Single Shot vs Multi Shot Backs

[jvora]

Hello All :

Chances are that the topic of single vs multi shot backs have been discussed - Did a brief search and did not find the answer I was looking for, so here is my most basic question :

Not sure how physically moving the sensor in micro mm for each of the exposures helps in rendering better detail and color than keeping the sensor stationary and exposing 3 times for each of the RGB channels ?

Not sure I understand the concept - Kindly provide an explanation !

Thanks,

Jai

[michaelnotar]

weither the sensor moves or uses a filter to capture each color channel, i dont think it doesnt matter. different companies, different patents. just two ways to skin a cat.

[shutay]

Hello All :

Chances are that the topic of single vs multi shot backs have been discussed - Did a brief search and did not find the answer I was looking for, so here is my most basic question :

Not sure how physically moving the sensor in micro mm for each of the exposures helps in rendering better detail and color than keeping the sensor stationary and exposing 3 times for each of the RGB channels ?

Not sure I understand the concept - Kindly provide an explanation !

Thanks,

Jai

Incorrect. It does not expose each of RGB channels 3 times. It exposes each of R, G and B in each pixel location once so that you have a complete set of R, G and B at each pixel location. This means slightly better spatial resolution, but much better colour resolution.

Normally, the Bayer filter means that out of say, 10 million pixels captured, 5 million will be green channel only, 2.5 million will be red only and the remaining 2.5 million will be blue channel only. Then the software - whether in-camera, or the RAW image processor you run on your computer - will interpolate the data so that it fills in the blanks, so that each image pixel location has red, green and blue data, but it is manufactured data based on mathematical models of what they expect to be in each location where there is missing colour data.

Multi-shot removes the mathematical modeling and puts REAL DATA in each channel. That is why multi-shot images have so much more resolution (i.e., better per pixel resolution) than single shot backs, but it is offset by the limitations in what you can use multi-shot backs for.

[ixpressraf]

The Imacon 528c and 384c both move the sensor at pixel dimension. There are no moving parts in front of or attacheted to the camera. Therefore it works each time perfectly. I have attacheted an image i shot with my 384 back but a 528c will give you an even better file as it is a 22Mpmultishot and microstep.
There simply is nothing else that gives you the same flexibillity such as an ixpress does. The nice thing is that you can also use it outside the studio as a one shot back, it gives perfect files and they go almost for free second hand. ( I am selling my 528c becaurse i almost always take the 384c for shooting. i donnt like the rectangular size, i am more into the old hasselblad image)

[Dustbak]

As far as I know multishot takes 4 images, for each of the RGB color channels and twice for the green channel. The 2 takes for the green channel are also used to check whether there was movement or change during the exposures.

This was how it went with my 384. I now have a 39MS which has a slight different sequence. It takes a pre-shot, another bleep followed by 4 shots. Maybe David or Paul can explain what the purpose is for the pre-shot and the small bleep? (was meaning to ask for this anyway).

I changed from the square to the rectangular because my shots are mostly that way  (next to that I am also rid of the CA with the new back).

The difference of single vs multi is extremely visible on garment or fabric.

[jvora]

Shutay :

I think you did not read/understand my post correctly -

The opening post states " . . . and exposing 3 times for each of the RGB channels ? " which clearly implies 3 exposures for EACH of the RGB channels and NOT 3 exposures for a composite RGB capture which would be absolutely meaningless !

This part is not the issue I had difficulty in understanding.

The post by "michaelnotar" clarifies what I was trying to understand by stating that both the aproaches amount to the same result/set of data - just that for a stationary sensor, a different filter ( i.e. R, G, B ) is employed for each of the 3 exposures VS a sensor using a bayer filter which has to move a pixel unit so that each pixel cell/well is exposed to each of the R, G & B "coatings" on the bayer filter.

OK - I now have got it and it makes sense !!

Thought I would clarify my original post - And thanks for your inputs !

Jai

Incorrect. It does not expose each of RGB channels 3 times. It exposes each of R, G and B in each pixel location once so that you have a complete set of R, G and B at each pixel location. This means slightly better spatial resolution, but much better colour resolution.

Normally, the Bayer filter means that out of say, 10 million pixels captured, 5 million will be green channel only, 2.5 million will be red only and the remaining 2.5 million will be blue channel only. Then the software - whether in-camera, or the RAW image processor you run on your computer - will interpolate the data so that it fills in the blanks, so that each image pixel location has red, green and blue data, but it is manufactured data based on mathematical models of what they expect to be in each location where there is missing colour data.

