Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
11-05-2007, 10:00 AM   #1
Senior Member




Join Date: Oct 2006
Location: India
Posts: 162
K10D performance at ISO 1600

Pentax Forum: Is it hype? Or does the K10D do just fine at high ISO? - photo.net
I don't know if posting links is allowed. If it is not, please delete this.

11-05-2007, 10:08 AM   #2
Veteran Member




Join Date: Aug 2007
Photos: Gallery
Posts: 1,563
The writer of that article states the following:

Getting the K10D as close to over exposure at ISO 1600 is the key. You don't want to blow out highlights but you want to get them to within a 1/3 stop of being blown out. Why? the camera does better if you over expose (mid tones) slightly but does a terrible job at more than 1/2 stop under exposed.

In my experience, when shooting in RAW you don't need to overexpose. You will loose dynamic range in the darker areas with JPG which is available in RAW. I wonder what format you is using in his shots.

- Bert
11-05-2007, 11:32 AM   #3
Inactive Account




Join Date: Nov 2007
Location: Oregon
Posts: 1
This agrees with the "Expose to the right" posting on Luminous Landscape.
Expose Right
11-05-2007, 11:43 AM   #4
Veteran Member




Join Date: Jun 2007
Location: California
Posts: 426
QuoteOriginally posted by bymy141 Quote
In my experience, when shooting in RAW you don't need to overexpose. You will loose dynamic range in the darker areas with JPG which is available in RAW. I wonder what format you is using in his shots.

- Bert
You're taking the picture slightly over exposed to reduce apparent noise in the picture.

Even if you shoot in raw, bringing the exposure up via post processing does NOT reduce noise like slightly overexposing the shot would. You would need to use noise reduction software which compromises sharpness details.

However, you can use RAW to dial the exposure back down before making it a JPG. Difference in shadow detail will be minimal in this case and the lower noise achieved is a far better compromise.

11-05-2007, 12:16 PM   #5
Veteran Member




Join Date: Aug 2007
Photos: Gallery
Posts: 1,563
QuoteOriginally posted by AVANT Quote
You're taking the picture slightly over exposed to reduce apparent noise in the picture.

Even if you shoot in raw, bringing the exposure up via post processing does NOT reduce noise like slightly overexposing the shot would. You would need to use noise reduction software which compromises sharpness details.

However, you can use RAW to dial the exposure back down before making it a JPG. Difference in shadow detail will be minimal in this case and the lower noise achieved is a far better compromise.
Ok, let me try and find the logic in the reasoning behing ISO 1600 overexposures.
Thinking about this, I do not agree.

I hope this makes sense...

1. Overexposing (without blowing out the highlights!) will decrease your available dynamic range since you'd need to correct the exposure in PP.
2. Overexposure at ISO 1600 is done to lower the noise effects.
. Since the noise stays the same, increasing the sensors signal (by applying more light = overexposing) increases the signal to noise ratio.
3. Overexposing with light can be done either by:
. a) longer exposure time (shutter speed), or:
. b) larger lens opening (diafragma).
4. Since increasing the Ev value is an electronic amplification of the sensor signal (including the sensor's noise) it does not effectively change the s/n ratio.
. Therefore it would overexpose but not improve the s/n ratio as is intended, you'd only loose DR.
5. Since we were using ISO 1600 in order to be able to either use less diafragma or faster shutter speeds in the first place, overexposing in ISO 1600 (using longer shutter times / larger appertures) brings us back to where we were!

This reasoning leads me to various possible options:
a) Why not use ISO 1250 and expose right instead of overexposing at ISO 1600? It would give you the same s/n ration and faster shooting.
b) Why not use ISO 800, UNDERexpose 1/3 and shoot RAW where you can use the added legroom to achieve the same s/n ratio and give you a stop more faster shooting?

Where do I go wrong with this??
Perhaps I will have to spend next sunday playing again with my camera.
Isn't this a great hobby?

- Bert
11-05-2007, 01:15 PM   #6
Veteran Member
WMBP's Avatar

Join Date: Dec 2006
Location: Dallas, Texas
Photos: Gallery
Posts: 2,496
The use of the term overexposure here is unfortunate. Exposing to the right means "overexposing" only if you make a fetish of the camera's idea of a "correct" exposure.

