[Donald]

It's not a subject I'm too versed in, but I share some issues.

The depth of field non-issue may be, as you point out, due to long exposure times circa 1900. The films' or previous wet chemistry took so long to expose didn't all shots have infinite focus?

I have shot digital with dark red filter, ignorantly attempting to borrow film black and white techiques into digital. In colour of course the images all came out red, converting to black and white the change was negligable from unfiltered.

I assume this, and your digital experiments, are due to fundamental performance of the Bayer array sensors' colour interpolation. Regardless of what light is captured do not its colour interpolation algorithm always and only attempt to produce best full spectrum? If so, in digital capture there's no trick with filters or lens that will have much affect.

I am heavily into high multiple exposures. Of hundreds of attempts not one red composition has satisfactory red colour, they're always limited tonal range, always going into purpley-pink. Everything I read about the Foveon sensor says this may be best solution in digital domain. Unfortunately everything I read about Sigma cameras they seem to be unmitigated pieces of shit, with one incomparable sensor!

[RioRico]

The depth of field non-issue may be, as you point out, due to long exposure times circa 1900. The films' or previous wet chemistry took so long to expose didn't all shots have infinite focus?

From my readings on the history of photographic technique, the answer is: no. Fast lenses (f/2.5, 1850s) and fairly fast emulsions (1-2 seconds, 1870s) were available; and portraitists from tintypists on up certainly knew how to use shallow DOF. The problem had to do with focus: actinic (UV-violet-blue) light focuses in a different plane than the green-yellow-orange we see best, so if you haven't prefocused or adjusted (like using the IR mark on a lens) you need to stop down for sharpness, resulting in longish exposures. Shooting fast with a 250/2.5 lens on 8x10 or 11x14, DOF is quite shallow, especially since your plate only captures a thin slice of spectrum. I think that's the cause of the Matthew Brady effect.

Again, what prompted this quest was reviewing shots I took long ago with a non-color-corrected 50/3.5 lens on 135 on B&W ortho/litho film processed for continuous tones, and noting that the DOF was much shallower than with Plus-X. I can't figure how that could happen, but I'm trying to replicate it digitally. That doesn't seem to be working, even with a 47B violet-blue filter, so I'll go back to film and see what happens. One factor may be emulsion thickness, reacting to light (a slice of spectrum) rather differently than do planar sensors.

I assume this, and your digital experiments, are due to fundamental performance of the Bayer array sensors' colour interpolation. Regardless of what light is captured do not its colour interpolation algorithm always and only attempt to produce best full spectrum? If so, in digital capture there's no trick with filters or lens that will have much affect.

Suppose: a sensor where different-color sites, or their filters/microlenses, are on different *levels*. Stick the blue out front, the red in back, to accommodate the slightly varied focal points of different frequencies. How would the captured image change vs. totally flat sensors?

Everything I read about the Foveon sensor says this may be best solution in digital domain. Unfortunately everything I read about Sigma cameras they seem to be unmitigated pieces of shit, with one incomparable sensor!

What are the quality of imagery from their spectrum-slicing forensic cameras?

This discussion is about to slow down even more, as I'll be gone for a few days. I'll have to wander around Lake Tahoe, shoot digital and film images, celebrate an anniversary, instead of concentrating on these vital forums. Oh snap. Cya.

[newarts]

Here's an idea of the color sensitivity/separation of the red (top), green(middle), and blue(bottom) channels from a sunlit scene (jpeg) - Cloudy WB, K100D.


There was some visible light slop into the right hand side of the image so treat this as semi-quantitative. As you can see the red sensitivity extends well beyond the visible at about 20% maximum. The blue slops over into green quite a bit. I'd think a green blocking filter in combo with splitting channels like this would do a good job for "actinic" photography.

This was done with a $5 grating spectrometer from eBay & a 24mm lens. The top image is a crop of the image from the spectrometer. The bottom graphs are intensity analyses of the color channels of the resultant jpeg by ImageJ.

Iowa Dave

[OregonJim]

And I'll try your idea of shooting with film reversed, altho I'm not sure which cam is primitive enough to handle an upside-down 135 cart.

The easiest way is probably with a reusable a bulk-load cartridge. Open a regular cart in a changing bag, and reload the film backwards into the reusable one.

