So the internal ND's for Pocket 6K Pro aren't really IRND's?

So the internal ND's for Pocket 6K Pro aren't really IRND's? Disappointing.

Says who? Tested how?

Says who? Tested how?

The built-in IRND filters attenuate the infrared light to the same degree as they reduce visible light. So they are not aggressively cutting all infrared. Some material such as black nylon seem to generate a lot more infrared than other dark material. But the material doesn’t have to be black, since we all know green leaves on trees also radiate infrared light.

So it is possible that the built-in filters work well in most situations but there may be times when an IR Cut filter is needed. The reason the IRND filters aren’t more aggressive is that they are designed to complement skin tones which can suffer some with an IR Cut.

I just wanted to quote Rick from another thread on the IR pollution as what he's saying seems consistent with what I'm seeing with my unit and with the CVP video. Personally I'm ok with the work around of using an IR cut filter on shots that need it and I'm finding that on a lot of shots it's not that noticeable but I know everyone will have their own view on this and how much of an issue it is.

I’d be more concerned about the focus shift. As mentioned by BM totally eliminating IR can have unwanted effects on skin tones so given the choice of best skin tone or accurate black on some synthetic black material I’d want the former esp as those comparisons with other cameras show the skin tone on the BM to be very good if not the best (way better than the Z6). Never encountered IR pollution on my BMPCC, P4k or P6k.

I have been researching more on this topic with respect to camera sensors.

Consumer displays being magenta-biased is implicit knowledge for many, explicit – though agonizingly so – for a few, like myself. Which is why it seems there’s wider propensity to discern “green cast”, but the magenta skew only becomes perceptible when discussing IR pollution issues.

Regarding camera sensors, this is on Arri’s FSND webpage: https://www.arri.com/en/camera-systems/ ... snd-filter

“Digital sensors require a small amount of this “far red” light to render skin tone as healthy and vibrant…”

However, just prior to that sentence, it says, “Film is insensitive to red at the edge of the visible spectrum.” So, that suggests that the notion itself is self-contradictory, because over half a century of using colour motion picture film never saw anyone complain that movies are showing unhealthy or dull skin tones, to my knowledge.

On a side note, one can argue that earlier film photography did have a normativity-based skin-tone bias – the Shirley Cards – and there was a significant variance in how vibrantly or dismally it rendered different types of skin tones. However, this does not seem to have been explored in terms of IR pollution or the widely used Kodak or Fujifilm stocks for filmmaking, especially after the 1990s.

Therefore, it appears that many cinema-camera manufacturers are caught in this self-fulfilling prophecy of creating digital sensors with more affinity to IR near and above 700 nm under the notion of producing great skin tone.

Now, this would not be a problem if it didn’t make the sensors susceptible to IR pollution owing to the very nature of how it works – cutting down more of the visible light would inevitably make IR more prominent.

So, it looks like we will have to either use FSND filters or make peace with IR pollution when using higher ND strengths for the foreseeable future.

Thoughtful analysis. When I researched this there was some variation on where red ends and infrared begins among people who specialize in their spectral knowledge.

Remember the IRND filters are used in the camera, not simply ND. So infrared is reduced the same as visible light. But with some material in some lighting, yes, you may need an IR Cut. I only shot black and white 16mm film so not able to comment on Hollywood and colour film stock.

Film stock is NOT sensitive to IR in the same way that digital sensors are.

ND filters didn't have to filter for IR, because IR wasn't a problem when shooting on film.

If you wanted to "see" IR on film, you actually had to shoot IR sensitised film AND generally use filters that ironically, are just like the very heavy ND's that are causing the IR to show up on Digital cameras today.

https://www.aerochrome.shop/product-pag ... rared-film

and you still need to generally use one of these when shooting film if you want to "see" IR show up.

https://hoyafilter.com/product/r72_infrared/

When shooting Infrared on film cameras, the lens makers used to have a different witness mark on the lens because you CAN'T SEE INFRARED and the shorter wavelengths could mean your shot was out of focus without the manual correction.

In other words the visible focus must be OUT for the infra red part to be IN FOCUS.
https://www.lifepixel.com/infrared-phot ... r-focusing

Anyhow, digital cameras see IR, film cameras do not unless you load IR sensitive film AND you filter for it in some way.

Now....we can't SEE infra red. We're bad at it.

There's no easy way to light meter for it as well. You can stand outside on a cloudy day or a sunny day and depending on the time of the day and cloud cover there could be a lot of infra red, or very little infra red and your light meter will tell you the EXACT same exposure. Your false colour says the same exposure. Your histogram will be the same. None of these tools register the AMOUNT of infra red.