What exposing to the right really means, is pushing as much data into the top bracket of the shot's dynamic range. Rationale for doing so is provided in the Luminous Landscape article. As I said in the thread over at photo.net cited here in an earlier message, I've done it both ways, and pulling detail down from the highlights produces much better results than pulling detail up from the shadows. Shooting sports, I shoot TAv and add +1/3 EV to bias the exposure to the right a smidge.

Will
11-05-2007, 02:25 PM   #7
Veteran Member




Join Date: Aug 2007
Photos: Gallery
Posts: 1,563
QuoteOriginally posted by WMBP Quote
The use of the term overexposure here is unfortunate. Exposing to the right means "overexposing" only if you make a fetish of the camera's idea of a "correct" exposure.

What exposing to the right really means, is pushing as much data into the top bracket of the shot's dynamic range. Rationale for doing so is provided in the Luminous Landscape article. As I said in the thread over at photo.net cited here in an earlier message, I've done it both ways, and pulling detail down from the highlights produces much better results than pulling detail up from the shadows. Shooting sports, I shoot TAv and add +1/3 EV to bias the exposure to the right a smidge.

Will
Sorry Will, I don't understand.

You are saying that the camera's metering is incorrect, right?
However, I understand that the overexposure is done to lower noise, not to correct the exposure.

I cannot find the Lumious Landscape article you mentioned. There are many, but none tell me how to improve high ISO / low light shooting. Can you point that one out for me please?

Therefore I cannot see anything wrong with my reasoning, since increasing the Ev value is no more than signal amplification of the whole signal. You do not add any light to the sensor. Hence the s/n ratio stays the same.

So, why not lower the ISO value to 1250 and shoot straight.

Attachment 5753

Looking at the DPR graph, there is much more to gain in the dark end. Frankly (and I'm not so experienced as you are) I found recovering from overexposed areas resulting in wrong colour distribution many times.

- Bert

Last edited by bymy141; 11-29-2007 at 09:33 AM.
11-05-2007, 02:54 PM   #8
Veteran Member
WMBP's Avatar

Join Date: Dec 2006
Location: Dallas, Texas
Photos: Gallery
Posts: 2,496
QuoteOriginally posted by bymy141 Quote
Sorry Will, I don't understand.
Bert,

That makes two of us. ;-)

I don't really understand this stuff. Won't say that I don't care -- if I had time and more technical know-how, of course, I would like to understand it. I speak only on the basis of my own experience (specifically, my experienced post-processing photos taken in low-light, as well as my experience fixing a small but pretty regular handful of badly exposed shots) and from info that I've read in articles like the Expose Right article. My practical conclusion is that pushing the histogram to the right provides more useful detail, and having more useful detail means that I can post-process shots more effectively.


QuoteQuote:
You are saying that the camera's metering is incorrect, right?
However, I understand that the overexposure is done to lower noise, not to correct the exposure.
Not, it's NOT a noise-reduction technique. It's a general exposure technique. Happens to help with noise, but I guess I would describe that as a by-product. The idea isn't that you should expose to the right when the light's low. The idea is that you should expose to the right all the time.

Does this mean that the meter is wrong? Not at all. When it's freezing outside, my Fahrenheit thermometer tells me the temp is 32 degrees -- not 0 degrees. Is the F scale wrong? Not if you know what the numbers on it mean. You need to know that the camera's meter is trying to get a medium-gray reading from the part of the picture that you're metering. Change the part of the picture, and the meter changes. Is it wrong for the meter to disagree with itself in this way? Obviously not. Exposing to the right simply works with the meter -- and the histogram -- in order to get the best capture of the data.


QuoteQuote:
I cannot find the Lumious Landscape article you mentioned. There are many, but none tell me how to improve high ISO / low light shooting. Can you point that one out for me please?
Sorry, it was mentioned in a previous post here, a few messages before my last message. Here's the URL again:

Expose Right

Note that the article is about exposure with digital cameras only and it's also about shooting raw. NOTE ALSO that, if you do take this advice to heart, you will almost always end up with shots that initially look washed out/overexposed. Post-processing is an essential part of the expose to the right idea. If your goal is to get photos that look as close to perfect the moment you download them to the computer, forget about this article.


