mike earussi
Veteran Member
Would an arc light work?
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Spectrum Profile lines
- Thread starter D Cox
- Start date
DavidWright2010
Senior Member
Maybe. Here's the spectrum of a Xenon arc:
You'd want a lamp that filtered out the UV and IR.
David
DavidWright2010
Senior Member
And I had thought, back then, 'when comparing a foveon with a Bayer sensor, why not do several?'
So I included Pentax, Nikon, and iPhone.
Let me look over my results and see if I saw a measurable difference in those three.
dbateman
Senior Member
Interesting idea.
I woukd say you are getting there. How I would test this.
1. Get a pinhole lens, like the pinhole pro that came out with 58mm thread. I have one and its nice.
2. Cut a 1000 line/mm grating into an old circular polarizers filter holder. You need to be able to turn the grating. And place on your pinhole lens.
3. Get a hallogen bulb. Now you can buy them inside E26/E27 A19 bulbs. These work well.
4. Distance the camera on a tripod from the bulb to get the full color spectrum. Capture your images. Then test all your cameras at same distance.
5. Use the free 4channels, which can be extracted from the libraw download. Its in the bin folder. The program is command line, but will output tiff images of the individual color channels.
6. Then align and look at the spectrum shifts in full image to individual channels.
I have done all the above, but steps 5 and 6 with a full spectrum Olympus Em1 to look at different filter spectal cut offs. So I will have to see what image data I have for my camera.
I woukd say you are getting there. How I would test this.
1. Get a pinhole lens, like the pinhole pro that came out with 58mm thread. I have one and its nice.
2. Cut a 1000 line/mm grating into an old circular polarizers filter holder. You need to be able to turn the grating. And place on your pinhole lens.
3. Get a hallogen bulb. Now you can buy them inside E26/E27 A19 bulbs. These work well.
4. Distance the camera on a tripod from the bulb to get the full color spectrum. Capture your images. Then test all your cameras at same distance.
5. Use the free 4channels, which can be extracted from the libraw download. Its in the bin folder. The program is command line, but will output tiff images of the individual color channels.
6. Then align and look at the spectrum shifts in full image to individual channels.
I have done all the above, but steps 5 and 6 with a full spectrum Olympus Em1 to look at different filter spectal cut offs. So I will have to see what image data I have for my camera.
Electronic flash has a spectrum near to that of daylight.
The channel you need for my approach is the Hue channel in HSV or HSL space.
Last edited:
dbateman
Senior Member
Sorry Dcox, I wasn't clear. I was responding to Saltydogstudio.
Raw digger would be best to quantitatively analyze the data to determine CFA spectral response. However, not many know that 4channels will output the individual color channels and is free.
You could then use histogrammer or ImageJ, which are also free for analysis. But would be more work.
Dcox, I have not looked into the Hue channels as you are and I am not too familiar with that analysis method.
Raw digger would be best to quantitatively analyze the data to determine CFA spectral response. However, not many know that 4channels will output the individual color channels and is free.
You could then use histogrammer or ImageJ, which are also free for analysis. But would be more work.
Dcox, I have not looked into the Hue channels as you are and I am not too familiar with that analysis method.
mike earussi
Veteran Member
Why? A cutoff filter is already built into the sensor so you wouldn't be photographing the UV and IR anyway.Maybe. Here's the spectrum of a Xenon arc:
You'd want a lamp that filtered out the UV and IR.
David
saltydogstudios
Senior Member
Interesting.
I also have a full spectrum Olympus E-PM1.
So here's what I'm thinking. I have a bunch of cameras I want to compare. If I use diffraction grating, how can I accurately line up the frequency ranges across cameras? I'm just trusting what the camera sees because the diffraction grating is so sensitive to slight movements. That is - how do I know the spectrum isn't shifted a bit from one camera to another if I try to line up the spectrums in post production?
Would I be better off with a prism and either a strobe or a constant light source? This way I could project it on to a surface - keeping those constant and putting the camera wherever makes sense for the given lens (I'm dealing with different lens mounts). Like - even if it's just a sheet of paper, the edges of the paper can be lined up in post production so I know that I'm lining up the "rainbow" the same in each file.
This way it's easier to line up the images in post and get a more apples-to-apples comparison.
I can play with the diffraction grating for projection once I get it.
--
"Wait let me comb my hair and put on a tie."
It Gets Nerdy: https://medium.com/ice-cream-geometry
Sometimes I take photos: https://www.instagram.com/sodiumstudio/
Last edited:
DavidWright2010
Senior Member
The classic method is to generate the spectrum of something that gives narrow emission lines. You may be able to use a fluorescent light, once your apparatus is set up:Interesting.
