Using primary color light filters

[LightningInAJar]

I wanted to try taking 3 camera images using red, green, and blue filters and combining them as color channels to see what that might look like.

A) Should this produce more accurate colors, particularly in low light?

B) Do they even sell filters that only let light pass that can activate the red, green or blue photosites?

 

[Tom.G]

B) Do they even sell filters that only let light pass that can activate the red, green or blue photosites?

Yes; color filters for cameras, also color filters for theater spotlights. The latter are often called 'Gels'.

A couple brand names are ROSCO and CineFilters. Both at least used to have sample packs available.

Good luck though (but I would like to see your results). The human eye color response is a bit more complex than that.

https://www.physicsforums.com/posts/6034710

This threads may also be of interest:
https://www.physicsforums.com/posts/7127139

Those, and many others found using the SEARCH feature.
https://www.physicsforums.com/search/21870330/?searchform=1&q=color+response+eye&o=relevance

Cheers,
Tom

 

[sophiecentaur]

A) Should this produce more accurate colors, particularly in low light?

You would need to define what you mean by "accurate". The present TV (and other) systems can reproduce 'accurately' colours which lie within the gamut of the RGB primaries on the screen under the right conditions. Filters are available with many different characteristics (in cameras etc).

You have to consider the fact that a good colour reproduction system has to deal with all spectral content over a continuous range of wavelengths . You cannot have gaps between the filter passbands or you could lose some wavelengths. It is impossible (a practical problem) to produce three analysis filters that have 'rectangular' passbands. Hence they use filter characteristics which overlap each other. The filters are similar to the responses of the human eye sensors.

The components in the overlap are 'processed together' and the result gives a flat response over the whole visible range when recombined on a TV display. Try starting with this wiki link.
A diifferent set of filters may be used for astrophotography; some of the filters are very narrow band for picking out particular spectral wavelengths.

 

[russ_watters]

I wanted to try taking 3 camera images using red, green, and blue filters and combining them as color channels to see what that might look like.

A) Should this produce more accurate colors, particularly in low light?

B) Do they even sell filters that only let light pass that can activate the red, green or blue photosites?

Yes, this is standard practice for astrophotography, so there are tons of options out there if you do a search for "rgb filters". Example:

https://www.highpointscientific.com/brands/astronomik-filters/lrgb-filters

 

[sophiecentaur]

A couple brand names are ROSCO and CineFilters.

Those filters are suitable for a fun demo to see how mixes of levels of three primary sources can produce a range of colours using three (good) white light sources on a white projection screen, each with a different filter. But using gels for analysis is very approximate because (for the OP's requirement for 'More accurate' results) they can't be relied on to mimic human colour analysis.

Colourimetry is a lot more complicated that it first appears. The familiar RGB values are a common intermediate step which gives a chance that a camera (one make with its own filters) image data gives a convincing image on a display (synthesis) with its particular set of primaries.

 

[Andy Resnick]

I wanted to try taking 3 camera images using red, green, and blue filters and combining them as color channels to see what that might look like.

A) Should this produce more accurate colors, particularly in low light?

B) Do they even sell filters that only let light pass that can activate the red, green or blue photosites?

I depends on your camera. If you have a typical color camera there is already a Bayer filter present and your idea will not work. If your camera is monochrome, you are in business.

 

[LightningInAJar]

By more accurate colors I mean with Christmas lights at night most of the scene is dark, but the lights sometimes blow out to white. I was just wondering if a physical filter could aid in getting colors more correct, particularly in low light when I probably need longer exposures.

And I assume as far as filters I would need three filters that are centered around the 3 most sensitive wavelengths, and offer the same overlap as the photosites do so there is no gap between them.

What are the best commercial camera filters for this? I was hoping to find slide filters from Cokin as I already have the slots lens adapter.

 

[sophiecentaur]

By more accurate colors I mean with Christmas lights at night most of the scene is dark, but the lights sometimes blow out to white.

It seems to me that the exposure control on your camera is over-cooking the picture. With a mainly dark scene the poor old camera can only assume it should choose a long exposure. That's nothing to do with the analysis (Bayer) filter; it's just horses for courses. See if you can override (reduce) the auto expose. Adding further filters will reduce the light energy reaching the sensor so the exposure will automatically increase further.

There is a further problem, of course. If you plan to insert all three of your gel filters, on top of each other, over the lens of your camera then you will admit 'no light' into it.

A similar problem arises when people try to take a photo of the Moon. That beautiful subtle silver object turns out like a fuzzy white disc.

 

[LightningInAJar]

It seems to me that the exposure control on your camera is over-cooking the picture. With a mainly dark scene the poor old camera can only assume it should choose a long exposure. That's nothing to do with the analysis (Bayer) filter; it's just horses for courses. See if you can override (reduce) the auto expose. Adding further filters will reduce the light energy reaching the sensor so the exposure will automatically increase further.

There is a further problem, of course. If you plan to insert all three of your gel filters, on top of each other, over the lens of your camera then you will admit 'no light' into it.

