| Author | Thread |
|
|
08/16/2007 10:33:11 PM · #26 |
Originally posted by zxaar:
Originally posted by _eug:
Originally posted by rswank:
Originally posted by kirbic:
It's a splendid idea where ... a static scene is being photographed.
|
Could a camera be fashioned so that the image is reflected on two image sensors, one at the "correct" exposure and the other at the 4 stop overexposure done via increasing the ISO to achieve better results |
If you had two sensor you'd need 2 shutters working at 2 different speeds. |
why do you need two shutter speeds, why can't you stop reading data from 'short exposure' shutter. Is there any problem in this approach. |
So you'd be using an electronic shutter. Basically the same thing. You'd still have 2 different timings. |
|
|
|
08/16/2007 10:50:20 PM · #27 |
Originally posted by _eug:
why do you need two shutter speeds, why can't you stop reading data from 'short exposure' shutter. Is there any problem in this approach. |
So you'd be using an electronic shutter. Basically the same thing. You'd still have 2 different timings. [/quote]
sorry I meant to write : 'short exposure' sensor.
I feel even with one sensor it could be done. The larger exposure time to me means larger time for data averaging on a pixel.
Now lets say that take first image data from sensor at time t1, which is my short exposure. and write the data to say, raw1 file (from whatever data i collected from the sensor), now I take data again from the senor at time t2 and write raw2. (which is our large exposure).
Can I use these two raw files to do work further.
I think (I may be wrong) that for the exposure time we keep collecting the information from sensor. As soon as the exposure time is over we could just stop reading the information from sensor, we do not need to close the shutter, let the light be falling on the sensor, even after we stopped collecting the information for first exposure at t1. (and if someone wish the information for larger exposure is still being collected).
(I know that in sony R1, same sensor is used for live preview, so the shutter actually never closes. The exposure to shot is just the data collection time from sensor.)
|
|
|
|
08/16/2007 11:06:37 PM · #28 |
Originally posted by rswank:
That is the million dollar question here.
I would think that the signal to noise ratio of the shadow region of a RAW file from a Canon MKIII would be lower than that of the same region of an image shot +4 EVs.
Hell maybe even better on my noisy D70.
And if not today I would expect it sometime in the near future.
Reason being is that for the first image we're dealing with a maximum potential of only 128 'levels' of data for the blackest shadow regions compared to 2048 'levels' of potential data for those same areas of the image.
(btw, I got this from the luminious landscape article on exposing to the right - but doh! it's also in the OP's link haha)
The other thing unclear is the affect that a different noise pattern would have on doing the image combine.
I see nothing in the OP's link that says different noise patterns would cause a problem to his algorithms. Maybe I'll email him and ask. |
Should test this out. Read below.
Originally posted by SamDoe1:
I have a great idea though, why don't we actually try this and find out? Haha. |
Originally posted by rswank:
I'd love to!
You have a couple of APC sensors to mess around with that you don't mind ruining? :P |
No need to destroy any cameras or sensors... Just do the exact same method of shooting the 4 stop difference. But instead of changing the shutter speed, kick the ISO up 4 stops. Do the same with the shutter speed, but leave the ISO the same for both images.
Check out the difference.
Originally posted by rswank:
I'm really trying to figure out if similar results can be done with a single shutter release as doing 2 so you are not relegated to static subjects and require a tripod.
(I guess you'd also need some crazy ass camera!)
This idea also hinges on the ability to project the same image on two sensors at the same time.
Is this optically possible? I'm sure it is but what would the mechanism be, combination of prisms and mirrors I assume? |
I'm sure it's definitely possible, but I'm no expert in optics so I couldn't tell you how it would be done. I can ask around to my optics friends and see what they say if you want.
|
|
|
|
08/16/2007 11:22:42 PM · #29 |
I don't mean to burst anybody's creative bubble, but all this has very much to do with the intricate workings of cameras. If it were possible (and economically viable) to reduce noise, the manufacturers would figure out how.
Kirbic has spelled it out: you need more photons via longer collection time or larger pixels.
I imagine using a prism to the image into two either degrades quality or is prohibitively expensive.
The notion of two exposures in a short time is interrupted by the problem of collecting all that data off the sensor. As I understand it, this is already a limiting factor.
You're better off figuring out how to get your subject to stay still for two exposures, or rigging up two cameras next to each other. =) |
|
|
|
08/16/2007 11:27:50 PM · #30 |
Well right now the "million dollar question" is whether or not using different ISO settings makes a difference compared with using different shutter speeds.
