Jump to content
UltravioletPhotography

Sigma SD14 Channel Response under Various Filter Combos (Done!)


Andrea B.

Recommended Posts

Editor's Introduction: Our member David Bateman (dabateman) made sets of Sigma SD14 photos under various filter combinations in order to investigate the possibility of separating UV reflective data recorded in the Foveon blue layer and UV induced IR fluorescence recorded in the Foveon red layer. Additionally, David wanted to look at how UV, Visible and IR light are recorded by the Foveon.

 

I'm providing the raw composites and raw histograms for the photos since David doesn't have Raw Digger (yet). I'm going to put each filter set in a separate post. Then David can add his explanations and comments.

 

Please be patient while we get the photos and comments in place. :D There will probably be some edits or rearrangements along the way.

I'll remove the In Progress flag from the title when David declares the topic to be properly posted.

 

A big Thank You to Dabateman for giving me the opportunity to play with some SD14 raw files!!!


 

Gear: Sigma SD14 + Asahi Ultra-Achromatic-Takumar + Convoy 365nm UV-Led (filtered)

The Convoy was filtered with U340 (2mm).

 

Exposure: f/8 @ISO-100 for various times.

 

Filters:

  • U330 (1.5 mm): UV + IR, dual bandpass
  • S8612 (2 mm): UV + Vis, blue-green IR-blocker
  • IR720: IR longpass (small amount of Red)
  • 2A: Tiffen Haze, blocks some UV

.

Software: The raw composites and raw histograms were created with Raw Digger. RD settings included Gamma 2.2 and Autoscale so that the demosaiced images are visible. No camera color profile was used. The histograms use EV along the x-axis. EV0 is based on the SD14's data range so that histograms are easier to compare. Data along the y-axis is linear (actual count, binned at 1/64 interval).

Images were cropped and resized in Photo Mechanic and labeled in Photoshop Elements 11.

 

Purpose of Showing Adjusted Raw Composite: In low light there is often not much to see in a raw composite even after application of the usual contrast adjustments following demosaicing. So I have supplied a "pushed" raw composite which better shows the raw recording in its non-white-balanced colours. (Of course, an adjusted raw comp should be be taken as accurate.)

 

 

POSTS

  • Control Set: UV Reflection + IR Emission
    I think that this set of SD14 photos definitely illustrates the possibilities of Foveon separation of wavelengths well. UV is very strong in the Blue layer as evidenced by the floral UV-signature recorded there. Blue-cyan fluorescent fibers show up in the Green channel. And IR fluorescence of the flower petals is recorded in the Red channel. Very nice!
    LOOK FOR: the adjusted version following each monochrome raw channel photo.

.

Link to comment

CONTROL SET: UV Reflection + IR Emission

 

DISCUSSION:

 

Under UV illumination from the 365nm UV-Led, this U330 filter should capture both the UV reflected off the subject and any IR emitted from the subject. This photo set acts as a control against which the primary fluorescence sets may be compared. See Infrared Fluorescence, Visible Fluorescence.

 

We would expect the Foveon sensor to record the relatively shorter reflected UV light on the top blue layer and the relatively longer emitted IR light on the bottom red layer. The histograms support this. You can see a broad range of reflected UV in the blue and a narrow range of emitted IR in the red. That fits what we know about reflection and fluorescence (typically narrower, peaked).

 

Raw channel extractions in monochrome are also shown as further support for the separation of the reflected UV and emitted IR. The raw Red channel doesn't show much unless pushed because the count and breadth of the red is low compared to the blue. Similarly for the Green channel.

 

The flower shows a slightly UV-absorbing center. The emitted IR has washed it out a bit in the 3-channel photo. The next reflected UV photo set better shows the UV-signature of this flower without resorting to channel extractions.

 

Now, should there be some evidence of the visibly fluorescent blue-cyan fibers and dust on the flower which is so easily seen in later photo sets? Maybe, but I'm thinking that the reflected UV and emitted IR has possibly overwhelmed that. (I don't have a Sigma converter in order to try to better bring out possible blue-cyan fluor.) Also, the U330 visible transmission drops off pretty quickly, so catching cyan emissions here wouldn't be very easy.

 

ADDED A WHILE LATER: Whee !! When I extracted the R, G and B channels in Raw Digger and applied a bit of a push to bring out what was recorded in the G channel, those fluorescent fibers popped out. Cool beans !!!

 

Three sets were made at exposure lengths of 2", 4" and 30".

The 30" set is shown here.

