Comparison of the color rendering of two Canon DSLRs (500 D and 6 D)

[Kai]

Today I would like to report something about UV sensitivity and the color differentiation of my camera bodies.

First to the cameras:
It is a Canon EOS 500 D (in Japan EOS Kiss X3, in North America EOS Rebel T1i), an APS-C camera with approx. 15 MP, and a Canon 6 D, a full-frame Camera with approx. 20 MP.

I de-filtered the 500 D according to the instructions from Gary Honis (http://dslrmodifications.com/rebelmod450d1.html). As a replacement for the removed filter, I used a quartz cover glass from microscopy (cut to size).

The 6D I had de-filtered professionally (http://www.astro-modifikationen.de). The filter structure of the 6 D is more complex. Only the top two (dust vibrator and original white balance filter) of the three filters were removed. One protective glass remains on the sensor.
 

Unfortunately I have neither a calibrated spectrometer nor a suitable monochromator for measurements. Of course I am interested in individual measurement curves for filters and lenses as well as sensitivity curves for sensors, but – on the other hand - for photography only the interaction of all parts is relevant to me.

This is why I have chosen the following method for my comparisons, which comes very close to the real situation when photographing in sunlight:

I used the midday sun as a light source. The light falls through a slit onto a grating (made from a DVD). I then focused the spectrum generated in this way through my UV filter (here: Baader-U) and the lens (50 mm magnifying lens, mounted on a suitable helicoid) on the camera sensor.

The spectra can be seen twice each, above with the camera-internal white balance to sunlight, below with white balance from reflected sunlight to matt PTFE.

So I can see which parts of the sunlight are used in which intensity for the picture (with a white motif) and which wavelengths correspond to which false colors. The Fraunhofer lines disturb the overall impression a little, but also allow an internal wavelength calibration.

Canon EOS  6 D-FS

 

Canon EOS 500 D-FS

Conclusions

The sensitivity of the EOS 6 D-FS does not reach as deep into the UV range as the 500 D-FS. This is probably not due to the sensor, but to the remaining filter glass.

The color differentiation is stronger with the 500 D-FS.

The longer wave range is shown more violet with the 500 D-FS and more bluish with the 6 D-FS. The shorter-wave area is shown greener on the 500 D-FS than on the 6 D-FS.

Overall, with this method I can clearly see which filter-lens-camera combination can display which UV area.
 

In further posts I can show lens comparisons and comparisons with other light sources.

[Stefano]

It is always nice to see false green below about 350 nm. Also notice how the lavender in the 500 D is redder than in the 6 D. I think, although this is controversial, that redder lavenders often indicate a deeper UV reach. But don’t assume that this is always true.

[StephanN]

This is interesting. I've just received my 5DSR full-spectrum back from Makario, and hope to do some similar comparisons with my 6D-FS soon.

 

Are you using the WB-JPEGs from the camera or did you manage to colour-correct the raw-files later in Lightroom, Photoshop, etc?

[Kai]

@Stefano
Doesn't that have to be the case? With white balance on PTFE, the sum of all colors is achromatic ("gray"). With the deep UV comes (more) green. Accordingly, the near UV must be more purple/red to compensate.
 

@StephanN
I developed the raw images (CR2) in dxo PhotoLab without any further filtering or processing.
I'm looking forward to your results, especially regarding the UV depth.
How many filters were removed from your camera(s)?

[Stefano]

1 hour ago, Kai said:

@Stefano
Doesn't that have to be the case? With white balance on PTFE, the sum of all colors is achromatic ("gray"). With the deep UV comes (more) green. Accordingly, the near UV must be more purple/red to compensate.

Yes, that’s correct. More green means more magenta.

[Andrea B.]

Good reporting, Kai. We can make good use of information like this.

 

I did not realize a DVD could be used in this way!

[JMC]

Yes, I strongly suspect this observed difference is driven by the material being used for the coverglass on the sensor.

 

I know that with my 5DSR that there was a big difference between the original coverglass and when I had it replaced with a fused silica one, as discussed in the thread here -

 

[diant]

22 hours ago, Kai said:

Today I would like to report something about UV sensitivity and the color differentiation of my camera bodies.