Now if I remember correctly, if you use ISO 100 for your base image 4 stops up would be 200, 400, 800 right? Or is it 1600? Stupid question I know.
|
|
|
|
08/16/2007 11:44:45 PM · #31 |
Originally posted by SamDoe1:
Well right now the "million dollar question" is whether or not using different ISO settings makes a difference compared with using different shutter speeds.
Now if I remember correctly, if you use ISO 100 for your base image 4 stops up would be 200, 400, 800 right? Or is it 1600? Stupid question I know. |
1600
R.
|
|
|
|
08/17/2007 12:00:06 AM · #32 |
Or use neatimage and cloning to remove the noise?
Edited to say: I mean, the point of this seems to be an easier way to get a noiseless picture, but with everything you're talking about it seems almost like it'd be more of a hassle than just to use photoshop skills to fix it up. Though, I'm far less advanced than you all are so what do I know. Just my humble opinion.
Message edited by author 2007-08-17 00:03:23. |
|
|
|
08/17/2007 12:02:08 AM · #33 |
Originally posted by Atropos:
Or use neatimage and cloning to remove the noise? |
Both of those methods destroy pixels in the image that you'd rather not have disturbed. The fastest possible lens and the best lighting will give the best results. Depending on NI, etc., will give you soft planes and oversharpened edges and not all noise will go away.
Message edited by author 2007-08-17 00:05:55.
|
|
|
|
08/17/2007 12:03:35 AM · #34 |
| this thread should be renamed from "Zero Noise Photography" to "Zero Noise Digital Photography" =_=" |
|
|
|
08/17/2007 12:05:40 AM · #35 |
Originally posted by smurfguy:
The notion of two exposures in a short time is interrupted by the problem of collecting all that data off the sensor. As I understand it, this is already a limiting factor.
|
Yes very much. |
|
|
|
08/17/2007 12:55:13 AM · #36 |
The article could be the basis to make an interesting tutorial.
Just as a side note, here is another discussion on the same article
Flickr Strobist Discussion |
|
|
|
08/17/2007 09:54:25 AM · #37 |
Originally posted by smurfguy:
If it were possible (and economically viable) to reduce noise, the manufacturers would figure out how.
|
Don't get you here... this happens with each generation.
Compare a MKIII to a 10D or D200 to a D100.
Originally posted by smurfguy:
Kirbic has spelled it out: you need more photons via longer collection time or larger pixels.
|
I'm having a hard time agreeing that increasing the ISO would not increase the amount of actual image information at a given "zone" compared to the original base0 EV pic.
I understand that to get more actual photons you need longer exposure or biger sensor or larger aperture but "getting more photons" isn't the point of this technique as I see it. What the technique is doing is to remap areas of the scene in the sensor/converter that had a low number of bits dedicated, to now have more bits to dedicated to the original dark FStop of the image.
In the base0 pic we are trying to increase the details of say Zone2 (from the zone system).
Suppose that this maps to the 5th (and final in this hypothetical camera) F-stop so only 128 levels are available.
Now we increase the ISO 4 stops and all zones now move up 4 stops.
So now what was originally Zone 5 is now zone 1 and has 2048 levels available.
The point of increasing the EV is to remap the data allocation not increase the number of photons hitting the sensor, no?
The fact that more photons do hit the sensor is what increases the EV value not that more photons are necessary to do this.
Originally posted by smurfguy:
I imagine using a prism to the image into two either degrades quality or is prohibitively expensive.
|
This thought exercise isn't to see if it is physically possible right now but whether at all.
Also who cares if "it is prohibitively expensive"?
The point is to determine if it is possible to do the 4 Stop Technique with a shutter actuation.
(thought I'm not sure why it would be prohibitively expensive considering all of the crazy optics going on w/ many elements inside a lens).
Originally posted by smurfguy:
The notion of two exposures in a short time is interrupted by the problem of collecting all that data off the sensor. As I understand it, this is already a limiting factor.
|
If that is a limiting factor then that is just a technological one that will go away with future generation converters and again should not be a concern for a thought exercise.
Originally posted by smurfguy:
You're better off figuring out how to get your subject to stay still for two exposures, or rigging up two cameras next to each other. =) |
That doesn't provide any value to the discussion.
Like I said, my hijack is to try and figure out if it's possible to do the 4 Stop Technique with one shutter actuation for cases where it is NOT possible to take two shots (which are more than not).
|
|
|
|
08/17/2007 11:29:39 AM · #38 |
Originally posted by rswank:
If that is a limiting factor then that is just a technological one that will go away with future generation converters and again should not be a concern for a thought exercise. |
On the contrary, when trying to solve a problem, the constraints are your main concern. You can't just assume "technology" will take care of it.
