Jump to content
UltravioletPhotography

Multispectral vs Tricolor


Andrea B.

Recommended Posts

EDITs on 17 June 2021.

 

If those of you who are interested would kindly contribute a working description of the terms multispectral image and tricolor image, I would greatly appreciate that.

 

I want to get consensus descriptions nailed down so that we can refine tags, Stickies, tutorials and our general discussions such that everyone understands what we are talking about.

 

Here is what I was thinking, but I am entirely open to changing my own thoughts

about these descriptions. :lol:

 

In my descriptions I am making the assumption that everyone knows

what is an RGB channel stack !!!

 

DESCRIPTION

A Tricolor Image is an RGB channel stack of 3 images

where each image is made from a different sub-region within one of the following broad intervals. Typically three narrowband filters are used to photograph the subject. Ideally, the narrowband filters do not overlap or overlap very little.

  • Ultraviolet [10-400 nm]
  • Visible [400-700 nm]
  • Infrared [700-104 nm]

The Tricolor goal is to assign visible colors to photos made with "invisible" wavelengths. The selection of sub-regions may of course be restricted to smaller intervals within these larger categories. 

Tricolor Example: Red [300-320 nm] + Green [330-350nm] + Blue [360-380 nm]
where each filter is a narrow 20 nm UV-bandpass.

 

DESCRIPTION

A Multispectral Image is typically thought of as an RGB channel stack of 3 images - UV, Visible and IR -where each image is made within the following broad intervals using either a broadband or narrowband filter. Other combinations can also be considered multispectral such as the UV, Vis and Vis example below.

  • Ultraviolet [10-400 nm]
  • Visible [400-700 nm]
  • Infrared [700-104 nm]

The Multispectral goal is to produce an image which might represent the outcome of photographing a subject in mixed light or to make an image which emulates the way an animal or insect might see.

Multispectral Examples:

  • Red-UV + Green-Vis + Blue-IR
    where each image is made under a broadband filter such as the BaaderU or an RG780.
  • Red (entire UV image) + Blue (Visible Green Channel) + Green (Visible Blue Channel)
    Mix it up and see what happens.

 

The 7 Electromagnetic Wavebands:

Nobody can quite agree on the endpoints, so please don't let this bother you.

This is not Science Class. There won't be a pop quiz.

For example, Wikipedia likes to start Infrared at 750 nm. But any UV/IR photographer would  start it at 700 nm because wavelengths in 700-750 nm contaminate Visible (and UV) photos.

  • Gamma Rays
  • X-Rays
  • Ultraviolet [10-400 nm]
  • Visible [400-700 nm]
  • Infrared [700-104 nm]
  • Microwaves
  • Radio

 

Some Sub-Wavebands:

Same warning - don't sweat the endpoints.

This is just a bit of guidance for general discussion.

 

Near UV.........[300-400 nm]

Middle UV......[200-300 nm]

Far UV............[122-200]

Extreme UV....[10-121 nm]

 

UV-A.....[315-400 nm]

UV-B.....[280-315 nm]

UV-C.....[100-280 nm]

 

Near IR (NIR)...........................[700-1400 nm]

For the typical reflected IR photographer I think we would use 800-1100 nm.

Shortwave IR (SWIR)..............[1400-3000 nm]

Mid-wavelength IR (MWIR).....[3000-8000 nm]

Longwave IR (LWIR)................[8000-15000 nm]

 

NOTES:

 

1) Neither Tricolor nor Multispectral imagery is new. It's been done since the film (both still and movie) days. It is definitely much easier to make such images these days using digital files.

 

2) Yes, certain single dual bandpass filters or really wide filters might produce results which could be considered either Tricolor or Multispectral. Some light from such a filter is likely going to "contaminate" some other light, so the result may vary from what can be obtained with an RGB stack. Besides which, where's the fun?

Link to comment
Andy Perrin
Andrea, that's way too restrictive for the Tricolor. Like, what would you call my SWIR-NIR images (R=1500nm, G=1064nm, B=780nm) of which there are several examples in our forum? Do we need to distinguish multispectral from tricolor in the first place? I don't think you can do it with any rigid specification of the bands and I don't see the need for it?
Link to comment

That may be a nice definition. Probably we should expand the wavelength range to include SWIR, MWIR, LWIR, UVC, and so on. Andy already tried this in SWIR...

 

I was about to post this but Andy did it before me... anyway, I think it is actually hard to give precise definitions.

Link to comment

Yes, I definitely should expand the possible wavebands. That is my fault. I will go back and add the outer IRs !