[RioRico]

A quick update for the pervs following this thread: It looks like the 47B filter (passes only UV-violet-blue light) on the K20D does a good job of bringing back that old-timey look. I did a spotty bit of RAW+JPG shooting the last few days, and I have some preliminary results to talk about. I haven't had time yet to work on the images, and might not for a few days, so y'all will just have to wait.

Exposure-mode and focus are critical. What doesn't work with a 47B filter: AF (there's not enough light for the K20D to autofocus with any accuracy); the viewfinder (too dim); P-mode (extreme EV adjustments are needed); color.

What works: Set the digital filter (Custom Image) to B&W-blue. Use LiveView and manual focus. If the CI isn't set, you'll hardly see anything on the LCD screen or viewfinder. (The UV-pass filter newarts mentioned above probably wouldn't be usable, as even the 47B passes very little visible light.) Use Av-mode (and watch the ISO climb) or M-mode (and shoot slow). Focus manually. Hope.

I sloppily cut that flimsy Wratten gel filter for a 49mm ring, so the fastest lenses I can use it on are a 28/2.8 (DOF isn't very shallow), 37/2.8, 50/1.4, 85/2, 135/3.5, and 200/5.6 (sharp but not too fast). I'll likely give a 40-80/2.8-4 a whirl too, maybe with a 180 fisheye - that should be weird enough. I've so far only tried the 28 and 50, just enough to build those preliminary DoWork/Don'tWork lists. Now it's time to take all those lenses somewhere interesting and do some semi-serious test-shooting. Stay tuned for further developments.

Note: I also tried the 47B on my 'NightShot' Sony V1. The mix of UV and IR is definitely odd, but I can't say that it 'works', at least not for anything resembling reality. But reality is over-rated, right?

EDIT: Iowa Dave, thanks so much for generating those graphs! I'll have to look for a spectrometer and gen a chart for the 47B filter - or I'll cheat and point to the chart here: http://newportglass.com/kodak/kod47B.htm - Adorama has a 75x75mm 47B for US$14 here: http://www.adorama.com/LEB47B3.html which could be easily trimmed to fit most lenses. (Hint: cut the get filter to just fit within a lens' front ring, then screw in a lens hood to hold it in place.) I think a combination of the 47B and selecting the Blue channel, should give a full effect. Maybe a lighter CC filter, like the fairly common (and cheap) 80A or 80B, to block/most some of the Green and Red, and using the Blue channel, would also be effective, without cutting the light so many stops. That could make autofocus more feasible. I'll try that out Real Soon Now.

The easiest way is probably with a reusable a bulk-load cartridge.

Ah, bulk-loading. I gave away my darkroom gear, including bulk carts and changing bag, back around 1992, when I thought I'd forsaken chemical photography forever. What a FOOL I was! Gave it all to a girl and never even got kissed. [Slams head against wall.] But wait, an idea comes to me! [Slams head again.] Ah yes! I just received a Sprocket Hole Revolution 135-620 adapter set, a couple pieces of whittled wood -- which I used as templates for crafting 135-120 adapters from a dried-out plastic felt-tip pen. I can easily fit a 135 cart backwards into my 120 Zeiss Ikonta-B folder, or the 620 Ansco Panda brownie or Anscoflex TLR. Now I can get sprocket holes AND diffused color-shifted reversed-emulsion images, all in one! BONUS: The Anscoflex was designed by Raymond Loewy, creator of many rounded-edge Modern products, so I'll look especially Hip while using it. Holgaroiders got nothing on me! (Holgaroid - that sounds like a nasty medical condition. Ewww...)

[newarts]

Here's a semi-quantitative/schematic comparison of the spectra of a Wratten 47B, a "typical Bayer filter array"*, and K100D measured spectra (red, green, blue channels, top to bottom.)


I squished the x axis of each spectrum so they are comparable. The main point is that the Wratten 47B's bandpass is narrower than the blue channel of these Bayer filters.

Using the K100D's blue channel for Actinic photography includes more green than the W-47 would allow.

Using just the blue channel of the K100D works pretty well however; take a normal photo and select only the blue channel in post-processing.

Iowa Dave

*http://micro.magnet.fsu.edu/primer/digitalimaging/cmosimagesensors.html