A similar problem arises when people try to take a photo of the Moon. That beautiful subtle silver object turns out like a fuzzy white disc.

I was using manual control. I wasn't letting camera determine exposure.

What do astrophotographers use the color filters for? More for visual effect or to actually clarify the image accuracy?

 

[sophiecentaur]

I was using manual control.

So you were probably overexposing it manually. If the christmas lights were nicely visible in daylight then your exposure should have been chosen as if it were daylight. Camera have a problem with high contrast ratios so you have to help them by placing wanted objects well within their range. Photos of the Moon should be set up as if or an object in full sunlight; not obvious until one has tried.

You haven't responded to the question about the camera. Is it a colour or a monochrome camera?

What do astrophotographers use the color filters for? More for visual effect or to actually clarify the image accuracy?

Three main reasons. The usual reason is to deal with light pollution. That will improve contrast by reducing the background. (for instance sodium D lines from street lamps. Although many people use standard DSLRs, monochrome cameras are very often used. Then there will be three or more separate exposures to pick out particular colours of , for instance nebulae.
There are certain spectral lines which can be enhanced to show regions / features of hydrogen gas.

An interesting fact about the early days of astrophotography with film is that, until panchromatic emulsion was invented, photos of the heavens only showed the shorter wavelengths, missing out IR in particular.

 

[LightningInAJar]

So you were probably overexposing it manually. If the christmas lights were nicely visible in daylight then your exposure should have been chosen as if it were daylight. Camera have a problem with high contrast ratios so you have to help them by placing wanted objects well within their range. Photos of the Moon should be set up as if or an object in full sunlight; not obvious until one has tried.

You haven't responded to the question about the camera. Is it a colour or a monochrome camera?

Three main reasons. The usual reason is to deal with light pollution. That will improve contrast by reducing the background. (for instance sodium D lines from street lamps. Although many people use standard DSLRs, monochrome cameras are very often used. Then there will be three or more separate exposures to pick out particular colours of , for instance nebulae.
There are certain spectral lines which can be enhanced to show regions / features of hydrogen gas.

An interesting fact about the early days of astrophotography with film is that, until panchromatic emulsion was invented, photos of the heavens only showed the shorter wavelengths, missing out IR in particular.

It is a Canon Rebel so it's full color. I don't know quite what you mean as Christmas lights typically aren't on during the day so I can't say how well they could be seen during daylight. I guess I just want the lights to appear a certain size of light source, but don't want to blow out the color. It seems like blue and green lights lose their saturation and brightness sooner compared to the warmer colors.

I assumed the wavelengths of light on earth are too similar to those skyward to what filters help the most for astrophotography? Reducing red light primarily?

 

[sophiecentaur]

It is a Canon Rebel so it's full color. I don't know quite what you mean as Christmas lights typically aren't on during the day so I can't say how well they could be seen during daylight. I guess I just want the lights to appear a certain size of light source, but don't want to blow out the color. It seems like blue and green lights lose their saturation and brightness sooner compared to the warmer colors.

If the camera is "full colour" then it will not be helped in this situation by adding more filters.

If the lights look ok to your eye then choosing the correct exposure will make the lights similarly visible. I recommend finding what exposure the camera gives in normal daylight (in the location of the lights) and then use the same exposure at night with the lights on. Are these Public lights which may not be on in the day but try late afternoon.

If the lights appear to be oversized then that is yet another confirmation that they are overexposed. The same happens with pictures of the Moon in auto expose mode- the Moon just appears very bright and oversized with no detail. When a picture of the Moon is correctly exposed (it's an object in full sunlight and should be exposed appropriately) then the background stars will be very faint because the camera doesn'.t have sufficient contrast range

The camera may well have different overload characteristics when the lights are burning out. Instead of arguing your case again and again, why not do the experiment with a range of underexposure settings. I think you may have an erroneous model in your head about how the camera and sensors work. Your suggestion of adding filters is a red herring.

 

[LightningInAJar]

If the camera is "full colour" then it will not be helped in this situation by adding more filters.

If the lights look ok to your eye then choosing the correct exposure will make the lights similarly visible. I recommend finding what exposure the camera gives in normal daylight (in the location of the lights) and then use the same exposure at night with the lights on. Are these Public lights which may not be on in the day but try late afternoon.

If the lights appear to be oversized then that is yet another confirmation that they are overexposed. The same happens with pictures of the Moon in auto expose mode- the Moon just appears very bright and oversized with no detail. When a picture of the Moon is correctly exposed (it's an object in full sunlight and should be exposed appropriately) then the background stars will be very faint because the camera doesn'.t have sufficient contrast range

The camera may well have different overload characteristics when the lights are burning out. Instead of arguing your case again and again, why not do the experiment with a range of underexposure settings. I think you may have an erroneous model in your head about how the camera and sensors work. Your suggestion of adding filters is a red herring.

How would an image look differently lets say during the daytime of nonilluminated things with appropriate exposers if it's just a color image versus 3 B&W photos using 3 filters and combined as channels afterwards? Would they look all that different?

