[sarka]

I have no problem being corrected on this, but we can't really improve the dynamic range of the image this way, because it's still either a 12-bit raw or 8-bit jpeg, right?
The dynamic range stays the same unlike HDR which involves generating a higher bit-depth result from multiple images of lower bit-depth.
But maybe we can make better use of the dynamic range by canceling out noise with this method? Like most of the variety of options on our cameras, this has an application in some circumstances and should be used when it will do some good.

[WerTicus]

Okay here is a link to a full 9mb jpg (rared) right from the camera no processing at all:

MEGAUPLOAD - The leading online storage and file delivery service

It was taken with:
a pretty good tripod,
2 second delay timer,
9 shots, ev comp on
D range on,
iso 200,
f8, (according to photozone this is the optimal resolution for this lens)
2 second exposure,
FA 50 f1.4 lens.

You can see the dogs lungs blurring that part of his fur, otherwise no motion in the image. You can read 'cat 5e' on that cable but there is not even any text visible to the naked eye from the cameras position. Also i didn't realise how much fur and sand there is... looks really bad but most of it isn't visible (the room is pretty dark, curtains mostly drawn etc)

[Jewelltrail]

Werticus: Okay here is a link to a full 9mb jpg (rared) right from the camera no processing at all:

It took me about 10 tries before I could pass the test of entering the 4 correct letters/numbers. Then I downloaded the file, with much anticipation, but could not open it--it is not a jpg--it is some kind of compressed file.

[Jewelltrail]

I have no problem being corrected on this, but we can't really improve the dynamic range of the image this way, because it's still either a 12-bit raw or 8-bit jpeg, right?
The dynamic range stays the same unlike HDR which involves generating a higher bit-depth result from multiple images of lower bit-depth.
But maybe we can make better use of the dynamic range by canceling out noise with this method? Like most of the variety of options on our cameras, this has an application in some circumstances and should be used when it will do some good.

I agree--you can never really change the DR of any given camera. My wording was poor--but you can, perhaps, make better use of a camera's DR though this means. Isn't that what the K20's so-called DR button does?

[WerTicus]

its a .rar file containing a jpg

winrar opens it and its a very good compression tool for all sorts of compression types.... it can make zip files half the size of windows for example (free program)
http://www.rarsoft.com/download.htm

[newarts]

I agree--you can never really change the DR of any given camera. ...

I think that's not mathematically true but may be practically true.

The lower limit of camera DR is usually due to noise sources other than from the image itself. If this noise is consistent in its properties (its characteristics don't change from exposure to exposure), a correction may be possible.

But most likely not practical.

Iowa Dave

[procyon]

I think that's not mathematically true but may be practically true.

The lower limit of camera DR is usually due to noise sources other than from the image itself. If this noise is consistent in its properties (its characteristics don't change from exposure to exposure), a correction may be possible.

But most likely not practical.

Iowa Dave

This correction is called dark frame subtraction.
Basically the workflow is:
1) Shoot n images of the scene (the bigger n, the better)
2)Shoot n images with the lens cover on (and possibly even covered with a dark bag or similar to avoid any light reaching the sensor). These images have to be shot using exactly the same settings.
3) Average the images of the scene, thus getting rid of any random noise (provided the number n is large enough).
4) Average the images shot with lens cover on. This results in a dark frame that consists of only the noise that is exactly the same in every image (noise from sensor readout electronics, faulty pixels etc.).
5) Subtract the dark frame from the average image of the scene, thus getting rid of any systematic noise also.

And the result is a very clean image.

This system is very common in webcam astronomy, where cheap webcams are attached to telescopes to record a video clip which is later processed as I decribed. Registax is one software that can handle these clips sometimes consisting of thousands of frames (webcams have lots of noise ...) that need to be aligned correctly, bad frames (blurry because of atmosphere turbulence etc.) need to be removed and so on.

[Jewelltrail]

Werticus: its a .rar file containing a jpg

winrar opens it and its a very good compression tool for all sorts of compression types.... it can make zip files half the size of windows for example (free program)
WinRAR archiver, a powerful tool to process RAR and ZIP files

YOU DA BEST! Thank you!

[RioRico]

I did many shots at ISO 100 & the final pic was so clean of noise that I was amazed--it looked almost not real, and seemed to lack sharpness.

