Sign in with
Sign up | Sign in
Your question

DSLR and infrared

Last response: in Digital Camera
Share
Anonymous
May 20, 2005 3:03:01 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

Is it possible to remove all visible colour wavelengths
in software post-processing, leaving you with only the infrared image,
rather than use an infrared filter? I figure you could
use a narrow enough aperture on the lens to accommodate
the infrared and visible light focus points.
-Rich

More about : dslr infrared

Anonymous
May 20, 2005 5:33:05 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

RichA wrote:
> Is it possible to remove all visible colour wavelengths
> in software post-processing, leaving you with only the infrared image,
> rather than use an infrared filter? I figure you could
> use a narrow enough aperture on the lens to accommodate
> the infrared and visible light focus points.


The photosites are filtered to record RGB (CYM in some cameras). Once
filtered and counted, it is just data, the camera, the file, the PS s/w
have no idea what contribution specific wavelengths made to the 'count'
in any photosite. (When playing it back, it's recreated in the computer
monitor or printer as a color).

Whatever IR is recorded is where the sensitivity of those channels
(mostly the red channel I suppose) spills into the IR part of the
spectrum. Then, by adding an IR friendly filter to the camera, you
block as much of the visible wavelengths as possible leaving IR only.
That IR can be 'counted' in any channel (mostly the red, I believe).

At that point the gain of a channel in IR is very low (beacuse its not
centered on the filter for that channel), so a longer exposure is
required to get anything. To make it worse, some cameras have an IR
filter over the sensor, so an even longer expsosure is required to
record the IR from a scene.

http://www.electro-optical.com/bb_rad/bb_rad.htm

Since the camera or photoshop does not *know* that that recording was
IR, you have to seperate it out. Again, I assume its mostly in the red
channel as that is closest to IR.

In effect, PS believes whatever 'count' is in a R,G or B channel
represents red, green or blue, not IR.

Cheers,
Alan.


--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.
Anonymous
May 20, 2005 9:47:06 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

RichA <none@none.com> writes:
> Is it possible to remove all visible colour wavelengths in software
> post-processing, leaving you with only the infrared image, rather
> than use an infrared filter?

No.

Think about it: There are no "infrared" photosites in your camera,
only monochrome photosites masked by red, green and blue filters
(that to some extent also are transparent to near IR light),
When looking at the raw sensor data, you don't know which bits that
were generated by light in the near infrared band, on which bits
that were generated by light in the visible band.

It is possible to fake the infrared film "look" of Kodak HIE pretty
good in Photoshop (boost green channel, desaturate, blur, add grain),
but it has nothing to do with a real near-IR image.
--
- gisle hannemyr [ gisle{at}hannemyr.no - http://folk.uio.no/gisle/ ]
------------------------------------------------------------------------
Kodak DCS460, Canon Powershot G5, Olympus 2020Z
------------------------------------------------------------------------
Related resources
Anonymous
May 21, 2005 4:38:59 AM

Archived from groups: rec.photo.digital.slr-systems (More info?)

Alan Browne <alan.browne@freelunchVideotron.ca> writes:
> Whatever IR is recorded is where the sensitivity of those channels
> (mostly the red channel I suppose) spills into the IR part of the
> spectrum. Then, by adding an IR friendly filter to the camera, you
> block as much of the visible wavelengths as possible leaving IR
> only. That IR can be 'counted' in any channel (mostly the red, I
> believe).

Not necessarely. On my Oly 2020Z, all three channels seem to
be equally sensitive to IR. On my DCS460, it is the blue channel
that is most sensitive, and the red is the least sensitive.

How coloured dyes (such as those used in beyer filters) respond
to near-IR wavelengths is quite unpredictable.

> http://www.electro-optical.com/bb_rad/bb_rad.htm

That page is totally irrelevant if we talking about the type of image
you shoot with standard cameras and Hoya R72 filters.

This guy talks about blackbody radiation, which is the infrared
energy radiated by an object due to its temperature. Capturing
blackbody radiation requires a very special camera, and is
known as "thermal photography".

