DCS 560 Infrared

Discuss older Nikon-based Kodak digital SLRs, including DCS 100, DCS 200, NC2000, DCS 400/600/700-series, etc. Ask questions, post general comments, anecdotes, reviews and user tips.
Post Reply
Ashley_Pomeroy
Posts: 119
Joined: Fri Aug 08, 2008 10:04 pm
antispam: No
Location: England
Contact:

DCS 560 Infrared

Post by Ashley_Pomeroy »

I was curious to see how sensitive my DCS 560 is to infrared and whether I could produce infrared photographs. The older DCS 460 I used to own tended to make everything look very pink, but I never really tried any intentional infrared photography with it because (a) the camera didn't have a screen, and working out a table of exposure and focus settings would have been long-winded and (b) I don't have a fast wide angle Nikon lens. I do however have an Olympus 24mm f/2.8, which is sharp at f/2.8, and on my DCS 560 it's about 31mm, which is still wide. Not amazingly fast but then again neither am I. So I bought a 49mm 760nm infrared filter and go to work, la:
Image

Image

In general at f/2.8 and ISO 80, in bright sunshine the exposure times were 1/10, 1/15. Higher ISO values are decent but tend to record what must be amplifier noise at the right edge of the frame. I had a surprisingly high number of sharp images at 1/15, I assume because the DCS 560 is so massive that it dampens vibrations. As far as I could tell there was no advantage to closing the viewfinder blind. In practice I held the camera up to my face, even though the viewfinder was too dark to see through, because I could stabilise the camera more effectively that way. It was chimp-city otherwise. The DCS 560's histogram seems to have a natural "expose to the right" bias and so images that looked underexposed on the screen were in actual fact not. I used to hate Stereolab - too snobby - but in retrospect, and now that they have split up and are no longer a threat, I quite enjoy some of their music. Can't stand the band.

Oh yes, the Olympus 24mm is a relic of the OM era and infrared infinity focus is at the left edge of the 6ft mark, at least with my adapter and camera and lens. With visible light infinity focus is at the ledge edge of the infinity mark.

If I stick a hot mirror filter in front of the infrared filter the result is much darker, which is obvious in retrospect, although it's nice to know that the hot mirror filter actually does something. Having said that, even in bright sunshine I find that the DCS 560's colours are generally good even without any filter at all, except for the occasional burst of flare around the edges of bright lights. Interesting to compare it with the sample images in DPReview's review of the DCS 760; they seemed to use a lot of sharpening and must have been output from Kodak's software.
Image

Image

I used a custom white balance taken from the grass. If I whack up the saturation, the ground turns green and the sky turns red, but the colours are blotchy and horrible. My Fuji S2 can produce better colour infrared images, with longer exposures, but it's nice to have a camera that can do this hand-held.
Image

Image

The big problem is noticeable vertical banding. I don't know if this is because of poor sensitivity to red, or if it's a consequence of the way that the camera reads data from the sensor. It's obvious in the following image (although horizontal, because the image is in portrait orientation):
Image

If only there was a really good ≤20mm ≤f/2.0 lens that is sharp at f/2.0 and costs less than £1,000, assuming such a lens exists. But then again it would be cheaper to have e.g. a Canon 1000D modified by Lifepixel, in which case speed would be less of an issue.

My next idea will be to experiment on a human subject. I have an Olympus 50mm f/1.4. I need to work out the correct focus settings for a couple of metres, and I need to find someone who will stand a precise distance from me. Preferably someone who radiates infrared, or reflects it. The Human Torch would be a good subject.

Otherwise the DCS 560 is pretty good. The images interpolate up to 48mb without looking awful although I can't imagine many photo libraries would be too happy to see shots taken with the camera in the twenty-first century.
Stan Disbrow
Posts: 601
Joined: Fri May 20, 2005 7:33 pm
Location: Raleigh, NC USA

Re: DCS 560 Infrared

Post by Stan Disbrow »

Hi,

Yes, you have a bit of artifacting due to the bucket-brigade method of moving the voltages from pixel capacitor-to-pixel capacitor on the sensor. That's the worst issue with a CCD as I see it, though. Otherwise, it's working pretty darned well.

Later!

Stan
Amateur Photographer
Professional Electronics Development Engineer
Ashley_Pomeroy
Posts: 119
Joined: Fri Aug 08, 2008 10:04 pm
antispam: No
Location: England
Contact:

Re: DCS 560 Infrared

Post by Ashley_Pomeroy »

Damn those buckets. DPReview's review of the DCS 620X has a lot of stuff about how the Kodak sensors were "full frame" (rather than "interleave transfer") but my brain seized up as I read the words and looked at the pictures, and what does it matter anyway. It is how it is.