Multi-shot removes the mathematical modeling and puts REAL DATA in each channel. That is why multi-shot images have so much more resolution (i.e., better per pixel resolution) than single shot backs, but it is offset by the limitations in what you can use multi-shot backs for.

[jvora]

Dustbak :

Very cleaver way to check for change / movement between exposures and that in reality its not 3, but 4 exposures !

Many thanks for clarifying this - cool !

Jai

As far as I know multishot takes 4 images, for each of the RGB color and twice for the green channel. The 2 takes for the green channel are also used to check whether there was movement or change during the exposures.

This was how it went with my 384. I now have a 39MS which has a slight different sequence. It takes a pre-shot, another bleep followed by 4 shots. Maybe David or Paul can explain what the purpose is for the pre-shot and the small bleep? (was meaning to ask for this anyway).

I changed from the square to the rectangular because my shots are mostly that way  (next to that I am also rid of the CA with the new back).

The difference of single vs multi is extremely visible on garment or fabric.

[Photostudent]

As far as I know multishot takes 4 images, for each of the RGB color and twice for the green channel. The 2 takes for the green channel are also used to check whether there was movement or change during the exposures.

This was how it went with my 384. I now have a 39MS which has a slight different sequence. It takes a pre-shot, another bleep followed by 4 shots. Maybe David or Paul can explain what the purpose is for the pre-shot and the small bleep? (was meaning to ask for this anyway).

I changed from the square to the rectangular because my shots are mostly that way  (next to that I am also rid of the CA with the new back).

The difference of single vs multi is extremely visible on garment or fabric.

continued....

Four exposure are superimposed and combined into one image because  there are twice as many green pixel on the sensor as red and blue. the greeb channel is capture twice for every position in the image.two green channels are avraged before the combined with red and green. the lighting need consistency in all four exposures.

Micro scan(Sixteen shot):- whole surface of sensor is not sensitive to light. data fall between the light sensitive area of two individual pixel are lost. in such cases micro scanning helps by shifting the sensor in step of half a pixel and thus also capturing the data fall between two individual pixel.a total of 16 exposures are required to capture full green, red and blue information for every position in image area.

[shutay]

Shutay :

I think you did not read/understand my post correctly -

The opening post states " . . . and exposing 3 times for each of the RGB channels ? " which clearly implies 3 exposures for EACH of the RGB channels and NOT 3 exposures for a composite RGB capture which would be absolutely meaningless !

This part is not the issue I had difficulty in understanding.

The post by "michaelnotar" clarifies what I was trying to understand by stating that both the aproaches amount to the same result/set of data - just that for a stationary sensor, a different filter ( i.e. R, G, B ) is employed for each of the 3 exposures VS a sensor using a bayer filter which has to move a pixel unit so that each pixel cell/well is exposed to each of the R, G & B "coatings" on the bayer filter.

OK - I now have got it and it makes sense !!

Thought I would clarify my original post - And thanks for your inputs !

Jai

Oh yes, that's true. I didn't see that possible interpretation.  My apologies!

[jvora]

Shutay :

Cool man - No apologies needed !


Jai

Oh yes, that's true. I didn't see that possible interpretation.  My apologies!

[PdF]

Personnaly, 90% of my work is made in multishoot, 7 or 8 % in multistep (16shots), and the rest in 1 shot.

Because it's really better (more precision, best colors, exemption of moiré, etc...).

But how many multishoots back are now available ? One Sinar and one Hasselblad. A shame.

PdF

[Henry Goh]

Personnaly, 90% of my work is made in multishoot, 7 or 8 % in multistep (16shots), and the rest in 1 shot.

Because it's really better (more precision, best colors, exemption of moiré, etc...).

But how many multishoots back are now available ? One Sinar and one Hasselblad. A shame.

PdF

How is a multi shot done when shooting with studio strobes?

[bdp]

The strobes fire 4 or 16 times. It is 4 or 16 separate captures. Nothing can move for the time it takes to capture the sequence - camera or objects on the set. A 16 shot capture can take around a minute.

See this thread regarding the multishot concept:  http://luminous-landscape.com/forum/index....c=28148&hl=

Ben

[john milich]

so in theory, with any digital camera, one could make three exposures using separate rgb pass filters in front of the lens, superimposing them in PS?

[thsinar]

that use to be like this, with the first digital backs being monochrome and using RGB filters, then the colour image being created in the capture software. Today all sensors of MFD backs have the colour filters directly on top of the pixels, according to a certain pattern.