Originally posted by rswank:
"getting more photons" isn't the point of this technique as I see it. |
But it absolutely is. The whole idea is to get those shadows up to brighter, more detailed levels of exposure. For a given camera setup, you can do this in three ways: increasing aperture, lengthening shutter, or raising ISO.
ISO is just a multiplier. It multiplies noise right along with the image. Shutter and aperture, on the other hand, are used to let in more light thereby increasing the signal to noise ratio. ISO will not help you in the quest for reduced noise, and should be the photographer's last resort in a low-light environment.
Changing aperture, as explain in the post, will change the DOF and result in an unnatural image.
Shutter is the only option here, because you must have more photons in those dark areas to increase the signal to noise ratio.
So your question: Can it be done without the second, longer exposure. Answer: For a given camera setup, no.
Alright, well, what if I change my camera setup?
Well ok, maybe you're on to something, but keep in mind: in physics, every conversion will cost you. Friction and noise ensure that things cannot even be evenly exchanged.
For example, with your prism idea, the best-case output would be two images that are half as bright. So you'd need to expose twice as long anyway. But it's worse than that - the glass itself would cause more noise in your signal. And from a financial perspective, you've got an expensive prism and twice as many sensors. Ouch.
I'm not trying to derail your train of thought. Just trying to keep it on the tracks of reality. =) |
|
|
|
08/17/2007 11:31:00 AM · #39 |
Ok here's the way I see it. First of all, you need more light, you need more photons to hit that sensor in order to see anything in the shadow regions. Increasing the ISO essentially increases the gain of the sensor and therefore it's sensitivity. In doing this, you also increase the amount of electronic garbage (noise) that you see in the recorded image. This is very similar to radio gain or any other electronic gain.
Now here's where a major problem comes in. What if you want to shoot your base image at ISO200 or ISO400? Unless you have a camera body capable of producing very nice and useable images at ISO3200 and/or ISO6400, you're relatively screwed. The problem with using ISO to change it is that it introduces a great deal of unwanted electronic noise, especially in the shadow regions, as you go up the scale which is exactly what we want to avoid.
I will try this tonight and post some photos with 100% crops.
Message edited by author 2007-08-17 11:31:50.
|
|
|
|
08/17/2007 03:03:08 PM · #40 |
Originally posted by smurfguy:
Originally posted by rswank:
If that is a limiting factor then that is just a technological one that will go away with future generation converters and again should not be a concern for a thought exercise. |
On the contrary, when trying to solve a problem, the constraints are your main concern. You can't just assume "technology" will take care of it.
|
I'm not "assuming technology will take care of it" what I am saying is that you don't limit yourself to consider what might be scientifically possible because of current technological limitations.
Identifying what CAN be done leads to innovations that erase those technological limitations.
Can you imagine what that attitude would have quashed in Da Vinci's mind? But this is off the main point here, which is that increasing ISO would work for the 4 Stop Technique.
Originally posted by smurfguy:
[quote=rswank]"getting more photons" isn't the point of this technique as I see it. |
But it absolutely is. The whole idea is to get those shadows up to brighter, more detailed levels of exposure. For a given camera setup, you can do this in three ways: increasing aperture, lengthening shutter, or raising ISO.
ISO is just a multiplier. It multiplies noise right along with the image.
[/smurfguy]
This is where I haven't seen any specific rebuttal to why my thinking is wrong in that raising the ISO also increases the amount of data in a RAW file dedicated to the areas we are trying to get more detail out.
To summarize what I've more verbosely asked/stated above:
Increasing the ISO would remap areas of the scene in the sensor/converter that originally had a low number of bits dedicated to now have more bits to dedicated to it.
No? And if not why?
Would the now blown-out data that was once zones 5 6 and 7 still take up the same number of levels even though they are blown out?
What used to be shadow area is now midtones or highlights (Zone 5 or6) thus being stored out of a higher maximum data pool of up to 2048 levels instead of the 128 when they were indeed shadows (Zone 2).
Originally posted by SmufGuy:
For example, with your prism idea, the best-case output would be two images that are half as bright.
|
Is that optically true?
Is there any other way via mirrors to reflect an image onto two different planes?
I would think so but don't know optics.
Originally posted by SmufGuy:
But it's worse than that - the glass itself would cause more noise in your signal.
|
Why? The image goes through much more glass via modern 9+ element lenses.
Originally posted by SmufGuy:
And from a financial perspective, you've got an expensive prism and twice as many sensors. Ouch.
Just trying to keep it on the tracks of reality. =) |
I could care less about expense.