What use to happen is that in the early days of digital camera, we'd use the same film era ND filters we're all used to using. Except, hey wow, the cameras all of a sudden a lot more sensitive! High speed film stock USED to be ISO / EI 500. Most daylight scenes where shot with ISO / EI 50 stock and that was in the last decade. Go back to the 50s and 60's and it was more like ISO / EI 12 and 100 would be a high speed stock.

Now days 800 is "native" and dual ISO camera have extraordinary ISO numbers.

SO.....

The cameras are a LOT more sensitive. When I started shooting on film in the early 90's, you never used more than an ND0.9 (3 stops) ever. Sometimes you'd use a Pola if you needed an extra stop. A typical full filter kit would be a three filter kit of ND0.3, ND0.6 and ND 0.9.

Now I carry a filter kit that has eight filters and goes to ND3.0, or 8 stops ! We've gone from 3 stops being plenty to needing more than 8 stops of ND.

Making ND filters is REALLY hard and it get's harder and harder with each extra stop of density because it becomes hard and hard to be consistent with the amount of light coming in. You'll notice that you see the biggest copy to copy variations in filters at the heaviest stops. They can often vary in both colour AND in density. So seven stops of ND might be six and half or seven and half stops in actuality and green or magenta.

SO

Film era ND making never cared about IR because they only blocked the visible light that the film stock sees.

Now we have these cameras that are more and more sensitive, meaning you need more and more ND to stop but perversely, the ND's are only blocking the visible light so they still let the same amount IR though. If it's film, then who cares....but

Digital sensors care.....

SO

Every stop of ND halves the amount of visible light. An ND0.3 is letting in half the amount of light as no ND. An ND0.6 is half that again, so only 1/4 of the amount of light is coming though.

Now you have a heavy ND filter that's letting the SAME amount of IR light but only 1/32 (eg with ND1.5) of the visible light. The RATIO of visible to IR light has dramatically changed. The ND is blocking all the visible, passing all the same amount of IR so now the sensor sees all the same amount of IR that previously was buried by the IR filter in the camera amongst the visible but because you've so drastically reduced the BALANCE between the visible and IR, it's showing up because YOUR ND'S DON"T WORK on IR.

So we started seeing IRND filters....but they are in-consistent because different sensors have different IR responses and you start to see IR pollution still affecting cameras even though you're now using IR filters!

If you look at the first image here, you can see an Alexa with a straight 1.5ND clearly having infra red pollution issues. Look at the lady on camera right's black top. Alexa HAS an IR cut filter though so we're still seeing IR is a problem on a camera that has an internal IR filter already. Right next to it is a full spectrum ND filter (labelled here as IR cut)
https://www.flickr.com/photos/johnbrawl ... 3844822085

So lately we've seen the way ND's are manufactured changing. Instead of the old style where it was a gel that was sandwiched in between two pieces of glass, filter makers started SPRAYING on the coatings.

Mitomo TRUE ND, whom many still consider to be the most neutral cinema ND's did it first. They made ND's that were consistent all the way to ND2.1 and instead of calling them IR CUT ND's we started seeing the terminology around FULL SPECTRUM.

In other words, not an ND with an IR filter, but ND that cut's both visible AND IR by the same amount, solving the original problem from the 35mm FILM ND days intend of adding an IR filter to a visible light ND filter.

Mitomo are great and super insanely expensive. A full set is about 10K. And they are really super super hard to keep clean because of the way the coating interacts with dirt and fingerprints. Mitomo are Japanese and have some great examples on their site if you can navigate it.
https://www.mitomo.co.jp/products/nd-filter-true-nd-en/

Since then we've seen other manufactures use the same technique, and starting to spray the INSIDE of two layers of glass to remove the cleaning problems off the Mitomo.

NISI, Firecrest, Panavision / LEE and Arri have all now released full spectrum style ND's.

Hoya PROND's are also using this spray technique and a good cheap way to access this third generation ND filter system.

ND's are still not perfect. They will vary copy to copy. Even the Mitomo do. The amount if IR that is cut is still up for grabs because there's no easy way to asses how much IR is in a given scene, nor a way to meter it and each camera sensor responds differently to it.

The internal filters on the G2 are the same as the P6K. They are considered full spectrum ND's as per the background above. Sometimes you will need a bit more IR attenuation because BMD have stuck with a recipe that thus far has given them a great reputation for their look and colour science. Part of that is very mild IR cut as a starting point.

I don't tend to use the internal ND's on my cameras much, even the Alexa when I use them because I prefer more control and want single stop adjustment.

Considering a lot of people use Variable ND on cameras in this price range routinely the internal ND's are far far better than what a variable ND is doing to your image.

I find it amusing because everyone that's posting "problems" with their internal ND is rarely bothering to show what ND's they compare them to, because I'm pretty sure 90% of the time the solution they have is no better than the internal ND's.

For that matter the new rush to eND also has it's problems as well, much like variable ND, which is why you don't see variable ND and eND used much on the larger sets.

