Skins still redish with IR rich light and B+W filter

[pascal_meheut]

Whilst shooting at a party, I noticed that the skins were still redish using a B+W 486, firmware 1.102 and several profiles. So I made a test: I shot myself in RAW mode holding a WhiBal with flash and with an 2400K halogen light...

Even after doing the WB on the WhiBal in C1, the skin is much redish with the 2400K light than with the flash. Such phenomenon does not occurs with the DMR so I guess the 486 does not filter all the IR.

As anyone else noticed this ? Or tried with a Leica filter to see if they are doing a better job ?

BTW, anybody knows where to order them from online in Europe ?

[wparsonsgisnet]

Pascal, when you mention several profiles, are you using Edmund's?

I am using Edmund's profiles and find them very successful. I don't know if he is working with the Leica-branded IR-cut filters, but he was using B-W 486's, I believe.

[lct]

Originally Posted by pascal_meheut

...anybody knows where to order them from online in Europe?

No sorry but E39 (#13410) and E46 (#13411) are available from time to time at B&H.
Take 3 days from order to delivery in France.
They'll ask you to send them a scan or a photocopy of your credit card if you're a new customer though.
LEUVAIRE39B
LEUVAIRE46B

[eronald]

Originally Posted by wparsonsgisnet

Pascal, when you mention several profiles, are you using Edmund's?

I am using Edmund's profiles and find them very successful. I don't know if he is working with the Leica-branded IR-cut filters, but he was using B-W 486's, I believe.

People are welcome to try my profiles under Tungsten lighting and give me their opinion on whether there is a problem.

My own people tests under Tungsten lighting look ok, with just about the quality I'd expect from this prelim profile.

I don't think Pascal has my current test profiles, he has his own set, which have different qualities, and I expect this is what he uses.

The color variation he mentions can be due to IR leakage or to metameric effects. If it were IR I would expect to see it also in the Jpeg files.

Edmund

[pascal_meheut]

It does the same in jpeg. And this is not a profile issue because even if do not have the last ones, the difference occurs with the C1 generic profile, Edmund previous profiles (for 1.092 and before), mine and the one posted by dfarkas (which is basically the same as mine)

[Eoin]

I have been thinking this for a while, I notice a magenta/purple tone in applications like lightroom cs2-3. I believe it is down to the matrix values leica are supplying within the DNG.

Let me put it another way, if I use the proper matrix values held within the DNG to make a raw support profile within Aperture and leaving aside the camera calibration multipliers for the moment, I'm finding the balance is skewed towards the purple magenta. The multipliers can then be used to town it down but it's still skewed.

For your own curiosity shoot some purple-red subjects filtered or unfiltered and process it through LR or CS2/3 and see if they are actually true to life. For the life of me I can't get them to appear on screen or print even close the the genuine article.

[eronald]

I too find there is a problem with red skin casts often appearing on the M8, although in my view this problem is not limited to Tungsten light. In fact one of the main reasons for using edited profiles such as my previous Tuned2 is to keep this skin red under control. By popular request, Tuned2 did it in a brutal way for 1.092 firmware.

This was the reason why I made my first edited Leica profiles. FOR MYSELF. I already had an IR filter but my images were off with respect to skin tone.

And I am once again going to have to release a profile which does it for the current firmware. Maybe tonight. One reason the current profile is called prelim is because it does not do this yet.

Existing profiling tools are designed to improve good color, not to fix broken color. So in the end it's up to Leica to deal with the color rendering of their camera.

Leica will have to deal with this sooner or later rather than continue for years to provide workarounds. I believe this means finding a sensor/IR filter combo which has better *visible color* rendering properties. THIS IS NOT A COMPLAINT, IT'S ADVICE.

Edmund

Originally Posted by Eoin

For your own curiosity shoot some purple-red subjects filtered or unfiltered and process it through LR or CS2/3 and see if they are actually true to life. For the life of me I can't get them to appear on screen or print even close the the genuine article.

[Jamie Roberts]

Just to put this in perspective, most digital cameras don't do so well on caucasian skin tones under tungsten light; every digital camera I've owned needed to have adjustments in PS for skin tones.