 

 

Filtered UV/Vis/IR

All photographs: Copyright 2018 by David Bateman (dabateman).

 

Filter: U330 only

 

As shot

sd14_uat_u330x150_uvLed365fltr_f8_30sec_iso100_20180716.jpg

 

Raw composite

sd14_uat_u330x150_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0101.jpg

 

Adjusted raw composite

sd14_uat_u330x150_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0201.jpg

 

Raw histogram

sd14_uat_u330x150_uvLed365fltr_f8_30sec_iso100_20180716rawHistoLx.jpg

 

 


 

Blue channel (top)

Obviously a UV-signature.

sd14_uat_u330x150_uvLed365fltr_f8_04sec_iso100_20180716blueChan01.jpg

 

Blue channel adjusted

sd14_uat_u330x150_uvLed365fltr_f8_04sec_iso100_20180716blueChanpnAdjHHH.jpg

 

 

Green channel (middle)

sd14_uat_u330x150_uvLed365fltr_f8_04sec_iso100_20180716greenChan01.jpg

 

Green channel adjusted

Exactly what we wanted to see! Just took a little push to bring those fibers out.

sd14_uat_u330x150_uvLed365fltr_f8_04sec_iso100_20180716greenChanpnAdjHHH.jpg

 

Green channel adjusted and sharpened

A little sharpening may add noise but helps show the fibers.

sd14_uat_u330x150_uvLed365fltr_f8_04sec_iso100_20180716greenChanpnAdjHPO01jj.jpg

 

 

Red channel (bottom)

Definitely some Infrared fluorescence.

sd14_uat_u330x150_uvLed365fltr_f8_04sec_iso100_20180716redChan01.jpg

 

Red channel adjusted

sd14_uat_u330x150_uvLed365fltr_f8_04sec_iso100_20180716redChanpnAdjHHH.jpg

 


 

Added Note: Let me point out that adjusting these channel photos by bringing in the white/black points needs to be done carefully. I do this in Capture NX2 so that I can monitor how far I'm going and not over-do it. There is a little bit of art to this procedure. So, while we do see that there is real information in an otherwise grey blob of a composite frame or channel frame, we should try not to force the matter. (Did this make sense? I hope so.)

Link to comment

CONTROL SET: Reflected UV

 

 

DISCUSSION:

 

The U330 + S8612 filter stack captures any reflected UV or emitted blue or green visible light from the subject under UV illumination from the 365nm UV-Led. Any IR emissions are blocked. The UV-Led light is so strong that I think any visible blue-cyan fluorescence of the dust and fibers on the flower can't be recorded in a 2 second interval.

 

These photo set acts as a control against which the primary fluorescence sets may be compared. See Infrared Fluorescence, Visible Fluorescence.

 

We would expect the Foveon sensor to record the (relatively) shorter UV light on the top blue layer. The histograms support this. Look at the log versions to see the very large amount of blue. The red and green are narrow and spiked. (I'm not sure why. Specular?)

 

The flower is certainly showing a UV-signature in both photo sets. The center is UV-absorbing and the petals are moderately UV-reflecting. Both histograms are similar, so removal of the U330 has not changed much what was recorded except that you can see the fluorescent blue-cyan fibers in the second set.

 

 

Two sets were made at exposure lengths 2" and 4".

Only the 2" set is shown here.

 

 

 

Filtered UV/Vis

All photographs: Copyright 2018 by David Bateman (dabateman).

 

Filter Stack: U330 + S8612

 

As Shot

sd14_uat_u330x150_s8612x200_uvLed365fltr_f8_02sec_iso100_20180716y.jpg

 

Raw composite

sd14_uat_u330x150_s8612x200_uvLed365fltr_f8_02sec_iso100_20180716rawComppn0101y.jpg

 

Adjusted raw composite

sd14_uat_u330x150_s8612x200_uvLed365fltr_f8_02sec_iso100_20180716rawComppn0201y.jpg

 

Raw histogram: Linear

sd14_uat_u330x150_s8612x200_uvLed365fltr_f8_02sec_iso100_20180716rawHistoLy01.jpg

 

Raw Histogram: Log

sd14_uat_u330x150_s8612x200_uvLed365fltr_f8_02sec_iso100_20180716rawHistoLog01bb.jpg

 

 

 


All UV/Vis

Filter: S8612 only

 

As Shot

sd14_uat_s8612x200_uvLed365fltr_f8_02sec_iso100_20180716x.jpg

 

Raw composite

sd14_uat_s8612x200_uvLed365fltr_f8_02sec_iso100_20180716rawComppn0101x.jpg

 

Adjusted raw composite

sd14_uat_s8612x200_uvLed365fltr_f8_02sec_iso100_20180716rawComppn0201x.jpg

 

Raw Histogram: Linear

sd14_uat_s8612x200_uvLed365fltr_f8_02sec_iso100_20180716rawHistoLx01.jpg

 

Raw Histogram: Log

sd14_uat_s8612x200_uvLed365fltr_f8_02sec_iso100_20180716rawHistoLogbb01.jpg

 

.