Kai, your scale in "nm" is little wrong. You may check it by an astronomical solar spectra in UV (or by catalogues of solar lines). I marked some lines of UV solar specrtum here.

[StephanN]

22 hours ago, Kai said:

@StephanN

How many filters were removed from your camera(s)?

 

Well, if only I knew. Judging from the photos which I've taken so far (no real comparison yet, just looking at old photos), I would say that the reach of my 6D is about the same as yours, as I haven't been able to see deep green, just the green-yellow tint.

 

If I find the same with the 5DSR (and tbh, I fully expect this, like Jonathan in the thread he referred to above), then I'll have to contact them to see whether they are able and willing to remove more glass. At the moment I don't have the need for UV-B or -C, but who knows what the future might bring.

[Kai]

7 hours ago, diant said:

Kai, your scale in "nm" is little wrong. You may check it by an astronomical solar spectra in UV (or by catalogues of solar lines). I marked some lines of UV solar specrtum here.

Thank you for yor data. I orientated myself on a multi-calibrated spectrum of a mirror spectrometer [1]. But I didn't do it myself!

In comparison, your spectrum appears a bit distorted. That could of course come through the lens.

 

 

How did you assign the lines? I didn't succeed when I tried to use the tables of the NBS (now NIST) as a reference [2]. Although I went by the line width, there were just too many iron lines in the UV-A...  

 

 

[1] Tom Schnee, Johannes Felix Schnepp
https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwjQ3rGpvLHzAhVmg_0HHaTZC60QFnoECAMQAQ&url=https%3A%2F%2Fstudylibde.com%2Fdoc%2F16848367%2Fdas-sonnenspektrum---sch%C3%BClerlabor-astronomie&usg=AOvVaw3ASFW3J9yISiP7-ttFl5rF

 

[2] THE SOLAR SPECTRUM 2935Å to 8770Å - Second Revision of Rowland's Preliminary Table of Solar Spectrum Wavelengths

CHARLOTTE E. MOORE, M. G. J. MINNAERT, J. HOUTGAST, National Bureau of Standards Monograph 61, 1966

https://digital.library.unt.edu/ark:/67531/metadc13249/

[Kai]

2 hours ago, StephanN said:

 

Well, if only I knew. Judging from the photos which I've taken so far (no real comparison yet, just looking at old photos), I would say that the reach of my 6D is about the same as yours, as I haven't been able to see deep green, just the green-yellow tint.

 

If I find the same with the 5DSR (and tbh, I fully expect this, like Jonathan in the thread he referred to above), then I'll have to contact them to see whether they are able and willing to remove more glass. At the moment I don't have the need for UV-B or -C, but who knows what the future might bring.

You´re right.
For examinations, the greater UV depth of the 500D-FS is interesting, for most of the photos that I take (and will occasionally show), the spectral UV sensitivity of the 6D-FS is currently sufficient :)

[diant]

1 hour ago, Kai said:

How did you assign the lines? I didn't succeed when I tried to use the tables of the NBS (now NIST) as a reference [2].

Your second link in good enough. You should focusing your attention on "equivalent wide" column. You are right - Fe lines often dominate in many UVA regions.

But (for example) look at 372nm strong line in your spectrum (it is Fe I line). Look at the page 86 of your second link catalogue, second column, 8th line from the bottom:

3719,947 Angstrom - equivalent wide 1664. Huge value! That is our 372nm line :)

 

Or the nice doublet between 370 and 371nm.

See page 85:

- 3705,577 (eq wide=562 with 3706,037 at its right wing) - it is a left line of this doublet

- 3709,256 (eq wide=573) - it is a right line of this doublet

 

etc. etc.

[Kai]

On 10/4/2021 at 4:00 AM, Andrea B. said:

Good reporting, Kai. We can make good use of information like this.

 

I did not realize a DVD could be used in this way!

