Aerochrome color shifts SOOC with Foveon

[Christoph]

This is another shot... in a cloudy setting... I like the purple tones in the transition area between grass and the "dead" field...

[Christoph]

This is with a polarizer (and Orange 16). I'm happy that it's not really changing the colors... It's a bit more saturated. Interesting thing, which shows why the film was so effective in exposing camouflaged positions in war: Look at picture 3. The balcony. The blue below are fake plants. In visible light they appear the same green like real ones above.
 

 

 

 

[colinbm]

Another good series Christoph.
Who would have known, that the fake plants turn blue...!

[Christoph]

Some more photos from earlier today. Very happy with them. The only thing that remains an enigma for me is the supposed color change red -> green.

 

Why don't red cars turn green? Why would only certain red lights appear green? I also did a little research online on how Aerochrome images red. You hardly ever see green, and at most sometimes in red traffic lights. But on other pictures with red traffic lights, they are yellow. Probably it depends on different things - especially the filter that is used.

 

Maybe someone can explain the theory behind it. Fedia? You seem to have done tons of research about this... :-)

 

I think most is very consistent with Aerochrome or Kodak EIR. The color tones, especially of the vegetation, depends a lot on light, where it comes from, how much, shade or not shade et cetera. 

 

But I have to stress - it's the best digital Aerochrome solution I have seen so far. And it's SOOC. For whatever reason. It's a mystery to me what the FOV Classic Blue actually changes. And why the hell did Sigma not market this as digital Aerochrome? It would have been easy to make a deal with Midwest Optics to actually include a TB filter with every camera...

 

Typical taillight coloration: 

 

 

Some greenish parts in the taillights. The yellow passenger bus turns white. Also note the variation in the plants and the colors of the hedge:

 

 

A lot of red here... shot with the sun behind me at late afternoon:

 

 

Some pictures from some days ago, same settings and setup:

 

Red car:

 

 

self-portrait - the typical yellowish lips:

 

 

Blue car will turn reddish violett:

 

[Christoph]

Some more from today, early evening, maybe 2 hours before sundown... with polarizer

 

artificial soccer turf:

 

 

 

 

 

the blueish barrels on the left are actually green, the reddish on the right is blue:

 

 

you can see the traffic lights in the distance... red is yellow, but not green:

 

[Fedia]

Wow, amazing new batch of pictures Christoph !

 

Concerning the the abscence of green objects in your photos, I don't think a defect in your technique is the cause. In the contrary, your process seems incredibly accurate and reveals what IRG really looks like better than the blurry, grainy aerochrome images we can find on the net.

 

To me the cause of the abscence of green objects is that it is unlikely to find objects that reflect red but not IR. In fact most objects reflect IR : that's why the green barrels are more lavender than blue, and that's why the astroturf is purple and not pure blue. The dyes we use in objects are made to block some bands of visible light not IR, so all dyed objects are susceptible to reflect IR.

 

The common point of yellow, red, purple, orange, violet (the colors that you get with your technique) is signal in the red/IR channel. Colors without signal in the red channel would be green, turquoise, cyan and acid blue.

Those anti-red colors you can also have them but only with object that absorb IR. And it doesn't look like there is a lot of these kinds of objects around, the more obvious ones are car windows, clearly absorbing IR. Then there is metal and concrete, which are more neutral. Then as you mentioned there are artificial lights that may or may not emit IR depending if they are LED, fluorescent, tungsten or gas-discharge. So if a red street light appears yellow it probably means that it is an older model that emits IR as well as red, and if it appears green it probably means that it uses LEDs and emits a monochromatic band of red.

 

You may have noticed how soil, tiles, straw and skin appear less red (or less "warm") than they should with IRG of even just with fullspectrum Foveons, giving this distinctive cyanish tone to brown/beiges colors ? I suppose this is because they are not dyed objects and they gradualy reflect less and less light after a peak around orange or red. In the contrary red or orange dyed objects are expected to reflect as much in IR as they do in visible.

 

To give an exemple of this with something we know well : filters; have you ever seen a red filter that cuts IR ? I haven't. There are green filters that cut IR : Hoya X1, b+w 061, blue filters as well : hoya 80 series, but starting from yellow, filters begin to behave like lee gels, passing all the longer wavelegnths beyond their cutoff point. I think it is the same thing that happens with red dyed objects.

 

Other exemple, the filters on a sensor : The blue dye has a peak, so does the green. But once the hot mirror is removed you realise that the red dye doesn't have a peak and is just a cutoff filter, the slow slope of the curve being only due to the deacreasing sensitivity of the silicon further in the IR. The graph also shows that if these particular dyes were used in paint, a blue object shot in IRG would show red and a green one would show blue-violet.

 

 

I don't think SIGMA knows about what you just found, they never had aerochrome in mind when developping their sensor. Even the normal fullspectrum look is not advertised. The only thing that seems to be worth advertising is definition...