Link to comment

INFRARED and VISIBLE FLUORESCENCE

 

 

DISCUSSION:

 

In theory, the filter stacks below are meant to capture any emitted visible and/or IR light under under 365 UV-Led illumination. When the U330 filter is used, the emitted visible light, if any, would only be recorded if it were blue or green. The 2A filter is meant to block reflected UV, but does permit some passage of UV above 350 nm.

 

The first photo set (with the U330) does shows suppression of the fluorescent fibers and dust. As mentioned before the U330 transmission of blue/green does drop off very quickly, so it is more difficult to catch any cyan-blue fluor. When the U330 is removed for the second set of photos, the visible fiber fluorescence is much stronger.

 

 

 

Filtered Vis/IR

All photographs: Copyright 2018 by David Bateman (dabateman).

 

Filter Stack: U330 + Tiffen Haze 2A

 

As shot

sd14_uat_u330x150_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716y.jpg

 

Raw composite

sd14_uat_u330x150_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0101y.jpg

 

Adjusted raw composite

sd14_uat_u330x150_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0201y.jpg

 

Raw histogram

sd14_uat_u330x150_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716rawHistoL01y.jpg

 

 

 


All Vis/IR

All photographs: Copyright 2018 by David Bateman (dabateman).

 

Filter: Tiffen Haze 2A only

 

As shot

sd14_uat_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716x.jpg

 

Raw composite

sd14_uat_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0101x.jpg

 

Adjusted raw composite

sd14_uat_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0201x.jpg

 

Raw histogram

sd14_uat_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716rawHistoL01x.jpg

 

 

.

Link to comment

INFRARED FLUORESCENCE

 

 

DISCUSSION:

 

Both filter stacks (below) capture any infrared light emitted by the subject under 365 UV-Led illumination. There should be no transmission of UV.

 

The flower is clearly looking IR-ish in the photo sets. We would expect the Foveon sensor to record the emitted infrared light mostly on the red layer, I think. The histograms do support this. (With a Bayer sensor, there is some capture of Infrared light in the blue channel when using longpass filters above 850 nm.)

 

Removal of the U330 in the second set increases the brightness of the photo (spikes move to the right), but does not affect the distribution much.

 

Looking at the logrithmic versions of the charts shows that the red layer count is more broadly distributed than the blue or green. The log charts make it easier to see that more of the IR fluorescence is recorded on the red layer.

 

QUESTION: Why the green and blue?

 

 

 

Filtered IR

All photographs: Copyright 2018 by David Bateman (dabateman).

 

Filter Stack: U330 + IR720

 

Two sets were made at exposure lengths 15" and 30".

The 30" set is shown here.

 

As shot

sd14_uat_u330x150_ir720_uvLed365fltr_f8_30sec_iso100_20180716y.jpg

 

Raw composite

sd14_uat_u330x150_ir720_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0101y.jpg

 

Adjusted raw composite

sd14_uat_u330x150_ir720_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0201y.jpg

 

Raw histogram: linear

sd14_uat_u330x150_ir720_uvLed365fltr_f8_30sec_iso100_20180716rawHistoLy01mm.jpg

 

Raw histogram: logarithmic

sd14_uat_u330x150_ir720_uvLed365fltr_f8_30sec_iso100_20180716rawHistoLog01.jpg

 

 


All IR

All photographs: Copyright 2018 by David Bateman (dabateman).

 

Filter: IR720 only

 

As shot

sd14_uat_ir720_uvLed365fltr_f8_30sec_iso100_20180716.jpg

 

Raw composite

sd14_uat_ir720_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0101.jpg

 

Adjusted raw composite

sd14_uat_ir720_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0201.jpg

 

Raw histogram: linear

sd14_uat_ir720_uvLed365fltr_f8_30sec_iso100_20180716rawHistoLx01mm.jpg

 

Raw histogram: logarithmic

sd14_uat_ir720_uvLed365fltr_f8_30sec_iso100_20180716rawHistoLog01.jpg

Link to comment

VISIBLE FLUORESCENCE

 

DISCUSSION:

 

Both filter stacks (below) capture any visible light emitted by the subject under 365 UV-Led illumination. Any IR fluorescence is blocked by the S8612. The 2A filter is meant to block reflected UV, but does permit some passage of UV above 350 nm. Any recorded UV in these photo sets would be in the blue layer. The U330 filter alters the transmission ratio of blue and green, so we would expects its removal in the 2nd set to change the B/G ratio. And that does happen.