I also have a full spectrum Olympus E-PM1.
So here's what I'm thinking. I have a bunch of cameras I want to compare. If I use diffraction grating, how can I accurately line up the frequency ranges across cameras? I'm just trusting what the camera sees because the diffraction grating is so sensitive to slight movements. That is - how do I know the spectrum isn't shifted a bit from one camera to another if I try to line up the spectrums in post production?
For the actual light source for testing cameras, Mike (upthread) suggested a halogen light. I tried that setup last night with my prism and the spectrum displayed on the wall seemed promising - broad and featureless.
David
saltydogstudios
Senior Member
Thank you, I was considering halogen - as I happen to have one and it's fairly easy to set up. My other thought was a strobe like a Vivitar 285 HV or a Nikon strobe that I have - they're small enough that I should be able to baffle the the light a bit to create a more narrow beam if needed.The classic method is to generate the spectrum of something that gives narrow emission lines. You may be able to use a fluorescent light, once your apparatus is set up:Interesting.
I also have a full spectrum Olympus E-PM1.
So here's what I'm thinking. I have a bunch of cameras I want to compare. If I use diffraction grating, how can I accurately line up the frequency ranges across cameras? I'm just trusting what the camera sees because the diffraction grating is so sensitive to slight movements. That is - how do I know the spectrum isn't shifted a bit from one camera to another if I try to line up the spectrums in post production?
For the actual light source for testing cameras, Mike (upthread) suggested a halogen light. I tried that setup last night with my prism and the spectrum displayed on the wall seemed promising - broad and featureless.
David
Then project that onto a WhiBal grey card that promises to be spectrally neutral across the visible spectrum and a bit into IR/UV as well.
My understanding is that halogen lights are pretty broad spectrum, even into UV and a UV filter has to be put in front of them to prevent us from burning our eyeballs.
I noticed on a recent Jet Blue flight that my "black" bag showed up as purple under the halogen light. I'm sure I got a few hours of eyeball burning UV and will be bringing eye protection on future flights.
I'll experiment with both and choose whatever gives me the best results with the least effort.
--
"Wait let me comb my hair and put on a tie."
It Gets Nerdy: https://medium.com/ice-cream-geometry
Sometimes I take photos: https://www.instagram.com/sodiumstudio/
dbateman
Senior Member
Interesting.
I also have a full spectrum Olympus E-PM1.
So here's what I'm thinking. I have a bunch of cameras I want to compare. If I use diffraction grating, how can I accurately line up the frequency ranges across cameras? I'm just trusting what the camera sees because the diffraction grating is so sensitive to slight movements. That is - how do I know the spectrum isn't shifted a bit from one camera to another if I try to line up the spectrums in post production?
Would I be better off with a prism and either a strobe or a constant light source? This way I could project it on to a surface - keeping those constant and putting the camera wherever makes sense for the given lens (I'm dealing with different lens mounts). Like - even if it's just a sheet of paper, the edges of the paper can be lined up in post production so I know that I'm lining up the "rainbow" the same in each file.
This way it's easier to line up the images in post and get a more apples-to-apples comparison.
I can play with the diffraction grating for projection once I get it.
saltydogstudios
Senior Member
Interesting.I wouldn't use a strobe. It would take too many images to line up everything. A constant light source and live view on the camera will save you a lot of time.Interesting.
I also have a full spectrum Olympus E-PM1.
So here's what I'm thinking. I have a bunch of cameras I want to compare. If I use diffraction grating, how can I accurately line up the frequency ranges across cameras? I'm just trusting what the camera sees because the diffraction grating is so sensitive to slight movements. That is - how do I know the spectrum isn't shifted a bit from one camera to another if I try to line up the spectrums in post production?
Would I be better off with a prism and either a strobe or a constant light source? This way I could project it on to a surface - keeping those constant and putting the camera wherever makes sense for the given lens (I'm dealing with different lens mounts). Like - even if it's just a sheet of paper, the edges of the paper can be lined up in post production so I know that I'm lining up the "rainbow" the same in each file.
This way it's easier to line up the images in post and get a more apples-to-apples comparison.
I can play with the diffraction grating for projection once I get it.
My plan was to project the strobe through a prism or whatever onto a grey card. The grey card has markings on it.
I can use a constant light source, but I think once I get the initial setup, I can just use the grey card to set exposure in camera. I find this particular grey card is a bit above 18% middle grey, but very neutral and I trust it for setting white balance in camera and in post.
Yes - I expect that different lenses will have different characteristics and I won't be able to completely subtract the effects of the lenses on the image. Sigma's DP (dp) series for instance, it's impossible to change the lens.