And here is the real application - not reducing noise at high ISOs, but getting low ISO shots to be nearly grainless, or at least noiseless. In my film+chemistry days, I loved shooting REAL SLOW film, like ASA 10 Kodalith pushed to ASA 25 with Acufine. Portraits seemed to bulge. I'll have to try ME in the light tomorrow. Thanks!

[WerTicus]

there ya go the answer was staring at us in the face, we just figured out how to get iso25 on the pentax

does anyone know what drive mode is fastest for shooting 9 shots in a row?

[Jewelltrail]

Werticus: there ya go the answer was staring at us in the face, we just figured out how to get iso25 on the pentax

RioRico: And here is the real application - not reducing noise at high ISOs, but getting low ISO shots to be nearly grainless, or at least noiseless. In my film+chemistry days, I loved shooting REAL SLOW film, like ASA 10 Kodalith pushed to ASA 25 with Acufine. Portraits seemed to bulge. I'll have to try ME in the light tomorrow. Thanks!

Yes, ISO 25 has its applications. We need to rename the thread now. I can see using my remote and tripod to tap nice results with this, especially with landscape and night work. If you wanted something special, to print very large, this might be the way to go.

Werticus: does anyone know what drive mode is fastest for shooting 9 shots in a row?

If you want full-size pics, the "continuous" drive mode would be the one since "burst" mode shoots 21fps, but at small file sizes.

[Donald]

I shoot almost exclusively multiple exposure style, starting with film, now dslr, typically blending between 20 ~ 50 images. I haven't shot these noisey ultra long exposures, but I'm a small expert in multiple exposures.

There is lots of information regarding multiple exposure methods for long exposure noise reduction. Software applications, open source code, wikipedia page....


What ricehigh said earlier is substantially correct. If we are basing critique on this style with K10/20/-7 camera automatic function, no comment as I don't have these models. He is correct however in terms of actual photographic capture.

Multiple exposure capture is done either adjusting ISO, substantially limiting number of blendable captures, or via exposure value adjustment based on Log2.

In camera method probably uses variation of "cemnt" algorithm for auto-blending (acronym might not be competely accurate, sounds close). I am not in favour of this method as all multiple exposure capable dslr's are limited to 9 exposures, and it's an auto function with no user control.

Having shot somewhere between 50 ~ 100,000 multiple exposure shots I will tell you DSLR EV adjustment is a waste of time for the noise issue ricehigh states, but mostly due to each blendable digital capture starts with reduced dynamic range: stacking multiples of low DR equals low DR result.

Multiple exposure on film is totally different as the process is inherently additive, building light capture according to Log2 EV compensation, until proper exposure is reached.

Digital blending of multiple exposures is best done in post processing using properly exposed captures, consequently noise should not be an issue.

Opacity blending, average blending, or even post processed Log2 are fine; the last two work best for approximately up to 10 images as each subsequent image contributes less; blending more than 10 images only opacity screening is capable of stunning colours and massive dynamic range.

But there is caveat. Of all the digital multiple exposures I've done, red's have never performed well. I believe this is weakness of Bayer array GBGR and consequent interpolation; however, I'm no expert in sensor's but have tried enough of them, never with good red results. I'm looking for cheap Sigma Foveon camera to try out: omigod they sound terrible, but with awesome sensor.

[boom]

高的米： 您有许多好想法，但是我建议您是有點關於Pentax的更多positve。 因為您是很快投入Pentax名字下來，許多論壇成員不严肃对待您。 嘗試更加支援的方法，并且您將使更多人民听您，不用他們变得很惱怒。 歐內斯特

" 当水制服fire."，人类制服无情; Mencius 6A ：18.

I had a good laugh at the trashlated Chinese.

[Jewelltrail]

boom: I had a good laugh at the trashlated Chinese.

Yes, so did I!

[MNCurt-K10D]

I too have recently become interested in ME to reduce noise in poorly lit situations, after talking to a friend who bought a Sony DSC-HX1 in part for its "Twilight" Mode, which does an automatic frame stacking to reduce noise in low light level photography... I haven't been able to find any details on how the Sony performs this function, but it seems I should be able to do something similar with the K20D ME function. Also, looking at the Remote Assistance software documentation it would appear that tethered, the K20D can be used to collect and add or average a sequence of exposures - perhaps meant for astrophotography?

Anyway, since I am just exploring the capabilities of my K20D, all I can say is "Inquiring Minds Want To Know" on this whole topic... so can anyone speak to the Sony algorithm/implementation for image averaging? And what can be accomplished with the K20D and Remote Assistant? Inputs/suggestions will be appreciated!