Infra Red Again

[Jeff Kilgroe]

The Rosco Trucolor IR's don't require any stop compensation and they are cheaper to buy than an 80C or 80B.

:huh: But the Trucolor IR and an 80C / 80B don't serve the same function. <edit> I suppose both an 80x filter combined with an IR cut filter could be a good solution when shooting with very hot tungsten sources, but I haven't tried that.

I threw one in my kit just to bring it out and save the day when necessary.

Good call.

Correcting in post might be OK for drama, but on TV commercials where you have people comparing Pantone swatchbooks, it is a better career move to get it right in camera.

True, but if this is the case, you will be shooting under sunlight or full spectrum lighting and not tungsten or using correction for such. At least not with the RED One. The sensor is balanced at about 5000K. If you want to "get it right in camera", you will need to light accordingly and sunlight or full spectrum will perform much better than flos or HMIs that are color balanced to 5000~5700K.

But we're branching off on a tangent here. I was just saying that if the correction is necessary, and in lots of situations it will be, it is easier to compensate in post for IR tinting than it is for the spectrum of shooting with tungsten. At least that has been my experience with shooting RED. It has also been my experience that under severe IR conditions, color tinting is the least of your worries. I have encountered far more issues with halos and effects that look like CA or overall desaturation, but are caused by IR. Various effects with IR seem to differ with the lenses I'm using. But I suppose just as with visible light, IR wavelengths will propagate through one type of lens differently than another.

 

[Joe G.]

xb29

xb29

"Because the XB29 blocks cyan light (as well as yellow) you get better blue vs. green as well as better green vs. red."

This sounds like an exciting new look. Does anyone have examples to post?

 

[Dalibor Fencl]

everlasing question: filter vs post

everlasing question: filter vs post

I can easily see the advantages of blocking UV and IR, those parts of spectrum affects the sensor, but blocking cyan and yellow? Couldn't it be done easily in post too?

 

[Brent J. Craig]

I can easily see the advantages of blocking UV and IR, those parts of spectrum affects the sensor, but blocking cyan and yellow? Couldn't it be done easily in post too?

I'm gonna need someone to explain to me how removing parts of the visible spectrum could give you better color!? Wouldn't blocking cyan and yellow just make your shots look more red and blue? Purple?

 

[Dalibor Fencl]

I'm gonna need someone to explain to me how removing parts of the visible spectrum could give you better color!? Wouldn't blocking cyan and yellow just make your shots look more red and blue? Purple?

Well, yes, it's bit theory of of human sense to colors in it ...
but
there are three sensors in camera: Red, Green and Blue, no sensors for yellow, cyan, magenta and millions of other colors, nonetheless the camera can see them as a combination of RGB. Bit similar like human eye and brain, but not the same way, maybe there are too much of yellow, maybe not enough ... it's a mystery.

What's not mystery is distorted picture due IR, UV, that we can't see with our eyes, but the camera can.

 

[Dan Hudgins]

Cyan and yellow...

Cyan and yellow...

Well, yes, it's bit theory of of human sense to colors in it ...
but
there are three sensors in camera: Red, Green and Blue, no sensors for yellow, cyan, magenta and millions of other colors, nonetheless the camera can see them as a combination of RGB. Bit similar like human eye and brain, but not the same way, maybe there are too much of yellow, maybe not enough ... it's a mystery.

What's not mystery is distorted picture due IR, UV, that we can't see with our eyes, but the camera can.

The filters in the Bayer filter are not saturated, so when you use an IR and UV filter you cut into the Red and Blue filters part of the light, this makes the Red and Blue sensors bias toward Green making the colors less true and bright.

You can matrix, my software lets you fine tune the matrix to try to cut the blue and green out of the red, the red and blue out of the green, and the red and green out of the blue.

But if your red and blue signals have noise, there are only half as many sensors for red and blue as green, then when you matrix you are adding "negative" noise to the other two colors, in this case you add noise from the blue and red to the green and so on.

To get the minimum noise it is better to filter the light before it gets to the sensors, and to have the light signal from all three sensors close, hence the need for both a XB29 filter and a 80 series filter for use under tungsten, and a XB29 filter and a skylight or 81EF when shooting in open shade from blue sky.

To say that the sensor has SOOO much dynamic range that you can just use matrix, levels, curves, and gamma in post to correct the color is to settle for less than the camera can give you. Its best to use the maximum dynamic range of the sensor and roll off the head and foot of the transfer curves, like film, and not to clip the top or bottom.

You want to cut the cyan because if you just cut the yellow green and blue will give a bigger signal than red, and you will get better seperation of the green and blue colors.

The colors of most objects are not single band, so when you make the peaks for red, green, and blue more sharp with cyan and yellow cut you still get yellow and cyan in the image, they just become part of a more bright range of distinct colors.

I did some tests scanning color negative film using a Bayer sensor color camera just with the Bayer filter, and then in three color separations using Wratten 98,99, and 70 narow band RGB filters. I then tried to matrix the single exposure Bayer filter image to match the separation exposures as best as I could, but the seperation exposures looked better and cleaner. So although you can increase the saturation in post from a Bayer sensor, you can not get as clean a result without the three passband filter, or narrow band seperation filters. I am working with uncompressed RGB images, not YUV, 4:2:2, or some other chroma degraded formats.

Some Movie film scanners use Red, Green, and Blue LEDs for the light source rather than narrow band flters. You could light with just RGB LEDs and get the same sort of color enhansing effect since there would not be any UV, Cyan, Yellow, or IR light.

If you use white LEDs, make sure they do not have yellow or cyan LEDs in their makeup, you can check the light with a diffraction grating (you should just see red green and blue dots when you look at the LED throug it.)

---

Some lenses are very out of focus in the IR and UV, so even though film did not pick up the halo around the image, a sensor might. Because the Green and Blue sensors may pick up IR, you can get a off color or blue halo, so it is best to use a four band cut filter because like I said if you just cut IR and UV you push the two outside sensors toward green and get too much green signal.

How much improvement you see in the images would depend on the dyes used in the costumes, the kind of light used, and the lens and stop used, how much ND is used and such, but I cannot see the XB29 making the images look worse except for very wide angle lenses if you use it in front since at wide angles interference filters do not work in the corners of the frame, and that is true for other interference IR cut filters as well...