JB

Another consideration to keep in mind is that cut filters are not "brickwalls"... there is a tail and a shoulder. Also the exact wavelength of the cut may not be the best for the situation (each material has it's specifications which may not be the perfect frequency.) So the exact wavelength of the cut and the tail/shoulder may indeed give issue with rendering skin tones. Maybe the "better" way is the wide band ND... even if not perfectly flat it might be better than dealing with the curves of tail/shoulders and a cut point that may not be perfect.

Thoughtful analysis.

Appreciate the considerateness, Rick. It’s encouraging in an age of diminishing civil discourse.

...because IR wasn't a problem when shooting on film...

Thank you for detailing your experience and thesis, John. I find it informative.

It’s clear that increasing the light sensitivity of digital sensors came at the cost of making them susceptible to IR.

I suppose a term such as “IR prominence” is less of a misnomer than “IR pollution” or “IR contamination” because the latter two imply that IR is being added to the recorded spectrum. In reality, as it’s clear now, wavelengths near 700 nm are not as suppressed as those are between 400 and 650 nm when using regular NDs or less-refined IRNDs, leaving IR more perceptible.

However, the dissertation hypothesis here is that should we have sensors sensitive to wavelengths near 700 nm under some misconceived notion of achieving great skin tone?

If yes, then it’s status quo ante bellum.

If not, then I posit that it might be preferable for cinema-camera manufacturers to limit the sensitivity of sensors between 400 and 670 nm, with the start at 400 nm and a sharp climb from 410 nm towards full perception, followed by a gradual fall-off starting at 650 nm and a stop at 670 nm, or perhaps at 680 nm. This may not only reduce IR prominence, but also minimize the need for IRNDs.

My impulse is to explore the innate reasoning behind this. Should we really worry about making digital sensors more prone to deep-red wavelengths to satisfy an arbitrary interpretation of what can be considered “healthy”, “pleasing”, and “natural” skin tone – terms that themselves are inherently subjective, and disregard an entire era of colour motion picture film when they did not appear on anyone’s radar?

The most satisfying approach is testing yourself with some talent rather than shooting bushes in your backyard (like I’ve done before).

...because IR wasn't a problem when shooting on film...

Thank you for detailing your experience and thesis, John. I find it informative.

It’s clear that increasing the light sensitivity of digital sensors came at the cost of making them susceptible to IR.

I'm not sure that the spectral response and IR sensitivity of digital vs film has much to do with increasing sensitivity of digital sensors. Those aren't directly related as far as I understand.

Remember most sensors are inherently luminance based only and the colour is mathematically calculated from a CFA that is fitted to the front of the sensor. The spectral response of the sensor is therefore derived from that CFA and the maths behind it.

However, the dissertation hypothesis here is that should we have sensors sensitive to wavelengths near 700 nm under some misconceived notion of achieving great skin tone?

If you look at the first IR filters, they tended to be very aggressive.

Here's an Alexa (which like BMD cameras already has an IR cut in front of the sensor) with an early IRND filter.

This terrible green bias used to be considered normal for an IR cut filter.
https://flickr.com/photos/johnbrawley/2 ... 844822085/

If not, then I posit that it might be preferable for cinema-camera manufacturers to limit the sensitivity of sensors between 400 and 670 nm, with the start at 400 nm and a sharp climb from 410 nm towards full perception, followed by a gradual fall-off starting at 650 nm and a stop at 670 nm, or perhaps at 680 nm. This may not only reduce IR prominence, but also minimize the need for IRNDs.

But you're not contemplating that the amount of IR in an environment isn't tied to the amount of visible light in an environment. It's tricky once you start trying to accomodate for both.

It is possible right now to order specific wavelength IR cut filters. Maybe you want to take that approach?
https://schneiderkreuznach.com/en/cine- ... rue-cut-ir
(note the different cut off points)

Schneider make a few that specific the CUT, but actually you'll notice none of them CUT, they ROLL OFF the transmission. Look at the spectral response curves. (these aren't the filters above but you get the idea)
https://schneiderkreuznach.com/en/indus ... protection

I don't think this is good practice either.

The better approach is what Arri and other are doing now, where the whole spectrum is cut by the same amount, rather than tuning in specifically on how much IR is cut relative to visible.

Objectively, Arri and Blackmagic both enjoy reputations of having "good" colour when the conditions are right. I think the thing to work out is how to protect yourself from these specific IR pollution issues in the shooting circumstances.

JB

John, thank you very much for this very well written and informative post! That's all I want to say, there really is nothing more to add...

Film stock is NOT sensitive to IR in the same way that digital sensors are.

ND filters didn't have to filter for IR, because IR wasn't a problem when shooting on film.

[...]

JB

I'm pretty sure 90% of the time the solution they have is no better than the internal ND's.

I'm pretty sure this is a generous estimate.