 

We would expect the Foveon sensor to record the visible light on any or all of the three layers (red, green and blue) depending, of course, on the colour of the fluorescence. This particular flower seems to show no visible fluorescence in its petals. However, we all know that the dust and fibers in the photo should fluoresce in a cyan-blue color. And that is what we see.

 

Compared to the large amount of non-fluorescent area in the photos, the visible fluorescence appears almost specular. And indeed, when you look at the raw histograms, the R, G and B counts are quite tall and very narrow ("spiked"). We have always seen that fibers and dust fluoresce as bright blue or blue-cyan. That is somewhat evident in the linear histograms, but more obvious in the logarithmic (y-axis) histogram which shortens the spike height while lengthening the height of the low-count bins so they can be more easily seen.

 

QUESTION: So, where does the red come from?

 

 

 

Filtered Vis

All photographs: Copyright 2018 by David Bateman (dabateman).

 

Filter Stack: U330 + S8612 + Tiffen Haze 2A

 

As Shot

sd14_uat_u330x150_s8612x200_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716y.jpg

 

Raw composite

sd14_uat_u330x150_s8612x200_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0101y.jpg

 

Adjusted raw composite

sd14_uat_u330x150_s8612x200_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0201y.jpg

 

Raw histogram: linear

sd14_uat_u330x150_s8612x200_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716rawHistoL01y.jpg

 

Raw histogram: logarithmic

sd14_uat_u330x150_s8612x200_2a350lp_uvLed365fltr_f8_30sec_iso100_20180716rawHistoLog01.jpg

 

 

 


All Vis

All photographs: Copyright 2018 by David Bateman (dabateman).

 

Filter Stack: S8612 + Tiffen Haze 2A

 

As Shot

sd14_uat_2a350lp_s8612x200_uvLed365fltr_f8_30sec_iso100_20180716x.jpg

 

Raw composite

sd14_uat_2a350lp_s8612x200_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0101x.jpg

 

Adjusted raw composite

sd14_uat_2a350lp_s8612x200_uvLed365fltr_f8_30sec_iso100_20180716rawComppn0201x.jpg

 

Raw histogram

sd14_uat_2a350lp_s8612x200_uvLed365fltr_f8_30sec_iso100_20180716rawHistoL01x.jpg

 

Raw histogram: logarithmic

sd14_uat_2a350lp_s8612x200_uvLed365fltr_f8_30sec_iso100_20180716rawHisto01Log.jpg

 

 

.

Link to comment
I keep getting interrupted locally!! So I keep having to re-do sentences and discussion points.
Link to comment

OK I think I finally have the labels and discussion correctly set.

 

ToDo:

  • upload the rest of the photos
  • examine results -- conclusions?

Tentatively, I do see the expected results. And I'll explain why. But we lack a definite explanation for the spikes on some layers. Stray light? Leaky Foveon layers? How is the Foveon constructed? Does it use any dyes? Our illuminants, in spite of filtration, are not always pure. Our filters are also not always pure - especially when exposure lengths reach 30" or more. And so on.

Link to comment

Thank you Andrea for all your work.

The Foveon sensor does not use any dyes. The colours are detected based on depth penetration through silicon. But what I think the data indicates is that as light passes through a layer it is also detected. I wonder if the Sigma software uses a correction to filter that out based on duplication of detection. Something I will have to play with.

Link to comment

Using Image J with the DCRaw plugin, I am able to open the Sigma files up to the SD14 camera. There a bunch of settings you can set. I will show below two sets. The main settings selected are White Balance:None, Output color space: sRGB, Document mode (no color, no interpolation, Document mode without scaling (totally raw), Read as 8-bit, Do not rotate or scale pixels (preseve orientation and aspect ratio).

 

Then I either set Do Not automatically brighten image on or off.