QuoteQuote:
Looking at the DPR graph, there is much more to gain in the dark end. Frankly (and I'm not so experienced as you are) I found recovering from overexposed areas resulting in wrong colour distribution many times.
As I said above, my understand here is not technical. Perhaps Thomas Knoll is wrong (I'm not being sarcastic) or more likely I've misunderstood something basic here myself. I do know that whether the light is good or bad, if I can get the histogram biased to the right -- without blowing highlights that matter -- I can more often end up generating a result that is better than the result I'd get if I had exposed to the center of the histogram. And while I've rescued a depressing number of badly exposed photos in the last couple of years, since I started shooting raw, especially, I find I can do more with a badly overexposed shot than with a badly underexposed one. Unfortunately, because the camera handles lots of light much better than it handles too little light, I have more underexposed shots to deal with than overexposed ones. And finally, if the highlights are blown, they're blown and that's all there is to it. At the wedding I shot two weeks ago, I came out of church with the bride and groom and stupidly forgot to adjust my camera settings. Result: disastrously and irrecoverably overexposed candids of the couple on the church porch. Fortunately, these were not important shots....

Will

11-05-2007, 03:03 PM   #9
Veteran Member
WMBP's Avatar

Join Date: Dec 2006
Location: Dallas, Texas
Photos: Gallery
Posts: 2,496
QuoteOriginally posted by bymy141 Quote
However, I understand that the overexposure is done to lower noise, not to correct the exposure.
I already responded to this once, with noise in mind. But I want to add something -- and I hope this is something non-controversial that we can all agree with. The idea of "correct exposure" is very simple and clear if you're a camera, but if you're a photographer, it's often not simple or clear at all. For the camera, "correct" means +/-0 EV -- meter reading is in the dead center of the scale. For the photographer, "correct" means that it LOOKS right. Everybody knows the classic examples: to the camera, a correctly exposed bridal gown looks gray; correctly exposed dark suits worn by the groom and his groomsmen also look gray.

So I take the camera's meter simply as a tool, like a compass. The compass tells me where north is. IT doesn't tell me if I should GO north.

Will
11-05-2007, 03:21 PM   #10
Pentaxian
Lowell Goudge's Avatar

Join Date: Jan 2007
Location: Toronto
Photos: Gallery | Albums
Posts: 15,400
QuoteOriginally posted by bymy141 Quote
Sorry Will, I don't understand.


Attachment 5753

Looking at the DPR graph, there is much more to gain in the dark end. Frankly (and I'm not so experienced as you are) I found recovering from overexposed areas resulting in wrong colour distribution many times.

- Bert
I am surprised at this graph, as I have done the same thing at 800 and find both the high and low end have almost symetrical shape,

Note also that the slope of the JPEG line is misleading as this changes by about 30% as a function of contrast ratio selected
11-06-2007, 02:28 PM   #11
Veteran Member




Join Date: Aug 2007
Photos: Gallery
Posts: 1,563
QuoteOriginally posted by WMBP Quote
I already responded to this once, with noise in mind. But I want to add something -- and I hope this is something non-controversial that we can all agree with. The idea of "correct exposure" is very simple and clear if you're a camera, but if you're a photographer, it's often not simple or clear at all. For the camera, "correct" means +/-0 EV -- meter reading is in the dead center of the scale. For the photographer, "correct" means that it LOOKS right.
Will
Thanks for your elaborate reactions Will.

But I'm sorry Will, I think your are moving this disussion in the wrong direction.
First also sorry for you spoiling those wedding photos. However, you've probably made it up by shooting many more excellent pictures.

I agree with your statement in general, however in this case you do (!) overexpose also for the human eye, in order to reduce noise.
Noise reduction what it is all about, not having a picture with the right exposure.

Adding 1/3 Ev can be corrected in PP, easily. It is not about composition and exposure versus what the metering system is telling in this discussion. My point is that overexposing for reducing the noise can only be achieved by increasing the amount of light caputered by the sensor. Not by increasing the Ev value of your camera.

And in order to increase the amount of light, you spoil the reason you were shooting ISO 1600 in the first place. You could just as well lower your ISO value somewhat and shoot faster. Same noise levels, however with faster shutter speeds.

When I find the time (I hope next weekend), I'm going to try and setup an experiment with this. After I'm done, I'll post the results.

- Bert
11-06-2007, 03:18 PM   #12
Veteran Member
lol101's Avatar

Join Date: Jul 2007
Location: Garennes sur Eure France
Photos: Gallery
Posts: 900
I posted a comparson between K10 and 30D...

... in answer to a thread on ... VPN out there @ DPR

ISO 1600 shots under tungsten light.