With a few exceptions, they're all modern lenses and in their native mounts - camera manufacturers take their lenses into consideration when tweaking sensors, so I don't mind this too much.
Getting to a theoretically pure ideal that's beyond reproach methodologically is nice - but I'm never going to use these cameras without lenses so I don't feel too bad about having lenses as part of the equation.
I'll keep in mind the advice to use several lenses if I have them.
I just want to see a full rainbow on the grey card and use the edges of the grey card to align different cameras after the fact. Knowing the specific frequency cutoffs would be interesting, I'm more interested in seeing how each camera's CFAs are tuned - albeit with the lens attached.
The plan is just to get the grey card to fill the frame with a slight edge around it between cameras. Then in post I can easily line up the grey card images and be confident that the different frequencies are hitting the same spots between cameras.
Remember where I said that I didn't have a Leica to test with? That ought to answer your question. LOL. I'm trying to test cameras from the major manufacturers. Some are quite old but happen to be the ones available to me.
My main goal is to help me decide what mirrorless system I may want to invest in. Even if I'm not testing those cameras directly, a several-generations-ago Canon or Panasonic or whatever will be instructive since I'm hoping camera manufacturers tend to optimize for the same things across generations.
Though according to the internet rumor mill some camera manufacturers have more consistent colors (Canon) than others (Leica, Nikon).
This spectrum analysis test isn't really the main test - just something I thought I'd try out while I have so many cameras until I have to return/sell them - and if I could have something to defend myself - a single document I can point to where I gather as much evidence I can from anywhere I can that points to different sensors having different responses to light - so much the better to avoid those discussions in the future.
BTW DXO does a similar test.
https://www.dxomark.com/Cameras/Nikon/D850---Measurements
Their tests are pretty comprehensive, but they're also measuring something slightly different than what I'm looking for. If you look at the "Color Response" chart, it looks like they just shine pure R, G or B light into the camera and measure the response of each CFA color.
Similarly the SNR and Tonal Range charts test response to color across the spectrum.
What I'm looking for is how each CFA is tuned across that spectrum. Those DXO tests should be enough to convince anyone that not all sensors (meaning RAW files) have the same response to light.
--
"Wait let me comb my hair and put on a tie."
It Gets Nerdy: https://medium.com/ice-cream-geometry
Sometimes I take photos: https://www.instagram.com/sodiumstudio/
A lens can't change the wavelength of light. It can change the intensity at any wavelength, but that is irrelevant when you are looking at the accuracy of the hues.Interesting.I wouldn't use a strobe. It would take too many images to line up everything. A constant light source and live view on the camera will save you a lot of time.Interesting.
I also have a full spectrum Olympus E-PM1.
So here's what I'm thinking. I have a bunch of cameras I want to compare. If I use diffraction grating, how can I accurately line up the frequency ranges across cameras? I'm just trusting what the camera sees because the diffraction grating is so sensitive to slight movements. That is - how do I know the spectrum isn't shifted a bit from one camera to another if I try to line up the spectrums in post production?
Would I be better off with a prism and either a strobe or a constant light source? This way I could project it on to a surface - keeping those constant and putting the camera wherever makes sense for the given lens (I'm dealing with different lens mounts). Like - even if it's just a sheet of paper, the edges of the paper can be lined up in post production so I know that I'm lining up the "rainbow" the same in each file.
This way it's easier to line up the images in post and get a more apples-to-apples comparison.
I can play with the diffraction grating for projection once I get it.
My plan was to project the strobe through a prism or whatever onto a grey card. The grey card has markings on it.
I can use a constant light source, but I think once I get the initial setup, I can just use the grey card to set exposure in camera. I find this particular grey card is a bit above 18% middle grey, but very neutral and I trust it for setting white balance in camera and in post.
Yes - I expect that different lenses will have different characteristics and I won't be able to completely subtract the effects of the lenses on the image. Sigma's DP (dp) series for instance, it's impossible to change the lens.
Likewise, a halogen lamp has more red and less blue than daylight, but the blue and red wavelengths don't change. The very spiky spectra of fluorescent and arc lamps would make the results difficult to be sure of, however.
saltydogstudios
Senior Member
Thanks. I think the other poster may have been talking about IR/UV sensitivity actually - not entirely sure.A lens can't change the wavelength of light. It can change the intensity at any wavelength, but that is irrelevant when you are looking at the accuracy of the hues.Interesting.I wouldn't use a strobe. It would take too many images to line up everything. A constant light source and live view on the camera will save you a lot of time.Interesting.
I also have a full spectrum Olympus E-PM1.