 

Here is the Raw not brightened crop view of the flower using the 720nm only filter:

 

Blue Channel

post-188-0-33506400-1540103728.jpg

Green Channel

post-188-0-85793400-1540103747.jpg

Red Channel

post-188-0-79041700-1540103764.jpg

 

Here is the Raw Automatically brightened crop view of the flower using the 720nm only filter:

 

Blue Channel

post-188-0-15153500-1540103782.jpg

Green Channel

post-188-0-62632400-1540103813.jpg

Red Channel

post-188-0-11290400-1540103829.jpg

 

Here is the Raw not brightened crop view of the flower using the 1.5mm U330 filter:

Blue Channel

post-188-0-09587200-1540103861.jpg

Green Channel

post-188-0-69263300-1540103902.jpg

Red Channel

post-188-0-24283100-1540103915.jpg

 

Here is the Raw Automatically brightened crop view of the flower using the 1.5 mm U330 filter:

 

Blue Channel

post-188-0-36458800-1540103940.jpg

Green Channel

post-188-0-52994700-1540103960.jpg

Red Channel

post-188-0-33123400-1540103973.jpg

 

I have no idea how DCRaw automatically brightens the channels, or if the same setting is applied to each channel or if each is boosted as best as possible. When I tried to Brighten in Image J it looks very posterized.

Interesting the differences between Raw Digger and Image J. I think you can manipulate the output here without auto brightening to get at what I think you could use the Sigma Camera for. This may be because the SD14 takes Dark images in general. I think the off set is -2.4 EV, if I am remembering correctly.

 

You could isolate the not auto brightened Blue channel and the auto brightened Green and Red channels to see the fluorescence. The Dust is clearly fluorescing in the Green visible channel only and the flower is clearly fluorescing in the Red channel and maybe Green channel, but that could be IR leak. Pulling just the IR fluorescing red and reflected Blue images seems possible.

Link to comment

One more correction, my 2A filter is a Kodak Watten 2 gel that blocks all light below 400nm. It only allows 10% transmission at 410nm.

 

Also interesting observation. How many people use a phone to access this site?

I am about 95% on my phone. Looking above with my phone I can't see the auto brightened Green channel image from the u330 filter. But on my computer its very bright and the dust is clearly visible. I may just have to jack the brightness on my images if people are using mobile devices more.

Link to comment

I finally finished posting everything. Everyone please let me know of typos and errors. (I got kind of tired!)

 

Do check out Post #2. A very cool example of how the Foveon sensor does separate the UV, Visible and IR wavelengths.

Link to comment
With a Bayer sensor, there is some capture of Infrared light in the blue channel when using longpass filters above 850 nm

 

Actually IR falls into almost equally into the R,G and B channels on a Bayer sensor because the Bayer dyes becomes almost transparent to IR at that wavelength and so basically very little attenuation occurs.

 

QUESTION: Why the green and blue?

 

The Foveon effectively uses 3 stacked photosensitive layers. The junction depth of each photosensitive layer governs which wavelengths of light are absorbed theoretically meaning that Blue should fall in the shallowest junction and red in the deepest junction. Your IR results concur with my IR findings on the Foveon sensor. IR is actually absorbed in all three junctions as well as in the substrate below the lowest junction. Since the IR light must pass through the blue and green junctions to reach the red junction some anomalous absorption is likely to occur however it seems a bit too much is occurring. Perhaps this is a result of scattering at the junction interfaces, junction leakage and thermally induced leakage due to substrate heating (absorption of IR in the substrate has nowhere to go).

Link to comment
QUESTION: So, where does the red come from?

 

Sigma firmware definitely seems to screw with the data prior to saving RAW data, just as Nikon performs pre-conditioning of the colour channels prior to WB application. In 2009 I only had my Sigma a short while to perform testing but it was behavior like this, the high noise in UV, absorption of IR in all 3 channels, clunky camera controls and menu, marginal robustness of the body, crappy OEM software etc that led me to being completely satisfied with a modified Nikon sensor.

 

I really hope all this has improved as the Foveon sensor really does have the potential to be an awesome sensor for UV and IR work.

Link to comment
An interesting test that will more clearly indicate what is occurring, would be to image monochromatic light at 5 given wavelengths for UV, B, G, R and IR. Use the 3 standard values for B, G and R that are commonly utilized for correcting APO lenses and select say 365nm for UV and 830nm for IR. The results of this test will produce a much clearer understanding of how light is interacting with the junction layers AND firmware, than tests using broadband filters and flowers.
Link to comment
The colours are detected based on depth penetration through silicon. But what I think the data indicates is that as light passes through a layer it is also detected.

If light passing into a junction is detected then it no longer exists in its current energy state to move on to the next layer.