If interested, you can have a look there:

I thought I didn't have VPN but...: Pentax SLR Talk Forum: Digital Photography Review

It was fun doing it and my wife and children are away for tonight....


.... God I'm so bored!!!
11-06-2007, 04:02 PM   #13
Veteran Member




Join Date: Jun 2007
Location: California
Posts: 426
While I respect the time and effort you went into explaining your position with technical details, there are still factors that you did not account for. An important one being the end result that we actually see.

I'm unsure about adjusting the Ev value on the camera amplifying the signal (I've always believed that increasing the ISO setting was the only factor for amplifying the signal), because I've always noticed that my shutterspeed or aperture are what respond to Ev shifts granted I'm not in Auto ISO. So in that regard, it satifies your statement that "reducing the noise can only be achieved by increasing the amount of light caputered by the sensor".

QuoteQuote:
"Why not use ISO 1250 and expose right instead of overexposing at ISO 1600? It would give you the same s/n ration and faster shooting.

Why not use ISO 800, UNDERexpose 1/3 and shoot RAW where you can use the added legroom to achieve the same s/n ratio and give you a stop more faster shooting?
Should note though that a ISO 800 at 1/3 underexposure does not net you the same aperture or shutterspeed of 1600 at proper exposure. You would need 800ISO and one full stop of underexposure to acheive that. Underexposing that far will limit your dynamic range farther, possibly clipping some of your darks, and creates more PP work.

Take this for what it's worth, but taking a shots well underexposed (say a stop or more) corrected via PP, from my experiences with my K10D, still creates a lot of noise and often adds the unfortunate side effect of vertical banding (which would most likely not show on a higher ISO photo at correct exposure).

My understanding is that the sensor generates much more noise in the darker regions than the mid range at high ISO, so overexposing will push the darks closer to the less noisy mid range and therefore less recorded noise in such regions.

You don't have to agree Bert, but I'd be very surprised if your findings disproves that slight overexposure records less overall noise.
11-06-2007, 07:31 PM   #14
Veteran Member




Join Date: Jan 2007
Location: Northern California
Posts: 475
QuoteOriginally posted by WMBP Quote
Bert,

That makes two of us. ;-)

I don't really understand this stuff. Won't say that I don't care -- if I had time and more technical know-how, of course, I would like to understand it. I speak only on the basis of my own experience (specifically, my experienced post-processing photos taken in low-light, as well as my experience fixing a small but pretty regular handful of badly exposed shots) and from info that I've read in articles like the Expose Right article. My practical conclusion is that pushing the histogram to the right provides more useful detail, and having more useful detail means that I can post-process shots more effectively.


Not, it's NOT a noise-reduction technique. It's a general exposure technique. Happens to help with noise, but I guess I would describe that as a by-product. The idea isn't that you should expose to the right when the light's low. The idea is that you should expose to the right all the time.

Does this mean that the meter is wrong? Not at all. When it's freezing outside, my Fahrenheit thermometer tells me the temp is 32 degrees -- not 0 degrees. Is the F scale wrong? Not if you know what the numbers on it mean. You need to know that the camera's meter is trying to get a medium-gray reading from the part of the picture that you're metering. Change the part of the picture, and the meter changes. Is it wrong for the meter to disagree with itself in this way? Obviously not. Exposing to the right simply works with the meter -- and the histogram -- in order to get the best capture of the data.


Sorry, it was mentioned in a previous post here, a few messages before my last message. Here's the URL again:

Expose Right

Note that the article is about exposure with digital cameras only and it's also about shooting raw. NOTE ALSO that, if you do take this advice to heart, you will almost always end up with shots that initially look washed out/overexposed. Post-processing is an essential part of the expose to the right idea. If your goal is to get photos that look as close to perfect the moment you download them to the computer, forget about this article.


As I said above, my understand here is not technical. Perhaps Thomas Knoll is wrong (I'm not being sarcastic) or more likely I've misunderstood something basic here myself. I do know that whether the light is good or bad, if I can get the histogram biased to the right -- without blowing highlights that matter -- I can more often end up generating a result that is better than the result I'd get if I had exposed to the center of the histogram. And while I've rescued a depressing number of badly exposed photos in the last couple of years, since I started shooting raw, especially, I find I can do more with a badly overexposed shot than with a badly underexposed one. Unfortunately, because the camera handles lots of light much better than it handles too little light, I have more underexposed shots to deal with than overexposed ones. And finally, if the highlights are blown, they're blown and that's all there is to it. At the wedding I shot two weeks ago, I came out of church with the bride and groom and stupidly forgot to adjust my camera settings. Result: disastrously and irrecoverably overexposed candids of the couple on the church porch. Fortunately, these were not important shots....