I'm just not used to working with those yet as an editor. Andy has pioneered there in the outer IRs.

 

I don't think it is too restrictive to say that either you are working within one well-defined larger waveband

OR you are mixing things wavebands.

 

Also remember we are a photo forum. We don't need laboratory strength definitions here. Just some general guidance for members who want to experiment or play.

 

The wavebands mentioned have Generally Accepted Boundaries. If somebody needs to quibble about whether UV ends at 380 nm or at 400 nm, well, that's ok. If you like 380 nm, then use it. Same for the other beginnings or endings. That doesn't really affect the *idea* of tricolor or multispectral.

 

Do we need to distinguish multispectral from tricolor in the first place?

Well, I would like to TRY to do this.

Traditionally these kinds of photos have always been called multispectral in the photo forums up until recently when the Tricolor usage hit the forum for digital work. Tricolor in the past was used in the context of still film and movie film work.

By the definitions above a Tricolor is always a multispectral but not the inverse.

 

There is also usage of both terms outside our particular forum.

Link to comment

Andrea, that's way too restrictive for the Tricolor. Like, what would you call my SWIR-NIR images (R=1500nm, G=1064nm, B=780nm) of which there are several examples in our forum? Do we need to distinguish multispectral from tricolor in the first place? I don't think you can do it with any rigid specification of the bands and I don't see the need for it?

Sorry to go off topic but may I see those images you refer to? Is that the one with the icicles? I don't remember.

Link to comment

I think Andrea has it correct. Tricolor of the past was blue #47, green #58, and red #25. All in the visible.

Ir tricolor would be within the IR range 700, 800, 900 as sensor most of us have cut off at 1086nm, the limit for silicone.

 

Uv becomes odd then. Is tricolor uv just low, mid and upper UVA?

 

So my desire to do 254nm, 313nm and 370nm, is that multispectral or tricolor? Maybe multispectral, as 2 of those are outside what most can image.

 

Tricolor in UV, only being in UVA might just be too limiting. There isn't huge difference in 335, 370, 390, but there is some.

UVB is just bicolor than as only 303nm and 313nm is easy with Mercury bulbs. UVC is just 254nm, as main peak.

Link to comment

Well, UVB is a very narrow band anyway, being defined as 280-320 nm or 280-315 nm.

 

I think a more relaxed definition would be OK.

Link to comment

The FDA defines UVB as 290nm to 320nm. UvAii as 320nm to 340nm, then UVA as 340nm to 400nm.

https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=201.327

 

I have found that to be a little odd.

 

I wouldn't mind pushing UVB to 290nm to 340nm. That makes more sense to me.

As in Vacuum UV below 200nm. UVc 200nm to 290nm. UVB 290nm to 340nm and UVA 340nm to 400nm.

But thats only me.

Link to comment
Andy Perrin

See, you just get caught in tangled discussions about where to make arbitrary spectral cutoffs, and I don't think the distinction is that useful in the first place. Let's let them be undefined terms, which we have even within mathematics.

 

To paraphrase a famous judge who was ruling on something else altogether, "What's a tricolor? Well, I know it when I see it."

Link to comment

It is easy enough to add in a Shortwave-UV band.

 

**********

 

If Multispectral and Tricolor are “undefined terms”, then how would you teach someone to make a Multispectral or Tricolor image? How would you explain to a new member which tag to use? Quite obviously such images can be *easily* described in concrete terms, so call it a Description instead of a Definition. :grin:

 

***********

 

I don’t think the “logic” of photography, such as it is (ha-ha), requires us to consider Multispectral or Tricolor as primitives.

(Just kidding there…….)

 

**********

 

Actually the most descriptive terminology for either is probably “Channel Stack”.

If that were used, then you would just precede the phrase with what waveband is placed in the R, G or B channels.

Example: Vis/IR/UV Channel Stack or 254nm/340nm/375nm Channel Stack and so on.

 

“Multispectral” is rather vague unless you define/describe it. In some sense all images are “multispectral”.

“Tricolor” is rather meaningless because every digital image has an R, G and B channel.

 

Of course every digital image is also a Channel Stack.

So there you are, full circle.

Cue Quackydoodle, stage left, with trumpet……

 

**************

 

The spectral cutoffs are Generally Accepted as I wrote above. While I do grant that there is some small amount of variation on the UV side for where UV ends (380nm or 400nm) or where UVB begins, I don’t think it is entirely arbitrary. Articles, authorities, textbooks, etc state their endpoints so you always know what your context is going to be.

