Michael, why is the additional IR filter needed? I can't find a reference to that.

EDIT: OK, found a reference on the DVInfo site. Has it been established that IR is a problem, or just that people have latched onto this. Do you see a difference?

Serena: When filming in the desert with hot sun where ground and bush temperatures are high green bushes come out brown and sand gets much darker. HOWEVER only tomorrow am I back into the outback as you would say to test my filter. I'll let you know how it goes.

Ir filter should work, it clips the ir range(sub-red wavelength), but I would think a great UV filter would be critical.

Interesting. I thought IR blocking filters were standard on electronic cameras except for those used in astronomy. IR radiation from stars adds significantly to the luma signal, so reducing necessary exposure times. Canon make a model of one of their still cameras without an IR blocking filter specifically for astronomy, and you must add an IR blocking filter for terrestrial use. Since filters affect the transmission of specific wavelength bands (no transmission at longer than, say, 700 nanometres; visible red covers from 750 to 620 nM), I wouldn't have expected adding an additional filter to have any additional effect. Certainly at their temperature the sand and foliage will be emitting wavelengths much longer than that. However maybe the built-in filter doesn't cut as sharply as it ought and is letting through some of the near infra-red of the sun's radiation. If that is the case I would expect to find the problem in all sun-lit situations.
I'll be very interested in your results.

The problem is the difference in IR reflectivity of certain black materials causing a magenta shift. This isn't unique to the EX1. The problem has been a pretty hot (pardon the pun) topic on Reduser as well.
I've heard of people even having the problem when they've gelled windows. The gells cut the visible but not the IR light. Same thing happens with a camera's internal ND filters. You attenuate X stops of visible light but not the IR. Result is the amount of IR reaching the sensor increases X stops.

The IR blocking filter is standard on cameras, I believe it's the same hunk of glass as the OLPF filter.

In one way it's perhaps good that the EX1's internal filter isn't that good if that helps shooting at night with IR lights.

Bob.

The sensitivity of electronic sensors to IR is good in astronomy but not in terrestrial use (other than night). I would hope that everything longer than 700 nM would be cut, so the IR reflectance of materials would be irrelevant. Poor design when this isn't the case.

EDIT: http://www.sciencecenter.net/hutech/idas/uibar.htmthis is how it should be[/link]

I'm not so certain it's poor design. There seems to be two types of IR cut filters, ones that reflect the out of band wavelengths and the ones that absorb them. The former need to be used in front of the lens and in front of any other filters, those Hutechs seem to reflect a lot of the out of band wavelengths.
The absortive filters are the ones used inside the cameras as having IR bouncing around could cause problems. Factored into this seems to be that the amount of absorbtion is dependant on angle of incidence.
Certainly this doesn't seem to a problem that's easily solved with 100% success. The Lieca M8 has a problem with IR and not one that was easily solved.

There's some interesting reading on the problem with Red and IR in general here:
http://www.reduser.net/forum/showthread.php?t=9412
along with some pretty extereme examples of the effect.

Bob.

Bob,

Interetsing readin it is; thanks. However, it says:

Due to the significant amount of ND, the ratio between IR and the visual spectrum has been reduced

This is a statement that doesn't fit to the typical IR-related magenta-shift seen on the EX1: I usually spot it indoors with artificial (tungsten) lighting, where I don't use ND filtering at all.

It's exactly the same issue.
If the internal IR cut filter was perfect it wouldn't matter what the external NDs were doing. Note the comments from the designers of the RED camera. They too have had to make compromises in the design of the combine OLPF and IR cut filter. The designers of the Leica M8 faced the same challenge and might arguably have fallen way short of the mark. Or not. They seemed to have faced the challenge of a fixed lens with a very WA. An internal IR cut filter might have had to cope with filtering IR with a low angle of incidence creating a really bad problem.

Engineering is always about compromises. I'd rather add an external 486 filter than find out the performance of the camera was forever locked down by some glued in internal filter that was scattering light and causing other problems.

Also keep in mind that the internal filters require much better quality than an external filter by quite a large margin.

Bob.

My B+W 486 is due tomorrow; I'll keep you posted about the results.

Bob: IF the 486 is a reflective type that needs come at the front of the matte box then what I am doing may well have problems. I hope the 486 is an absorber. I have emailed B+W but no reply as yet. Anyway, as I told Serena I'm off into the desert testing today. Will be back Thursday PM and will check in.

