it.

Tom,

That sounds like good advice for saving an image that's this far gone with noise, if it happens to be in one particular weak channel. Please understand I was not trying to present an optimized result to argue one way or another, rather, some unmanipulated images shot using the OP's theory that underexposure then pushing in post would provide similar results to increasing ISO in camera. If I had wanted this shot to be technically good, I would have used ISO 100, and slowed the shutter down to give me back my three stops.

I've sent Julia the original NEF's, let's see what she comes up with, maybe using a better converter than the DNG/old ACR/CS hybrid I'm dealing with.

-- hide signature --

-Nick Davis
Please feel free to critique anything I post. I'm here to learn.

My galleries, such as they may be...
http://www.pbase.com/cycle61

Tom Christiansen Senior Member • Posts: 2,239
Re: Adobe problems

Julia Borg wrote:

Julia, might you happen to have any idea why Adobe converters
seem to do worse on Nikon noise than Nikon's converters do?

I tried to figure that out. it seems that applying of white balance
to RGB data through matrix transform is introducing more noise then
applying multiplication to individual channel data before
demosaicing. btw using multiplication only 2 channels are affected,
while with chromatic adaptation over RGB data all three channels are
affected and introducing additional cross-noise (noise from weak
channel affects stronger channels during demosaicin, and then,
additionally, during chromatic adaptation).

That makes some sense.

It's a bit frustrating to see people wave ACR on NEFs at you
as "proof" that Nikon is lousy at dealing with noise, rather
than saying ACR is.

RML demosaicing acts separately on luma and chroma; and then restores
exact values of channel data where possible. Adobe way does not
follow yellow brick road, accuracy is sacrificed for speed. repeating
demosaicing each time user changed WB is of course time-consuming.

In NC it drives me crazy on a crawling machine to change the
noise reduction or digital lighting, especially at zoom. Sometimes
I just walk away from the computer. They're pretty clearly trying
to compute the whole matrix, at least for display resolution, every
time you change white balance or contrast or mode. I understand
the tradeoffs. But I still think some caching would help. You should
be able to flip back and forth through history without so much pain.

Does everyone reverse engineer the exact
multipliers needed due to Nikon's idiosyncratic dye-color choices
between bodies by dissassembling Nikon's own converters, or do
they derive these number empirically, and potentially inaccurately?

I used disassembly for my table. however it turned out there are much
simpler methods

Set color temperature. Shoot known neutral. Use spectrograph
if necessary to measure how far from neutral resultant values appear
to be. Now see how far a given white-balance moves those
numbers, and in what direction. Now just apply inverse/reciprocal factors
and call those WB coefficients.

Something like that?

-- hide signature --

tom

Julia Borg Veteran Member • Posts: 7,280
Re: Adobe problems

In NC it drives me crazy on a crawling machine to change the
noise reduction or digital lighting, especially at zoom.

try staying at 50%, at least for WB, eV, DLight

But I still think some caching would help.

yes, but mainly they should have been using openGL and calculating in video processors.

Set color temperature. Shoot known neutral. Use spectrograph
if necessary to measure how far from neutral resultant values appear
to be. Now see how far a given white-balance moves those
numbers, and in what direction. Now just apply inverse/reciprocal
factors
and call those WB coefficients.

if a camera is supported through NC the easiest way is to change WB in NC to Uni. resulting custom coefficients are reciprocal values of WB coefficients the camera. a simple hack of NX allows the same behaviour.

-- hide signature --

Julia

Marianne Oelund Veteran Member • Posts: 7,765
Thank you for the additional data

Bill Janes wrote:

Marianne,

Your analysis is lucid, but I must take issue on one point and I will
present data that expand on your analysis.

Yes, I was trying to provide a simple answer which addressed only one point - tonal resolution.

Thank you for your time and trouble to provide the additional information about noise effects. It's nice to have all of this data collected together in one place, clearly presented as you have done.

cycle61 Senior Member • Posts: 1,550
ISO boost test shots re-processed

Julia was kind enough to re-process the original NEF's for all of us and provide a link to the new fullsize jpeg's. I am now quite certain that I don't know which is better, as I cannot really see a difference in the images. Hopefully there's some more discerning pixel-peepers out there than me. The images are in order from ISO 800 to 100. From Julia:

Hello Nicholas,

Rendered at all defaults through RAWMagick, all noise reduction set to
off.

http://www.pochtar.com/jb/NDavis/_LND6075.jpg
http://www.pochtar.com/jb/NDavis/_LND6076.jpg
http://www.pochtar.com/jb/NDavis/_LND6077.jpg
http://www.pochtar.com/jb/NDavis/_LND6078.jpg

Looks like light was changing a little between the shots, or your
camera has uncalibrated amplifier.

Please feel free to post the links if you wish so.

Julia, thanks again. I'll put up some 100% crops of the images, hopefully that'll add to the confusion

By the way, the shots were all taken within 30 seconds total elapsed time on an essentially cloudless day. Is the amplifier thing something I should look into?

-- hide signature --

-Nick Davis
Please feel free to critique anything I post. I'm here to learn.

My galleries, such as they may be...
http://www.pbase.com/cycle61

Keyboard shortcuts:
FForum MMy threads