If there's one thing I have learned from this process it is that I naturally lean to the left. I need to train my thinking brain to master my body and lean to the right slightly. There's a metaphor buried in there somewhere. Must resist.
Last edited by Ashley_Pomeroy on Mon Dec 21, 2009 7:29 pm, edited 1 time in total.
Stan Disbrow
Posts: 601
Joined: Fri May 20, 2005 7:33 pm
Location: Raleigh, NC USA

Re: DCS 560 Infrared

Post by Stan Disbrow »

Hi,

Yeah. There's no good answer, either.

If you use the CCD design, there's almost nothing to get in the way of the light and the photodiode. That means good sensitivity, relatively speaking. But, it means that some degradation occurs between areas of high light and low light during the process of reading the thing.

If you use the CMOS design, then you escape the reading issue as each site has it's own electronics for reading the thing. But, then all the extra electronics take up space, leading to more stuff in the way of the light and lower relative sensitivity.

As the barker at the carnival says: "You pays your money and you take your choices!" :P

At least the old Kodaks use Full Frame Transfer CCD's, not interleaved. All that means is that all CCD rows are read from one side of the chip and all at the same time. Interleaved reads the odd rows, and then the even rows. Also the odd rows are read off one side of the chip and the even rows off the other side.

The full frame transfer means you get less artifacting, but need a faster processor and associated electronics, which means more cost. It also means you have to shutter off the incoming light during the reading process. That's not an issue with using a modifed film body, of course. It also means a slow frame rate for shooting, but, again, not an issue for what Kodak was making these cameras for.

I recall the issues with the Canon 1D mark-I, which used an interleaved CCD. While it had a shutter, it still suffered from increased artifacting if one had high contrast in the shot. It was at least 3x as bad as what you're seeing. So, you can look at what you're getting and actually be happy about it. ;)

As far as leaning goes, maybe you need to shoot with a large-brimmed hat with the downhill side of the brim pinned to the crown and the uphill side left down. You'd look like a US Civil War officer, but that might balance you out! :P

Later!

Stan
Amateur Photographer
Professional Electronics Development Engineer
Ross_Alford
Posts: 105
Joined: Tue May 16, 2006 1:10 pm
Location: Australia
Contact:

Re: DCS 560 Infrared--a trick for monochrome IR

Post by Ross_Alford »

A trick with IR shots--

It may not work with a 760 nm IR filter, and may not work with some versions of the filter arrays on various digital cameras, but try shooting raw, and using dcraw with automatic white balance and -d (document mode, does not demosaic) to develop. With an 800 or so nm filter on my Nikon Coolpix 8400, which has had the IR cutoff filter removed, but has not had the Bayer array removed, this gives spectacular results. Apparently at least on that camera the R, G, and B Bayer filters all pass about the same amount of really long wavelength IR, and telling dcraw to use auto white balance levels out any small differences, so you end up with the equivalent of a pure monochrome sensor.

Using that trick, the detail resolved by the 8 mp coolpix 8400 is about the same or slightly better than I get in color images with a 12 mp D90. With a Hoya R72 (720 nm I think) it doesn't work, too much difference among R, G, and B pixels with wavelengths that short apparently. I haven't tried it with any of my old Kodaks, but anyway I would suggest having a go and seeing if it works.

Cheers,
Ross
NikonWeb
Site Admin
Posts: 1029
Joined: Fri May 20, 2005 4:12 pm
antispam: No
Location: Norway
Contact:

Re: DCS 560 Infrared--a trick for monochrome IR

Post by NikonWeb »

Ross_Alford wrote:..try shooting raw, and using dcraw with automatic white balance and -d (document mode, does not demosaic) to develop.
A while ago I posted a short and very basic description on how to use dcraw. May be handy for new users: viewtopic.php?f=2&t=328

Jarle
Ashley_Pomeroy
Posts: 119
Joined: Fri Aug 08, 2008 10:04 pm
antispam: No
Location: England
Contact:

Re: DCS 560 Infrared

Post by Ashley_Pomeroy »

This is fascinating - I've just been experimenting with DCRaw and it does genuinely appear that the -b trick mentioned by Ross works! Or at the least it can produce the illusion of extra detail. I didn't doubt him for a moment, mind you. Contemplate the following 100% crops from the middle of an image:
Image

On the left is ACR; on the right is DCRaw, run with "dcraw -b 0.3 -a -d -6 3N8B1381.TIF". This chucks out the interpolation, lowers the brightness, uses auto white balance, and produces 16-bit output. If I don't use auto white balance the image has a half-tone pattern (which is present if I feed in a colour image as well).

I placed the images side by side and then ran my standard unsharp mask settings on the whole thing (100, 0.5, 0). The image on the right does appear to have more detail than the one on the left. But it's grainier, because ACR appears to be using noise reduction, even with the luminance and colour sliders turned down. The grain in the DCRaw image is very obviously pixellated whereas the grain in the ACR image is smoother. It could be that the extra detail is just sharpened grain or enhanced contrast. But I'm not a scientist, this is outside my sphere. Very impressive though.
Post Reply