What hobbyist call "infrared photography" should really be called
"Near-infrared photography". It is about taking photographs of
/reflected/ infrared light in a narrow spectrum just above visble
light - from 700 nm up to about 950 nm.
--
- gisle hannemyr [ gisle{at}hannemyr.no - http://folk.uio.no/gisle/ ]
------------------------------------------------------------------------
Kodak DCS460, Canon Powershot G5, Olympus 2020Z
------------------------------------------------------------------------
Anonymous
May 21, 2005 2:26:28 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

Gisle Hannemyr wrote:

> Alan Browne <alan.browne@freelunchVideotron.ca> writes:
>
>>Whatever IR is recorded is where the sensitivity of those channels
>>(mostly the red channel I suppose) spills into the IR part of the
>>spectrum. Then, by adding an IR friendly filter to the camera, you
>>block as much of the visible wavelengths as possible leaving IR
>>only. That IR can be 'counted' in any channel (mostly the red, I
>>believe).
>
>
> Not necessarely. On my Oly 2020Z, all three channels seem to
> be equally sensitive to IR. On my DCS460, it is the blue channel
> that is most sensitive, and the red is the least sensitive.

That indicates that the 'out of visible filtering of the RGB color
filters is not controlled. eg: the blue filter blocks green and red but
out there beyond visible it is no longer filtering.

>
> How coloured dyes (such as those used in beyer filters) respond
> to near-IR wavelengths is quite unpredictable.

It appears to be uncontrolled, rather than unpredictable.

>
>
>>http://www.electro-optical.com/bb_rad/bb_rad.htm
>
>
> That page is totally irrelevant if we talking about the type of image
> you shoot with standard cameras and Hoya R72 filters.

I only posted the link to show the relationship of the IR to the red
channel.

<SNP>
> What hobbyist call "infrared photography" should really be called
> "Near-infrared photography". It is about taking photographs of
> /reflected/ infrared light in a narrow spectrum just above visble
> light - from 700 nm up to about 950 nm.

Probably. There is so little authoratitve information on the subject
with resepct to what wavelenghts are passing the RGB filters, the sensor
filters and being counted by the photosites. I haven't seen a gain
chart for the whole chain (in segments). Do you know where there is one?


--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.
Anonymous
May 21, 2005 6:39:48 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

> Whatever IR is recorded is where the sensitivity of those channels (mostly
> the red channel I suppose) spills into the IR part of the spectrum. Then,
> by adding an IR friendly filter to the camera, you block as much of the
> visible wavelengths as possible leaving IR only. That IR can be 'counted'
> in any channel (mostly the red, I believe).

I've noticed that my S3 records the near-IR in the blue channel - I thought
I must be mistaken until I read Gisle's post on this subject!

While tweaking the RAW image in the HyperUtility software if I pull the
middle of the red curve down and push the blue curve up, it increases the IR
effect: the areas of reflected IR are pale blue and those of least are deep
red.

Strange.

I'm happy with the result though which is what matters really isn't it!

Craig.
Anonymous
May 21, 2005 6:39:49 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

Craig Marston wrote:

> middle of the red curve down and push the blue curve up, it increases the IR
> effect: the areas of reflected IR are pale blue and those of least are deep
> red.
>
> Strange.
>
> I'm happy with the result though which is what matters really isn't it!

Certainly. The 'typical' presentation of IR photos is in B&W, but they
can be presented in any color you like since there is no 'color' at all
in the human sense.

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.
Anonymous
May 21, 2005 8:45:28 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

In article <d6ngej$2sr$1@inews.gazeta.pl>,
Alan Browne <alan.browne@freelunchVideotron.ca> wrote:
>Gisle Hannemyr wrote:

[ ... ]

><SNP>
>> What hobbyist call "infrared photography" should really be called
>> "Near-infrared photography". It is about taking photographs of
>> /reflected/ infrared light in a narrow spectrum just above visble
>> light - from 700 nm up to about 950 nm.
>
>Probably. There is so little authoratitve information on the subject
>with resepct to what wavelenghts are passing the RGB filters, the sensor
>filters and being counted by the photosites. I haven't seen a gain
>chart for the whole chain (in segments). Do you know where there is one?

It isn't just the filters and the sensor which control things
when you get to the far IR regions. There are only two bands which pass
the atmosphere cleanly. (They were 3-5 and 9-14 nM IIRC.) All other
"colors" are blocked to one degree or another.

Sensors for the longer wavelengths are also strange compared to
the visible light ones. I have seen (and used) CCD scanned sensors, but
the actual sensor material is something esoteric like a
lead-tin-telluride alloy, not a silicon photovoltaic sensor. Now,
consider the problems of cleaning such a sensor array given that the
alloy is a lot softer than lead-tin solder, and any deformation destroys
the crystal structure. But then -- these did not have interchangeable
lenses, and were typically in sealed enclosures which were dry-nitrogen
purged on assembly (in a clean room).