(Just for the record, in my 15 or so years of foto forums, UV is 300-400 nm and IR is 700-1100 nm.)

 

**********

 

ok, late night musings over…..

Link to comment

I think this is a valid discussion, if we agree that this distinction is no scientific classification, just an internal agreement withing UVP to facilitate search and discussion.

 

As an internal agreement, my vote goes to simplicity, i.e. for me, the following would make sense.

 

We define the usual three regions, purely for convenience and easy conversation:

  1. UV, everything below 400nm
  2. VIS, everything between 400 and 700nm
  3. IR, everything above 700nm

and then:

  • Tricolor: R, G, and B from the same region
  • Multispectral: R, G, B from at least two regions
  • Hyperspectral (???): arbitrary filters allowing light from at least two regions (not using the R, G, B at all)

 

If you want, you may add an additional specifier, like UVA-Tricolor, or SWIR-Tricolor, or UVA-NIR-Multispectral. This would make it clear at one glance what we're talking about, and searching would work well.

 

I know that we're talking about arbitrary stuff here, but, look, if you think that's nonsense, then let's also forget the distinction between UV, VIS, IR, Radio, X-Rays, Gamma-rays, etc., because all of these are electromagnetic waves, so why bother with names? Let's just rename the site to ElectroMagneticImaging.com and have done with it :wink:

 

And if you don't want to use this scheme, then don't, but your articles might then not be easily found by any new (and old) members. But we all know that if a group of people can't agree on some basic rules about communication, then there will be no useful communication, period. And rules are by definition arbitrary, they are just what the group agrees on as common ground. Why do we use English as common language here at UVP? Why don't we allow everybody to write in their own language? Clearly, because this would make no sense at all, as long we don't have the Babelfish :tongue:

Link to comment

Let's just rename the site to ElectroMagneticImaging.com and have done with it :wink:

Or InvisiblePhotography.com... we are no longer limited to UV here, but UVP is historical, such as tinfoil which is no longer made of tin but aluminium.

 


I agree with you, one can specify the bands used and this would make everything clear. So in Andy's case, he should write NIR-SWIR multispectral.

Link to comment

So in Andy's case, he should write NIR-SWIR multispectral.

 

As a matter of fact, in this case, I would actually write IR-tricolour, at least as far as thread-title and search-tags are concerned, because all is IR :smile:

 

I haven't followed the path to tricolour and multispectral yet, so perhaps I ought to shut up, but if I ever decide to follow up on this, and find that at UVP there is no common name for photos like this, and they are referred to as "NIR-SWIR multispectral", "NIR-SWIR-SWIR multispectral", "IR-tricolour", or "NIR-SWIR-SWIR tricolour", or something else completely, depending on who wrote the post, then I'd just shrug my shoulders and try to find information elsewhere.

 

Like I wrote, what do we want, a (more or less) scientific exact definition, or a common ground for discussions?

Link to comment
Andy Perrin

StephanN, that’s acceptable to me although I still think this whole discussion is pointless (we have gotten along fine up to now without formal definitions, and these feel unnecessary).

 

Stefano, it has occurred to me more than once that “invisiblephotography.com” is probably a more apt site name these days. I’m not insisting anyone change it, but food for thought...

Link to comment

When you are running a forum like this some day, you will see the need to organize posts and topics

so that people can get the information they need. :grin: :grin: :grin:

There is also a need to explain and teach some of the UV/IR topics.

Proper usage of photography terms is necessary.

 

As an example, the Stickies here have been accessed thousands of times.

The Stickies have existed for years before any of the recent books have been written,

and writers come here to get information. Birna and I have been "giving away" all this information

for "free". We have not written books*. We have not kept anything "secret". :grin:

 

So it is part of my job to attempt to keep information organized.

Call it a "public service" if you like. :cool:

 

*Please don't get me wrong. I love the books which have been written and have them all.

And I hope people buy them and find them useful for our wonderful UV/IR photography !!!


 

If I changed the name of the forum, then I would have to maintain

both the old URL and the new URL so that forwarding would work.

I don't think I'm going to do that.

 

Aside from which, most good URLs

referencing UV/IR photography and "invisible light" are already taken.

 

Website for invisiblelight.com.

I am not sure what "invisible light" has to do with pubic speaking. Oh well.

 

Website(?) for invisiblephotography.com

Somebody is simply holding this URL hoping to sell it for $119 (current)

and perhaps make a small profit.


 

StephanN: I like your simplification of the Definitions/Descriptions.

I am going to incorporate your suggestions. THANK YOU !!!