As I've said, I'm proposing this as a thought exercise to see if it is scientifically possible or not.
Can you imagine the amount of R&D money that went into the DSLRs we are using today?
Thanks for being devil's advocate btw, I appreciate being challenged and having to rethink my ideas/assumptions etc.
|
|
|
|
08/17/2007 04:02:21 PM · #41 |
This thread has turned waaaay too technical for me..
I plan on doing an easy test when I get home, someone mentioned it earlier.
I plan to take a properly exposed picture on a tripod in RAW , then take the same shot +2EV IN RAW and take it down to -2EV in software and "blend" the layers together. Take a 100% crop and see what I get.
Thanks for the tips.
Message edited by author 2007-08-17 16:02:46. |
|
|
|
08/17/2007 04:58:23 PM · #42 |
Originally posted by jfriesen:
This thread has turned waaaay too technical for me..
I plan on doing an easy test when I get home, someone mentioned it earlier.
I plan to take a properly exposed picture on a tripod in RAW , then take the same shot +2EV IN RAW and take it down to -2EV in software and "blend" the layers together. Take a 100% crop and see what I get.
Thanks for the tips. |
Why not take the second shot +4EV?
Doing any exposure adjustments in software wouldn't remap the shadow/dark areas we are hoping to get more detail out of to an F Stop value that has more levels of the RAW file assigned to them, right?
If this isn't correct I'll have to eat my hat as my argument hinges on this.
|
|
|
|
08/18/2007 08:03:07 PM · #43 |
Anyone know what method was used to combine the two images to get the one low noise image? I'm going to try this out tonight.
|
|
|
|
08/19/2007 03:54:33 PM · #44 |
Originally posted by SamDoe1:
Anyone know what method was used to combine the two images to get the one low noise image? I'm going to try this out tonight. |
I think that's his secret sauce... But one could imagine. Maybe convert both to HSV, and where V < (some threshold) substitute the H and S from the overexposed image. |
|
|
|
08/19/2007 04:03:45 PM · #45 |
Originally posted by SamDoe1:
Anyone know what method was used to combine the two images to get the one low noise image? I'm going to try this out tonight. |
Should not be any big deal. What you want to do is to use just the shadow areas from the +4 image. Here's how I'd proceed:
- Convert both images, using -4 exposure compensation on the +4 shot, and otherwise identical conversion parameters
- Layer the +4 exposure over the normal exposure
- Create a layer mask on the +4 exposure based on luminosity. You could do this in a number of ways. How I'd go about it would be to make a copy of the +4 layer and do a threshold operation on it with the desired level (trial & error) and then select the dark color. Now blur the edge as desired, and you've got a selection that you can use as a mask. |
|
|
|
08/19/2007 10:53:39 PM · #46 |
The article posted by the OP has a link to a beta download of his software.
Not to be a pest, but am I wrong in that overexposing even with just increasing the ISO would still remap the areas of interest in the image to now receive a bigger allocation of data in the RAW?
Isn't that the point of overexposing with increased shutter speed?
Beta software //forums.dpreview.com/forums/read.asp?forum=1018&message=24340277 |
|
|
|
08/19/2007 11:06:19 PM · #47 |
Using ISO will boost those dark areas into the more dynamic regions of exposure, but it will also boost the noise, probably overcoming any benefit from the extra dynamics. By increasing the shutter, you'll boost the light without the noise.
Note that, as the goalis minimal noise, he also says to use ISO 100 for both shots.
Maybe I'll try some test shots if I have some time... |
|
|
|
08/19/2007 11:13:02 PM · #48 |
Man all this talk, I wish I had a REAL camera.
 
Message edited by author 2007-08-19 23:14:30.
|
|
|
|
08/19/2007 11:15:55 PM · #49 |
As I said before, the ISO will do what changing the shutter speed will do. *But* if you're already shooting at ISO400, how high would you have to go to get that additional 4 stops of dynamic range?
|
|
|
|
08/20/2007 12:39:58 AM · #50 |
Alright, this shot should show what I mean. ISO may improve dynamic range, but shutter time improves dynamic range without extra noise.
Read the image notes for more details. Also, I accidentally used +2 stops of shutter time, and +4 stops of ISO, but it doesn't really matter. It still illustrates the point: shutter doesn't boost noise while ISO does. The relative amounts of noise are inconsequential.
Don't like it? Do you own test. =) |
|
Home -
Challenges -
Community -
League -
Photos -
Cameras -
Lenses -
Learn -
Help -
Terms of Use -
Privacy -
Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 07/19/2025 06:10:08 PM EDT.