And the sensors are commonly used at liquid nitrogen
temperatures to minimize resistive electrical noise.

Probably the most widely seen images through such systems are
the ones put on television during the gulf war and the more recent one
in Iraq, as the targeting cameras for the aircraft fired missiles were
far IR systems.

They are strange looking beasts, with lenses (totally opaque in
the visible spectrum) made of germanium or silicon.

And the only references to which I had access were not public
documents at the time, so I can't post a link to them. (They were not
even on the web back then, though they may well be by now.)

Enjoy,
DoN.
--
Email: <dnichols@d-and-d.com> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
Anonymous
May 21, 2005 8:55:45 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

In article <d6njl3$djr$1@inews.gazeta.pl>,
Alan Browne <alan.browne@freelunchVideotron.ca> wrote:
>Craig Marston wrote:
>
>> middle of the red curve down and push the blue curve up, it increases the IR
>> effect: the areas of reflected IR are pale blue and those of least are deep
>> red.
>>
>> Strange.
>>
>> I'm happy with the result though which is what matters really isn't it!
>
>Certainly. The 'typical' presentation of IR photos is in B&W, but they
>can be presented in any color you like since there is no 'color' at all
>in the human sense.

Or -- you can make a multi-band sensor array, and display the
near-IR in one color, the 3-5 uM in another, and the 7-14 uM in the
third for an interesting false color effect.

Of course, one problem is that the longer the wavelength, the
greater the diffraction which causes loss of resolution. So I would
probably put the near IR in the color to which the human eye is most
sensitive, so it would carry the most detail.

Enjoy,
DoN.

--
Email: <dnichols@d-and-d.com> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
Anonymous
May 21, 2005 11:02:03 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

"Alan Browne" <alan.browne@freelunchVideotron.ca> wrote in message
news:D 6ngej$2sr$1@inews.gazeta.pl...
> Gisle Hannemyr wrote:
SNIP
>> How coloured dyes (such as those used in beyer filters) respond to
>> near-IR wavelengths is quite unpredictable.
>
> It appears to be uncontrolled, rather than unpredictable.

In fact, it is predictable in the sense that these dyes are
transparent to IR. What complicates the predictability is the IR
filter that can have different cut-off points. The IR filter could be
omitted, or it is part of an assembly with the Anti-Aliasing filter
(e.g. Canon's assembly looks like
http://www.canon.com/technology/detail/digi_35mm/lo_fil...,
and incorporates both a dichroic mirror *and* an IR-filter).

In some camera's (e.g. the Canon 20Da) the characteristics for
extended-red/near-infrared filtration are changed, but those changes
will cause difficulties in regular pictorial photography.

And then there is the variable amount of reflected Near-IR light ...,
predictability is the issue.

Bart
Anonymous
May 21, 2005 11:02:04 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

Bart van der Wolf wrote:

>
> "Alan Browne" <alan.browne@freelunchVideotron.ca> wrote in message
> news:D 6ngej$2sr$1@inews.gazeta.pl...
>
>> Gisle Hannemyr wrote:
>
> SNIP
>
>>> How coloured dyes (such as those used in beyer filters) respond to
>>> near-IR wavelengths is quite unpredictable.
>>
>>
>> It appears to be uncontrolled, rather than unpredictable.
>
>
> In fact, it is predictable in the sense that these dyes are transparent
> to IR. What complicates the predictability is the IR filter that can
> have different cut-off points. The IR filter could be omitted, or it is
> part of an assembly with the Anti-Aliasing filter (e.g. Canon's assembly
> looks like
> http://www.canon.com/technology/detail/digi_35mm/lo_fil...,
> and incorporates both a dichroic mirror *and* an IR-filter).
>
> In some camera's (e.g. the Canon 20Da) the characteristics for
> extended-red/near-infrared filtration are changed, but those changes
> will cause difficulties in regular pictorial photography.
>
> And then there is the variable amount of reflected Near-IR light ...,
> predictability is the issue.

Unless you have the gain over wavelength curves for every bit of glass
in the chain and include the sensitivity of the sensor at those
wavelengths, it is unpredictable. (Which is not the same as saying it's
not repetable). Even where the manufs attempt some degree of IR
supression, enough gets through that a longer exposure will record it
(if the visible wavelengths are suppressed by filtering and don't swamp
out the IR).