Thank you, Andrea :)
If you separate the two layers of a blanc burnable DVD and rinse off the dye with ethanol, you get useful gratings for many beautiful experiments. This has been very popular in scientific education for many years.
In the picture you can see my Canon 500 D-FS with an intermediate ring, helicoid and magnification lens, a bundle of adapter rings, the rotating mount of an old linear polarizer and the grating adapted to it (adapter via 3-D printer).
Since the grating lines are curved, the spectra are also curved. A narrow section is usually sufficient. Because of the 1350 L / mm, the resolution is considerable in all cases.

 

 

As an illustration, I will soon post the spectrum of an electrical spark between a titanium and a molybdenum electrode compared to sunlight and the light from a halogen lamp. 

[Kai]

20 hours ago, diant said:

Your second link in good enough. You should focusing your attention on "equivalent wide" column. You are right - Fe lines often dominate in many UVA regions.

But (for example) look at 372nm strong line in your spectrum (it is Fe I line). Look at the page 86 of your second link catalogue, second column, 8th line from the bottom:

3719,947 Angstrom - equivalent wide 1664. Huge value! That is our 372nm line :)

 

Or the nice doublet between 370 and 371nm.

See page 85:

- 3705,577 (eq wide=562 with 3706,037 at its right wing) - it is a left line of this doublet

- 3709,256 (eq wide=573) - it is a right line of this doublet

 

etc. etc.

Yes - you are right. Than

 

21 hours ago, diant said:

Your second link in good enough. You should focusing your attention on "equivalent wide" column. You are right - Fe lines often dominate in many UVA regions.

But (for example) look at 372nm strong line in your spectrum (it is Fe I line). Look at the page 86 of your second link catalogue, second column, 8th line from the bottom:

3719,947 Angstrom - equivalent wide 1664. Huge value! That is our 372nm line :)

 

Or the nice doublet between 370 and 371nm.

See page 85:

- 3705,577 (eq wide=562 with 3706,037 at its right wing) - it is a left line of this doublet

- 3709,256 (eq wide=573) - it is a right line of this doublet

 

etc. etc.

 

Yes, thank you very much, you were right.
Now I have assigned a lot of strong lines and the result corresponds exactly to your correlation.
Here is the result:

 

The white lines on top represents the values from literature. The curve is the luminance curve of the picture. Direct sunlight, slit, grating, lens (f 75mm), ITOS DUG11X filter, EOS 500 D-FS.
 

[Kai]

On 10/3/2021 at 3:25 PM, Kai said:

Today I would like to report something about UV sensitivity and the color differentiation of my camera bodies.

First to the cameras:
It is a Canon EOS 500 D (in Japan EOS Kiss X3, in North America EOS Rebel T1i), an APS-C camera with approx. 15 MP, and a Canon 6 D, a full-frame Camera with approx. 20 MP.

I de-filtered the 500 D according to the instructions from Gary Honis (http://dslrmodifications.com/rebelmod450d1.html). As a replacement for the removed filter, I used a quartz cover glass from microscopy (cut to size).

The 6D I had de-filtered professionally (http://www.astro-modifikationen.de). The filter structure of the 6 D is more complex. Only the top two (dust vibrator and original white balance filter) of the three filters were removed. One protective glass remains on the sensor.
 

Unfortunately I have neither a calibrated spectrometer nor a suitable monochromator for measurements. Of course I am interested in individual measurement curves for filters and lenses as well as sensitivity curves for sensors, but – on the other hand - for photography only the interaction of all parts is relevant to me.

This is why I have chosen the following method for my comparisons, which comes very close to the real situation when photographing in sunlight:

I used the midday sun as a light source. The light falls through a slit onto a grating (made from a DVD). I then focused the spectrum generated in this way through my UV filter (here: Baader-U) and the lens (50 mm magnifying lens, mounted on a suitable helicoid) on the camera sensor.

The spectra can be seen twice each, above with the camera-internal white balance to sunlight, below with white balance from reflected sunlight to matt PTFE.

So I can see which parts of the sunlight are used in which intensity for the picture (with a white motif) and which wavelengths correspond to which false colors. The Fraunhofer lines disturb the overall impression a little, but also allow an internal wavelength calibration.