 

[russ_watters]

How would an image look differently lets say during the daytime of nonilluminated things with appropriate exposers if it's just a color image versus 3 B&W photos using 3 filters and combined as channels afterwards? Would they look all that different?

Astrophotograpers do it with monochrome cameras for three main reasons:

1. Higher resolution because every pixel gets used for every color.

2. Higher sensitivity when combining with a luminance (all colors/clear) filter, for dim objects.

3. Ability to take different exposures of different colors.

The only one you can do with an already color camera is #3 but you don't need separate filters to do it, just software to combine separate exposures of one color channel each.

 

[Andy Resnick]

I was using manual control. I wasn't letting camera determine exposure.

This (overexposure) is your actual problem here. When you manually expose, how do you meter the scene? For myself (not a professional), I use 'live view', zoom into a bright area, and dial in the exposure accordingly. Note- that fails if the bright area is monochromatic-ish, I assume b/c of the white-balance process.

But then, if I underexpose to save the highlights, most of the image is black, and that's primarily because our eyes are logarithmic detectors while solid state detectors are linear. That's one reason film is a superior detector- film has a nonlinear intensity response and so can better replicate how the image appears to your eye.

In digital, some people use HDR to better match the dynamic range. Others do this manually (use an app to manipulate a single image). I remember seeing a good online tutorial but can't find it now...

 

[LightningInAJar]

This (overexposure) is your actual problem here. When you manually expose, how do you meter the scene? For myself (not a professional), I use 'live view', zoom into a bright area, and dial in the exposure accordingly. Note- that fails if the bright area is monochromatic-ish, I assume b/c of the white-balance process.

But then, if I underexpose to save the highlights, most of the image is black, and that's primarily because our eyes are logarithmic detectors while solid state detectors are linear. That's one reason film is a superior detector- film has a nonlinear intensity response and so can better replicate how the image appears to your eye.

In digital, some people use HDR to better match the dynamic range. Others do this manually (use an app to manipulate a single image). I remember seeing a good online tutorial but can't find it now...

I have used zoom to manually adjust focus for distant objects, but as far as long exposures I'm not sure the view screen will give me an accurate idea of what I'll end up with. I took many photos with slight changes. I have exposure stacked to increase dynamic range, but honestly I don't care as much about the darker aspects if I can't yet get the lit part correct.

 

[sophiecentaur]

, but honestly I don't care as much about the darker aspects if I can't yet get the lit part correct.

You will get the "lit part" right if you use a much reduced exposure. Have you actually tried reducing the exposure by four 'stops'? You will not see the darker background detail but the lights will look better. Sorry to doubt you but when you 'reduce' exposure are you controlling both aperture and exposure time? Your description of results could imply that the camera mode you're using could be Aperture Priority or Exposure Time priority. You need fully manual control.

PS you don't have to visit the site of your experimental Cristmas light pictures. You can experiment using a light bulb in the home and 'control' the burn out by using exposure control. You should be able to read the printing on the bulb itself if you use a low enough exposure. With digital you can see results immediately. In the days of film you had to process images before you could see your results.

 

[Andy Resnick]

but as far as long exposures

Everyone has been trying to tell you to try using shorter exposure times.

 

[LightningInAJar]

You will get the "lit part" right if you use a much reduced exposure. Have you actually tried reducing the exposure by four 'stops'? You will not see the darker background detail but the lights will look better. Sorry to doubt you but when you 'reduce' exposure are you controlling both aperture and exposure time? Your description of results could imply that the camera mode you're using could be Aperture Priority or Exposure Time priority. You need fully manual control.

PS you don't have to visit the site of your experimental Cristmas light pictures. You can experiment using a light bulb in the home and 'control' the burn out by using exposure control. You should be able to read the printing on the bulb itself if you use a low enough exposure. With digital you can see results immediately. In the days of film you had to process images before you could see your results.

I was using manual but largely kept aperture pretty wide open I think. I actually don't remember. With a longer expose in like 6 seconds or longer range I imagine I might of got up to F/8. I would need to inspect the files to be sure though.

 

[LightningInAJar]

Everyone has been trying to tell you to try using shorter exposure times.

How short? ISO only as high as 800 to keep noise down? Wide open aperture?

 

[sophiecentaur]

How short? ISO only as high as 800 to keep noise down? Wide open aperture?

Normal experimental approach is to alter just one variable and observe the results. Then head in the appropriate direction. Then change another variable if necessary. Taking notes is a good habit.

Question: What would you have to do with exposure control if you wanted less light energy onto the sensor?

If ISO is changed during your exposure it will just introduce confusion. Sit down and see the effects of varing exposure time and aperture on the scene where you are on this forum. Observe what you have to do to get a darker and darker image. You have a misconception somewhere with this problem. Try to identify it and you can expect some success.

Everyone has been trying to tell you to try using shorter exposure times.

This forum is polite. That post should, perhaps be in upper case and the OP might get the message.