 

.

Link to comment

we are no longer limited to UV here...

 

Just for the record, we never have been! :lol: :lol: :lol:

Link to comment

Here is the re-write repeated here from the edits in the first post.

So you don't have to scroll up. La!!

 

A Big Thank You to StephanN for reminding me to Keep Things Simple.

See the new Descriptions below.

 

 

The 7 Electromagnetic Wavebands:

Nobody can quite agree on the endpoints

Please don't let this bother you.

This is not Science Class. There won't be a pop quiz.

 

For example, Wikipedia likes to start Infrared at 750 nm. But any UV/IR photographer would

start it at 700 nm because wavelengths in 700-750 nm contaminate Visible (and UV) photos.

  • Gamma Rays
  • X-Rays
  • Ultraviolet [10-400 nm]
  • Visible [400-700 nm]
  • Infrared [700-104 nm]
  • Microwaves
  • Radio

.

Some Sub-Wavebands:

Same warning - don't sweat the endpoints.

This is just a bit of guidance for general discussion.

  • Near UV.........[300-400 nm]
  • Middle UV......[200-300 nm]
  • Far UV............[122-200]
  • Extreme UV....[10-121 nm]

  • UV-A.....[315-400 nm]
  • UV-B.....[280-315 nm]
  • UV-C.....[100-280 nm]

  • Near IR (NIR)...........................[700-1200 nm]
    For the typical reflected IR photography I think we would use 800-1100 nm.
  • Shortwave IR (SWIR)..............[1200-3000 nm]
  • Mid-wavelength IR (MWIR).....[3000-8000 nm]
  • Longwave IR (LWIR)................[8000-15000 nm]

.

DESCRIPTION

A Tricolor Image is an RGB channel stack

for which each image is made from a different sub-waveband within one of the following broad intervals.

Typically three narrowband filters are used to photograph the subject.

Ideally, the narrowband filters do not overlap or overlap very little.

 

The Tricolor goal is to assign visible colors to photos made with "invisible" wavelengths.

The selection of sub-wavebands may of course be restricted to smaller intervals

within these larger categories. For example, a Tricolor SWIR image could be constructed

if appropriate narrowband filters were available.

  • Ultraviolet [10-400 nm]
  • Infrared [700-104 nm]

.

DESCRIPTION

A Multispectral Image is an RGB channel stack

for which each image is made from any sub-waveband, either broad or narrow,

chosen from the following wavebands.

 

The Multispectral goal is to produce an image which might represent the outcome of photographing a subject in mixed light or to make an image which emulates the way an animal or insect might see.

  • Ultraviolet [10-400 nm]
  • Visible [400-700 nm]
  • Infrared [700-104 nm]

.

Examples:

  • Tricolor: Red [300-320 nm] + Green [330-350nm] + Blue [360-380 nm]
    where each filter is a narrow 20 nm bandpass used to create a Near UV stack.

  • Multispectral: Red-UV + Green-Vis + Blue-IR
    where each image is made under a broadband filter such as the BaaderU or an RG780.

  • Multispectral: Red (entire UV image) + Blue (Visible Green Channel) + Green (Visible Blue Channel)
    Mix it up and see what happens.

 

.

NOTES:

 

1) Neither Tricolor nor Multispectral imagery is new. It's been done since the film (both still and movie) days. It is definitely much easier to make such images these days using digital files.

 

2) Yes, certain single dual bandpass filters or really wide filters might produce results which could be considered either Tricolor or Multispectral. Some light from such a filter is likely going to "contaminate" some other light, so the result may vary from what can be obtained with an RGB stack. Besides which, where's the fun?

 

3) In the olden days of UV forums, we had a wonderful stacking app written by Ben Lincoln.

Here's the link: https://www.beneatht...ace_Breaks.html

Someone should take that and update it.

Link to comment
Andy Perrin
Andrea, can we put the SWIR cut-on at 1200nm? That is where silicon's range ends, and it seems to me that if you care about search utility here, then that's valuable information to a searcher. Some people go as low as 900nm for SWIR (where InGaAs starts) but I think that's less useful on here than knowing where silicon ends.
Link to comment

will do.

Thank you for that input to make it more relevant to photography done here on UVP.

Link to comment

A Big Thank You to StephanN for reminding me to Keep Things Simple.

 

Well, you know, the famous KISS (Keep It Simple, Stuipid!!!!), in fact started out as KISSPLS (Keep It Simple for Stupid People Like Stephan), but people started using it in the sense of "A kiss, please!", so it had to be changed :grin:

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...