The Hutechs are http://www.sciencecenter.net/hutech/canon/ffmount.pdfinternal mounting[/link] behind the lens. These are dichroic filters that destructively absorb particular wavelengths, rather bounce it back out through the lens. Perhaps higher quality than generally installed in video cameras. Pity really. Guess I'd better take off the haze protective filter and install an IR in its place; general thought is that the UV aspect is unnecessary anyway.

That's just what I did Serena. Exchanged my UV/Haze for a UV/IR. B+W confirm it is an absorbing filter and works the same both ways.

My field test did not happen as we had a cold spell. In the lab measuring the throughput of the EX1 with no ND filter, the IR filter DID cut the red component as shown by the Vectoscope of the signal even at a cool 20C. I have not the facilty to measure IR directly but it did cut radient heat as tested from a heated hot plate.

It's interesting if you can record a body at 20 deg C, because that is emitting at around 10,000 nM (way into far-infrared). Granted that a black body emits at all wavelengths http://en.wikipedia.org/wiki/Image:Wiens_law.svgPeak radiant wavelength[/link], but the intensity of visible light from a 300K radiator will be extremely small. This is the reason I questioned the problem with your sand and foliage, the latter being unlikely to be hotter than 50C without dying (8780 nM). Accepting that the EX in-built IR filter transmits enough to distort the red end of its response, I suggest that this is reflected IR (ie. from the Sun) and not because things are hot. In terms of your filter cutting radiation from a hot plate, it wouldn't be any good if it didn't.

I did not explain my test very well and I was going to edit it but Serena answered first. What I was really saying that even at low temperatures like 20C where there may be little or no IR the filter still has a slight cut of red in the visible spectrum. Maybe Bill Ravens should re run his EX1 color correction tests with an IR filter on as I am sure the results will need to be compensated for the red cut.

Now here's an interesting question. Assuming the filter is attenuating part of the visible spectrum would WB correct for this?

Still waiting for my filter so I can't test this myself.

Bob.

Michael,
The difficulty here is in sorting assumptions from reality. I'm not saying there isn't IR radiated by a body at 20C, because there can be a great deal. This is a colour temperature of 293K, and is readily seen by thermal imagers. However it is predominantly at wavelengths in the far infra-red which I'm sure the IR filter in the EX wouldn't let pass, no matter how deficient it is in the near infrared. But that is my assumption. My only point is not to dispute that the EX lets in too much near infrared, but to point out that even hot desert sands don't get hot enough to emit radiation of relevant wavelength. But they will reflect IR from the Sun (or other hot source) and I suggest this is what appears to be getting through the camera's IR filter. Ideally the filter shouldn't transmit anything longer than 750 nM (http://en.wikipedia.org/wiki/Visible_spectrumvisible spectrum[/link]), but it seems that it is and this will contaminate the camera's colour rendition. So adding an IR filter, as you have done, is necessary. I'm just being picky about the source of the relevant IR.

Bob, probably true to suggest that not all IR filters are equal, and a "cheap" one may well begin to cut in the visible spectrum. In a general sense, cutting the red end of the spectrum should make the source seem to be a higher colour temp, so I suppose WB would try to compensate.

I have just run my first test with the 486 outdoors (moderately bright sunshine) and can confirm that IMHO, the effect IS visible in that the reddish cast (especially using Hisat matrix) is gone. I am using totally neutral matrix settings; Bill's presets are not valid any more with this filter (to my naked eye, at least).

In the evening, I'll run some more test in tungsten light to check whether my wife's black blouse is really black now, with the filter on.

Just a quick update:

Yes, the black clothes look really black, and deep blue are not magenta anymore, with the 486 filter attached in tungsten light (2600-3200K).

However, with all matrix settings being default (for both Hisat and Cinema matrices), red is not properly red anymore; will have to play with the settings and try to correct this.

Also, in tungsten light the corners indeed get greenish tint at fully wide.

All in all, the colours on my V1E looked much better, and without any filtering or the myriad of tweaks needed - this IR problem is something Sony should definitely be made to admit, and treat as a serious flaw on the EX1!

Another issue from astrophotography that may apply here is that IR contaminates all colours (not just red). Commonly we use monochrome CCDs and use dichroic filters to record separately luma (including IR) plus red,green and blue images. If an IR filter is not used with each of the RGB filters, those exposures are thrown out of balance because those filters are designed for the visual spectrum and do not exclude IR. We might expect similar in 3 CCD/CMOS cameras.

Isn't CT a kind of median value for the wavelengths from a black body? At a given temperature the wavelengths radiated by a hot body seem to depend on more than just the temperature.

Bob.