So here's what I'm thinking. I have a bunch of cameras I want to compare. If I use diffraction grating, how can I accurately line up the frequency ranges across cameras? I'm just trusting what the camera sees because the diffraction grating is so sensitive to slight movements. That is - how do I know the spectrum isn't shifted a bit from one camera to another if I try to line up the spectrums in post production?
Would I be better off with a prism and either a strobe or a constant light source? This way I could project it on to a surface - keeping those constant and putting the camera wherever makes sense for the given lens (I'm dealing with different lens mounts). Like - even if it's just a sheet of paper, the edges of the paper can be lined up in post production so I know that I'm lining up the "rainbow" the same in each file.
This way it's easier to line up the images in post and get a more apples-to-apples comparison.
I can play with the diffraction grating for projection once I get it.
My plan was to project the strobe through a prism or whatever onto a grey card. The grey card has markings on it.
I can use a constant light source, but I think once I get the initial setup, I can just use the grey card to set exposure in camera. I find this particular grey card is a bit above 18% middle grey, but very neutral and I trust it for setting white balance in camera and in post.
Yes - I expect that different lenses will have different characteristics and I won't be able to completely subtract the effects of the lenses on the image. Sigma's DP (dp) series for instance, it's impossible to change the lens.
Likewise, a halogen lamp has more red and less blue than daylight, but the blue and red wavelengths don't change. The very spiky spectra of fluorescent and arc lamps would make the results difficult to be sure of, however.
At any rate, people will use any excuse to dismiss a "study" if it doesn't match their pre-conceived notion of what a sensor is and I was just hoping to avoid that.
--
"Wait let me comb my hair and put on a tie."
It Gets Nerdy: https://medium.com/ice-cream-geometry
Sometimes I take photos: https://www.instagram.com/sodiumstudio/
whisky
Senior Member
Hi, Was just browsing the Forum after a lapse of several months, not much of interest to me nowadays as I am back in the SD15 era when I noted your post on 'Spectral Lines'.
Perhaps this image, created by my Phd brother in law, who was a laser scientist and instrumental in reducing the bulk of high power lasers for use in cutting and other applications might be of interest to you.
He is retired now and just retains his interest in matters which are well above my understanding. I believe he created this image using old set of binoculars he had kicking around. Hopefully you will be able to see the many spectral lines present.
spectral lines.
I can view them clearly on my monitor so hopefully some of you can as well.
Cheers.
Perhaps this image, created by my Phd brother in law, who was a laser scientist and instrumental in reducing the bulk of high power lasers for use in cutting and other applications might be of interest to you.
He is retired now and just retains his interest in matters which are well above my understanding. I believe he created this image using old set of binoculars he had kicking around. Hopefully you will be able to see the many spectral lines present.
spectral lines.
I can view them clearly on my monitor so hopefully some of you can as well.
Cheers.
I can see 'em, Jimmy. ;-)Hi, Was just browsing the Forum after a lapse of several months, not much of interest to me nowadays as I am back in the SD15 era when I noted your post on 'Spectral Lines'.
Perhaps this image, created by my Phd brother in law, who was a laser scientist and instrumental in reducing the bulk of high power lasers for use in cutting and other applications might be of interest to you.
He is retired now and just retains his interest in matters which are well above my understanding. I believe he created this image using old set of binoculars he had kicking around. Hopefully you will be able to see the many spectral lines present.
spectral lines.
I can view them clearly on my monitor so hopefully some of you can as well.
Cheers.
They look like absorption lines:
Loading…
astronomy.swin.edu.au
--
Ted
That's very nice. Plenty of absorption lines.Hi, Was just browsing the Forum after a lapse of several months, not much of interest to me nowadays as I am back in the SD15 era when I noted your post on 'Spectral Lines'.
Perhaps this image, created by my Phd brother in law, who was a laser scientist and instrumental in reducing the bulk of high power lasers for use in cutting and other applications might be of interest to you.
He is retired now and just retains his interest in matters which are well above my understanding. I believe he created this image using old set of binoculars he had kicking around. Hopefully you will be able to see the many spectral lines present.
spectral lines.
I can view them clearly on my monitor so hopefully some of you can as well.
Cheers.
It would be interesting to know exactly how he made the image. Presumably the spectrum was made with a diffraction grating, but how did he get a smooth photo of it ? Not with a Bayer sensor, I bet.
dbateman
Senior Member
whisky
Senior Member
I'm not entirely sure how he did it but it looks like he took the pic with a Canon. 
Will ask him next time I see him.
Cheers.
Will ask him next time I see him.
Cheers.
Keyboard shortcuts
- f
- Forum