 

Of Course I don't know what FOV classic blue does, I would like to. The foveon sensor has a lot of overlap beetween the channels, so substraction has to be done to get good colors. Maybe FOV classic blue substracts less green in the blue channel ? It's worth noting that the other colors changes (red to yellow, brown to gray) already appear in the regular Foveon fullspectrum look.

 

[Fedia]

7 hours ago, Christoph said:

You seem to have done tons of research about this

I'm litterally guessing everything

 

5 hours ago, Christoph said:

you can see the traffic lights in the distance... red is yellow, but not green:

Also with your settup the red/IR channel is still sensitive to red. it's not like on a bayer where you infrared channel (blue) is totally free of visible light. So if a monochromatic red light comes to your sigma's sensor it will be picked by both the green channel and the red channel. It's actually the white balance settings that will defffine which color it's going to be in the end. The red light can appear green only if the red channel is "cruched" with the intensity of the IR and has to have its gain lowered at a point where the green channel is more sensitive to red than the red channel itslef. I think you will not have the same acid yellows you can have on a bayer camera where red light is recorded by the red channel, giving a cleaner red signal even if it's poluted with IR.

[colinbm]

These are stunning Christoph.

 

PS, can you tell me again what camera & filter you used on these last batch of photos please ?

[Christoph]

Thank you for your encouraging words and explanations, Fedia. Much has become clear to me. Makes sense what you say :-)

 

@Colin: It's the Triple Bandpass (G, R, IR), Middle Rose by Lee to reduce green a bit, and Orange 16 for the contrasts. With a Sigma SD1 Merrill

[Fedia]

1 hour ago, Christoph said:

Middle Rose by Lee to reduce green a bit, and Orange 16 for the contrasts

So you are cutting almost 75% of green light with these filters combined. It's interesting to know that not that much green light is needed to make the grass appear pink. Did you notice that these two filters played a role in the green to blue color shift ?

 

Did you not mention a polarizer also ?

[Fedia]

It's striking when we compare your SD1 aerochrome technique to other digital aerochrome techniques, the luminance of the reds is completely different. You get crimson when on bayer camera we usualy get bright red.

 

I noticed darker reds when using the DB850 on my sigma DP1s. I think it's because pure red light is expected to be recorded in both the red channel and the green channel in a foveon sensor. But since the TB ot the DB  have a spike of IR at 850nm (which begins to be really really far from what the green channel can pick up). The IR signal ends up being recorded almost only in the red channel which leads the camera to think the 850nm IR light comming in is a very deep kind of red at the edge of humans vision, so it represents it as crimson.

[Christoph]

58 minutes ago, Fedia said:

So you are cutting almost 75% of green light with these filters combined. It's interesting to know that not that much green light is needed to make the grass appear pink. Did you notice that these two filters played a role in the green to blue color shift ?

 

Did you not mention a polarizer also ?

The photo on Page 1 with the blue traffic light was taken with TB and Middle Rose. I added Orange 16 to get a less pink vegetation - but the colors may have been due to time of day. The polarizer was only used for the contrasts in the sky...

[Christoph]

Fedia, you once asked for jpegs... here they are... SOOC with minimal adjustment of white balance in Sigma Photo Pro...

The first 2 from last Summer (with polarizer) the others from yesterday with autumn trees...

 

[colinbm]

Lovely to see Christoph
Nice to see the deciduous trees, rare in Australia.

[ulf]

I am impressed by all work done here for simulating Aerochrome and like results where red colours are rendered correctly as yellow.
I lost track of what types of cameras and filters that were used in the separate posts.

Are all of Foveon type?

 

When searching for the DB850 filter I found that two of those filters are available on eBay to nice prices here: https://www.ebay.com/itm/175439590389

Be aware of that it is important to have the correct sensor type and filter for this to work as expected!

 

Maybe Christoph and/or Fedia can make a clear summary of that?

 

BTW a polariser was mentioned in one of the posts.

Polarisers normally do not have a polarising effect at the upper NIR range around 850nm.

There they act as long pass filters with a one stop increased transmission compared to in the VIS range.

[Christoph]

32 minutes ago, ulf said:

I am impressed by all work done here for simulating Aerochrome and like results where red colours are rendered correctly as yellow.
I lost track of what types of cameras and filters that were used in the separate posts.

Are all of Foveon type?

 

When searching for the DB850 filter I found that two of those filters are available on eBay to nice prices here: https://www.ebay.com/itm/175439590389

Be aware of that it is important to have the correct sensor type and filter for this to work as expected!

 

Maybe Christoph and/or Fedia can make a clear summary of that?

 

BTW a polariser was mentioned in one of the posts.

Polarisers normally do not have a polarising effect at the upper NIR range around 850nm.

There they act as long pass filters with a one stop increased transmission compared to in the VIS range.

ulf, this is all Foveon, Sigma SD1 Merrill with TB550/660/850 and "Middle Rose" from Lee. Yes, SOOC works only with Foveon sensors.

The polariser has a big effect here... It really darkens the sky and boosts saturation and contrast, maybe even too much for my taste. 

Glad you like our trials and reports here...