In the case of the 20Da, they appear to want the IR in order to help the
overall image expsure; it is intended for astro work, after all.

Cheers,
Alan

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.
Anonymous
May 22, 2005 6:46:02 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

DoN. Nichols wrote:

> In article <d6ngej$2sr$1@inews.gazeta.pl>,
> Alan Browne <alan.browne@freelunchVideotron.ca> wrote:
>
>>Gisle Hannemyr wrote:
>
>
> [ ... ]
>
>
>><SNP>
>>
>>>What hobbyist call "infrared photography" should really be called
>>>"Near-infrared photography". It is about taking photographs of
>>>/reflected/ infrared light in a narrow spectrum just above visble
>>>light - from 700 nm up to about 950 nm.
>>
>>Probably. There is so little authoratitve information on the subject
>>with resepct to what wavelenghts are passing the RGB filters, the sensor
>>filters and being counted by the photosites. I haven't seen a gain
>>chart for the whole chain (in segments). Do you know where there is one?
>
>
> It isn't just the filters and the sensor which control things
> when you get to the far IR regions. There are only two bands which pass
> the atmosphere cleanly. (They were 3-5 and 9-14 nM IIRC.) All other
> "colors" are blocked to one degree or another.

I'm talking about the light that manages to reach the camera. Ideally,
you would want the Red filter to filter everything except red
wavelengths. And so on for G and B. What apparently is happening is
that the filters have another higher gain region outside the visible
region. So some of the OEM's slap on another filter over the sensor to
block that. And that filter is not 100% effective, so IR photography is
possible by longer exposures and by blocking the visible spectrum.

Cheers,
Alan.

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.
Anonymous
May 22, 2005 8:45:53 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

In article <d6qk11$jm0$1@inews.gazeta.pl>,
Alan Browne <alan.browne@freelunchVideotron.ca> wrote:
>DoN. Nichols wrote:
>
>> In article <d6ngej$2sr$1@inews.gazeta.pl>,
>> Alan Browne <alan.browne@freelunchVideotron.ca> wrote:
>>
>>>Gisle Hannemyr wrote:
>>
>>
>> [ ... ]
>>
>>
>>><SNP>
>>>
>>>>What hobbyist call "infrared photography" should really be called
>>>>"Near-infrared photography". It is about taking photographs of
>>>>/reflected/ infrared light in a narrow spectrum just above visble
>>>>light - from 700 nm up to about 950 nm.
>>>
>>>Probably. There is so little authoratitve information on the subject
>>>with resepct to what wavelenghts are passing the RGB filters, the sensor
>>>filters and being counted by the photosites. I haven't seen a gain
>>>chart for the whole chain (in segments). Do you know where there is one?
>>
>>
>> It isn't just the filters and the sensor which control things
>> when you get to the far IR regions. There are only two bands which pass
>> the atmosphere cleanly. (They were 3-5 and 9-14 nM IIRC.) All other
>> "colors" are blocked to one degree or another.
>
>I'm talking about the light that manages to reach the camera.

Well ... the 3-5 and 9-14 nM would reach the *camera*, but are
unlikely to make it past the first piece of glass. :-)

> Ideally,
>you would want the Red filter to filter everything except red
>wavelengths. And so on for G and B. What apparently is happening is
>that the filters have another higher gain region outside the visible
>region. So some of the OEM's slap on another filter over the sensor to
>block that. And that filter is not 100% effective, so IR photography is
>possible by longer exposures and by blocking the visible spectrum.

Note that it is pretty difficult to make a really wide-spectrum
filter, so most are made to cover only the part of the spectrum needed
in a given situation.

I've worked with scientific filters which were three layers:

1) A basic high-pass filter to block everything with a wavelength
longer (or shorter) than the general region of interest.

2) A more general color filter to select a region of what the
high-pass lets through.

3) An interference filter (which gives a very narrow bandpass or
cutoff) to sharpen the cutoff at the edge of response as much as
possible.

Having seen and worked with these, I'm not going to expect the
filter areas in the Bayer filter to do more than what is absolutely
needed, and this makes the separate IR filter in the D70 make lots of
sense compared to trying to block everything in that deposited pattern
filter on the sensor itself.

Enjoy,
DoN.
--
Email: <dnichols@d-and-d.com> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
!