Thierry

so in theory, with any digital camera, one could make three exposures using separate rgb pass filters in front of the lens, superimposing them in PS?

[shutay]

so in theory, with any digital camera, one could make three exposures using separate rgb pass filters in front of the lens, superimposing them in PS?

There are software in fact that you can use with any digital camera, where you shoot anywhere from 16 shots or more. Then you specify whether the camera was on a tripod or not, etc., and the software layers them on top of each other and aligns them. It uses the idea that if you were pressing the shutter with your finger each time, there would have been small and randomly different displacements from one shot to another. It then evens out the chroma noise between the shots and apparently, you can both reduce noise and increase detail.

The only catch is, I can't remember which software does this, although I know I have encountered them before. I seem to think it was Kolor Autopano Pro, but I can't find any reference to it on their website.

shutay

[ixpressraf]

There are software in fact that you can use with any digital camera, where you shoot anywhere from 16 shots or more. Then you specify whether the camera was on a tripod or not, etc., and the software layers them on top of each other and aligns them. It uses the idea that if you were pressing the shutter with your finger each time, there would have been small and randomly different displacements from one shot to another. It then evens out the chroma noise between the shots and apparently, you can both reduce noise and increase detail.

The only catch is, I can't remember which software does this, although I know I have encountered them before. I seem to think it was Kolor Autopano Pro, but I can't find any reference to it on their website.

shutay

That has absolutely nothing to do with multishot as done by a real MS back. This only multiplies the entire image as it was shot in one shot. MS shoots RGB color per color and that is what makes the entire difference.

[Jack Flesher]

so in theory, with any digital camera, one could make three exposures using separate rgb pass filters in front of the lens, superimposing them in PS?

Hi John:

There is some astral imaging software that does that very thing, though I have not ever experimented with it.

One software I played with a few years ago did a pretty credible job just stacking regular frames -- it was called Image Stacker: http://www.tawbaware.com/imgstack.htm. In the high-resolution mode it did add a bit of perceived resolution, and in fact the software suggested you do the captures with the camera a little loose in the tripod so there was movement between the captures.  It then used the movements to interpolate intermediate pixel data and create a larger file.  At the end of the day it worked, but the net gain in perceived resolution was pretty minimal over my normal uprezzing technique, so I did not explore it further.  I have not played with it in at least two years, so things could be different now.  The other kind of nifty ability was long exposures by adding up multiple frames.  It also used the acquired data to brighten the image while reducing noise. That was pretty clever and actually worked quite well. The one thing it didn't do was increase color depth -- at least that I could see.

I have seen results from MS backs, and IMO those gains are ones I would consider "real" in terms of detail and color fidelity.  But just like when using a scanning back, any of these options require the subject remains motionless (unless you want motion blur), which is tough to accomplish outside the studio.

Cheers,

[David Klepacki]

As already mentioned above, combining multiple images with software is not equivalent to a multi-shot capture digital back.  The multi-shot digital back guarantees physically captured information for the Red, Blue, and Green at every spatial location.  There is no interpolation of "missing" pixel information with a muti-shot capture back; all three color channels (R, G, B ) are produced from physically measured data.  On the other hand, software that combines multiple images is working with interpolated pixel information, since each of the captured images is first interpolated to obtain the missing pixel information.  Therein lies the difference; with the multi-shot back, 100% of the pixel information is physically measured.  With software combining of images, only one-third of each image is physically captured, and guessed pixel information (via interpolation) enters into the process.

[Hoang]

There are software in fact that you can use with any digital camera, where you shoot anywhere from 16 shots or more. Then you specify whether the camera was on a tripod or not, etc., and the software layers them on top of each other and aligns them. It uses the idea that if you were pressing the shutter with your finger each time, there would have been small and randomly different displacements from one shot to another. It then evens out the chroma noise between the shots and apparently, you can both reduce noise and increase detail.

The only catch is, I can't remember which software does this, although I know I have encountered them before. I seem to think it was Kolor Autopano Pro, but I can't find any reference to it on their website.

shutay

You can do this in Photoshop CS4 Extended (and in CS3 Extended too, I believe). When I shoot static subjects I take 10 shots using an intervalometer, load them as layers into Photoshop, convert the 10 (you can do more if you want) shots into one smart object, then change the start object blend mode to mean. It won't give you any more resolution but it does take away quite a bit of noise. You can get really clean shadows this way.