Will
The article explains this very clearly, but in summary, to keep it simple, the luminance histogram equals 8 bits of total exposure from zero (black) to 255 (pure white with no detail), which it does when shooting jpeg, but not RAW. This range is simply derived from using 8 bits to represent the entire range, which results in 256 distinct steps.

The important thing to remember is that with digital, 255 is a hard limit, at this level everything is pure white with no detail, so everything in your scene that is not pure detail-less white has to fit under this level, and everything that is not pure detail-less black has to fit above zero. This is true because the pixels, much like buckets, simply get full at some point (255 in our example) and cannot hold any more photons, so by deifnition, that has to be white. Any more photons (or more exposure) simply get shunted aside and thrown away.

The idea is to use all 256 levels to represent your scene, which will give you all of the dynamic range that the camera has, and will also give you the most number of different levels to represent the different tones in your scene as accurately as possible.

So, if you increase the exposure until the histogram is as far to the right as possible, you can use all 256 steps of the camera's range.

More importantly lets think of the 0-255 range in stops of light:

If you expose to the right and exactly fill the histo from black to pure white, one stop from the top is 128 steps because one stop from the top (255) is 1/2 as much light (the highlights).

Subtract another stop (minus 1/2 of the light) and you get: 128 - 64 = 64, so this stop is only 64 steps.

The next stop is 64 - 32 = 32 for the next range, and so on.

The 4th stop down from the top is 32 - 16 = 16 levels to represent this stop.

As can be seen, in just 4 stops from pure white you are left with only 16 steps to represent the tones in your scene that fall in this range. Fewer steps means that tones that fall in between will not be represented accurately, and since we are talking about the shadows, you quickly get blocked up mid-tones and shadows where the detail that was in the original scene is lost.

Now, with that in mind, what if you shot such that the histogram ended (on the right) at 128? Now the first stop down is 128 - 64 = 64, and all of the scene has to fit in exactly 1/2 of the available steps. 4 stops down now equals 8 steps. Compressing the scene lilke this means that the shadow end of the scene will have little detail.

In terms of noise, the issue is simply the number of photons captured by the pixel well versus the absolute noise that cannot be reduced below a certain level. Exposing to the right simply means getting as many photons into the pixels as possible for a given scene, which maximizes the signal to noise ratio (more good photons in the pixel while the number of noise photons remain the same). If the max number of photons is represented by 255 and the noise equals 1 (it is far smaller than this), then exposing to the right gives you a signal to noise ratio of 255:1, while exposing to 128 gives you 128:1, making the noise look 2X as bad.

I think that the idea behind using 1600 and exposing to the right versus underexposing at 800, is that the added amplification noise of the camera at 1600 ISO is not as bad as the reduction in signal to noise ratio that comes with one stop underexposure at 800 which then has to be bumped up one stop in software during PP. You can test this yourself by shooting the exact same scene exposed to the right at ISO 1600 and then one stop underexposed at 800. Open both images in your editing software and push the exposure up one stop for the ISO 800 shot and compare the noise.

Ray

Last edited by Ray Pulley; 11-06-2007 at 07:44 PM. Reason: Added notes about pixel capacity
11-06-2007, 08:22 PM   #15
Forum Member




Join Date: Sep 2007
Location: Upstate NY
Posts: 77
This is funny, as I was just shooting a show last night and I have found that this technique does indeed limit the amount of appparent noise at ISO 1600.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
camera, dslr, iso, k10d, photography
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
K10d Iso 1600 tarsus Post Your Photos! 5 05-25-2008 09:39 PM
ISO 1600 Performance dylansalt Pentax DSLR Discussion 14 03-24-2008 08:34 PM
K10D at ISO 1600 Canada_Rockies Pentax DSLR Discussion 21 01-02-2008 06:04 AM
K10D at iso 1600 Kemal Pentax DSLR Discussion 15 10-21-2007 12:37 PM
1600 ISO K10D new picture ercole69 Pentax DSLR Discussion 0 11-07-2006 04:26 AM



All times are GMT -7. The time now is 07:39 AM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top