Link to comment

If light passing into a junction is detected then it no longer exists in its current energy state to move on to the next layer.

 

Thats an excellent point. I do wonder why the leakage through layers. Since it seems to disapate, maybe its just heat we are seeing. The signal is more intense in the correct channel, than fades as further away. This is a problem for the green channel then, smack in the middle.

 

Shane what are your standard wavelengths. I could do 365nm, 405nm, I don't have a good 550nm source, but I might have a 632nm lazer pointer some where. I also have an IR flashlight that I can't remember the wavelength. But it may be in the 700s not 800s.

Link to comment

Do not forget that the Sigma SD14 cameras have a hot mirror that cuts in at 420nm. The image sensor has a glass cover of unknown transmittance. There are micro-lenses that also have an unknown transmittance. Then the first junction (blue) has a this layer of silicon to attenuate the UV & only allow the blue photons to enter. This all comes together to make the Sigma SD14 very unattractive for UV photography !

I have to mention the lens too has some limitation to UV as well.

Even removing the hot mirror & having a suitable lens, there is still the restriction on UV penetrating the cover glass, the micro-lenses & the top layer of silicon !

I am a big Sigma fan, but that is the limitations of it, not a go-to camera for UV ! Great for RGB & IR.

Cheers

Col

Link to comment

Shane: Perhaps this is a result of scattering at the junction interfaces, junction leakage and thermally induced leakage due to substrate heating (absorption of IR in the substrate has nowhere to go).

 

This certainly makes sense to me! Thanks for your various comments about how the Sigma Foveons work. :)

 

 

Shane: Sigma firmware definitely seems to screw with the data prior to saving RAW data

 

In principle I really do not like the RAW data to be screwed with!!! However, with the Foveon sensor perhaps it is necessary?

 

EDIT: Please disregard incorrect next sentence. My error!

Conditioning the raw data for WB application probably OK because we can still recover the raw in tools like Raw Digger.

 

 

Colinbm: Then the first junction (blue) has a this layer of silicon to attenuate the UV & only allow the blue photons to enter.

 

Oh my. That is not good news!

David did manage, it seems, to capture the UV signature of the flower. But we know that many floral signatures can be captured in the UV/violet under/around 400nm.

And this leads me once again to wanting some kind of reference standards which absorb UV/violet around 400 nm but which reflect around lower points. Like, a 350 nm peaked reflectivity would be nice for testing.

Alternately, high transmitting, narrowband filters would be super good to have. If only they did not have those mirrored surfaces.

Link to comment
Conditioning the raw data for WB application probably OK because we can still recover it in tools like Raw Digger.

No you cannot.

Preconditioning is applied before WB multipliers and what you see in Rawdigger is pre-WB but post pre-conditioning. In fact Rawdigger is useful for illustrating pre-conditioning because it can't present the RAW data without it.

Link to comment

No you cannot. Preconditioning is applied before WB multipliers and what you see in Rawdigger is pre-WB but post pre-conditioning.

 

Oops! I will go back and correct my error. Thanks!

Link to comment
Colinbm: Then the first junction (blue) has a this layer of silicon to attenuate the UV & only allow the blue photons to enter.

 

http://www.ultravioletphotography.com/content/index.php/topic/3015-sigma-camera-for-uv-reflectance-imaging/page__view__findpost__p__24237

 

A passivation layer (typically silicon dioxide, silicon oxynitride, silicon nitride etc) is applied over sensitive circuitry to protect the circuitry from environmental corrosion and offers minimal physical damage protection. These layers may or may not offer some form of UV protection (blocking) due to chemistry, however their physical thickness comes into play in the same way that the Foveon sensor works.....wavelength dependent absorption coefficient of silicon. The thicker the overlying passivation the more short wavelengths are absorbed.

Theoretically, there is no need to apply an overlying layer to specifically block UV on the Foveon sensor since the Blue channel photodiode has been designed at the appropriate thickness that most UV should never reach it.

 

In fact it would be hard to design a UV channel that would function well since it would literally have to be right at the surface of the sensor.

Link to comment
What does this preconditioning consist of, and how do we know it’s happening?

 

https://www.rawdigger.com/howtouse/rawdigger-histograms-overexposure-shapes

 

Several references in this article to preconditioning but here is a quote for one of them.

 

Quote

Cameras where color channel preconditioning is used are recognizable by the regular one-pixel gaps in the histograms of red and blue channels (sometimes, additionally, they have different clipping points for red, blue, and both green channels, as on fig.24)

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...