Canon EOS  6 D-FS

 

Canon EOS 500 D-FS

Conclusions

The sensitivity of the EOS 6 D-FS does not reach as deep into the UV range as the 500 D-FS. This is probably not due to the sensor, but to the remaining filter glass.

The color differentiation is stronger with the 500 D-FS.

The longer wave range is shown more violet with the 500 D-FS and more bluish with the 6 D-FS. The shorter-wave area is shown greener on the 500 D-FS than on the 6 D-FS.

Overall, with this method I can clearly see which filter-lens-camera combination can display which UV area.
 

In further posts I can show lens comparisons and comparisons with other light sources.

 

After diant's hint, I recalibrated the wavelength scales. Fits now :) 

 

6D-FS

500D-FS

 

[Nate]

I picked up some diffraction gradings and figured I'd test my setup. 

Canon 77D, EL Nikkor 80 old metal, and the Baader.  My lines match pretty much, but guessing my cover glass is stopping me from having anything useful under 350nm.

Would a different sensor glass make that much of a difference?

Kinda  displeased I was trying to go deep as possible with different filters, but had this limiting factor. I might contact the guy who modded my cams to see if he's worked with UV mods to put better UV pass glass in. 

[Andrea B.]

There is not much UV light around this time of year. Try this again in strong summer sunlight?

 

Also the BaaderU peaks around 350 nm and tapers off on both shoulders. That also truncates what little you can get on the left side of 350 nm. Similarly the EL 80/5.6 slopes down on the left of 350 and will add to the truncation.

 

Who converted your camera?

 

If you wanted to make a better test of the capabilities below 350 nm for this gear, then perhaps use a narrowband filter or one which peaks under 350 nm? 

 

 

[Nate]

8 minutes ago, Andrea B. said:

There is not much UV light around this time of year

I never thought of that, still is kinda low in the sky, will try in later months.

I did try the Antilia, as well as all filter combinations, but the left side never changed. Here's the Antilia done a different way, I used my door, just cracked a bit to the outside, it squeezed the lined more but still never got lower. That filter I believe goes bellow 300nm.

[ulf]

Depending on the type of grating there can be losses there too.

A long time ago I measured the transmission of a cheap transmission grating using a good IV-VIS spectrometer:

In combination with some filters like the Baader U there is not much light passing below 320nm to be used for investigating a cameras UV reach

[Nate]

8 hours ago, Andrea B. said:

Who converted your camera?

I had Isaac Szabo https://www.isaacszabo.com/ convert my two cameras to full spectrum before I got into UV and was only interested in NIR. Great speedy work.

 

@ulfThat is the grating I bought, pack of around 50. Your test looks like it would show the uv green on a nonWB shot. I even tried the Sun 30deg to the right, but only get yellow and no green. Here's that test   

 

[ulf]

AFAIK all these tests are WB. 

The different UV pass filter transmissions must be compensated for.

 

The false colours differs between RGB dye formulations in different camera brands.

The coating differences on the sensor cover glass is definitely also a limiting factor, just as Jonathan stated above.

[Nate]

Here's my results for the Baader, Nikkor 80 old

 

 

It is two different shots, but I was under the impression that daylight or auto WB still will show green below 350nm?

[Stefano]

Usually you have green below ~350 nm with a common UV white balance. At least, this is what I have seen with my camera and in the forum. All those grating/sparticle tests show some green if the lens reaches deep enough.

[ulf]

But no WB will not work at all as the filter and bayer matrix response will be very different from a normal VIS WB. It has to be done against something with a neutral UV reflectance like PTFE

[Nate]

@ulf I've rigged up my asi178mc and my asi462 to my Nikkor 80 and achieved focus. I just need the sun to come up for testing. the 178 has no cover glass as I broke it removing the uv/ir cut filter so going to try that one.

 

The 462 has this

 

So what I'm hoping, is that neither show anything deeper than my 77D, which they probably will, and might need advice on what to do if I want bellow 350nm with the 77D.

Thanks.