I'm not sure that the spectral response and IR sensitivity of digital vs film has much to do with increasing sensitivity of digital sensors. Those aren't directly related as far as I understand.

Remember most sensors are inherently luminance based only and the colour is mathematically calculated from a CFA that is fitted to the front of the sensor. The spectral response of the sensor is therefore derived from that CFA and the maths behind it.

To my knowledge, the demosaicing derived from the color-filter array is based on the sensor’s wavelength response – the luminance. By that extension, spectral response, light sensitivity, and IR susceptibility tend to be related.

Here's an Alexa (which like BMD cameras already has an IR cut in front of the sensor) with an early IRND filter.

This terrible green bias used to be considered normal for an IR cut filter.
https://flickr.com/photos/johnbrawley/2 ... 844822085/

I’m afraid I disagree – I don’t find it as terrible as a magenta bias, but of course, this is subjective.

But you're not contemplating that the amount of IR in an environment isn't tied to the amount of visible light in an environment. It's tricky once you start trying to accomodate for both.

Not sure I follow. From a wavelength perspective, IR and visible light are linked.

Anyway, this discussion has been fascinating. Thanks again, John.

And let’s face it – in the grand scheme of things, IR pollution is not a huge hurdle to creativity. And not even peripheral to an average audience member.

I find it amusing because everyone that's posting "problems" with their internal ND is rarely bothering to show what ND's they compare them to, because I'm pretty sure 90% of the time the solution they have is no better than the internal ND's.

Great info. .

At least the cheaper NDs like Nisi, Firecrest & Haida have some correction that's needed but could also be used to more creative ends.

I'm personally finding with the Pocket 6K Pro I'm only subtracting about 10 Magenta, either in camera or in the RAW tab, to match the 6 stops internal ND to the clear setting. Maybe also taking out 100K of blue from the 6 stop images. They're pretty neutral IMHO without taking into account IR.

I’m afraid I disagree – I don’t find it as terrible as a magenta bias, but of course, this is subjective.

And let’s face it – in the grand scheme of things, IR pollution is not a huge hurdle to creativity. And not even peripheral to an average audience member.

Here's a preproduction shot from an ad where I had some idea that the black coats would have some IR, and I used a look that emphasized green from the Firecrest ND and SLR Magic 50mm APO. Shooting into the light I pushed the greenish flare over her chin and the coat. The green also bleeds into the hat, but I think her skin tone still holds an interesting color vs the background.

URSA 4.6K SLR Magic 50mm APO 6500K

ursa_46k_hoya_ir_cut_slr_magic50_firecrest_12_1.7.1.jpg (830.37 KiB) Viewed 8210 times

To my knowledge, the demosaicing derived from the color-filter array is based on the sensor’s wavelength response – the luminance. By that extension, spectral response, light sensitivity, and IR susceptibility tend to be related.

Sensitivity is driven more by things like the way the circuits are gained, by other techniques like mixed amplification circuits, pixel size. None of these affect IR sensitivity per se, but they all make the sensor more sensitive.

I’m afraid I disagree – I don’t find it as terrible as a magenta bias, but of course, this is subjective.

I won't call it magentas bias. It's got a tint of magenta but actually what is happening shouldn't be confused with the green magenta swing on WB. IR pollution does can have a magenta tint for sure, but it's worse than that because different objects are responding in different ways to the amount of IR present in the scene. Colour reproduction is different. In the green filter example there's just a green wash over the scene and an absence of near visible red / infrared.

But you're not contemplating that the amount of IR in an environment isn't tied to the amount of visible light in an environment. It's tricky once you start trying to accomodate for both.

Not sure I follow. From a wavelength perspective, IR and visible light are linked.

Not the amount.

I can create a scene on the stage that is lit to say T22 @ 800 ISO and it will contain a lot of IR.

If I go outside on the same day and the amount of light also means I expose at T22 @ 800 but with less IR pollution...or more.

Same amount of "light"

The one on stage though will have a lot of IR because I'm using a tungsten source. Daylight won't always have the same amount, and it will change with cloud cover too. It's the same exposure, different amount of IR light.

JB

John thanks for your ongoing contributions to this and many other threads. Extremely informative - I’ve learnt a lot.


Sent from my iPad using Tapatalk

Brawley strikes again...

Keep it coming, mate...

...keep it coming.

Love productive threads!!!!

Here are some quite unscientific tests I did on my P6K Pro ND filters, comparing to IRND Firecrest ND filters (not vari-ND). Test methodology was to use shutter speed (and frame rate in one case) to cover the 2 stop difference. I did a tiny bit of tweaking of offset in Resolve to get them as close as I could in exposure, but no other changes at all. I've listed what I saw in the scopes under each frame, relative to the ND0 original:

(you can open the full res images in a separate window - I left them as PNG exports from Resolve)




Very slight push to magenta.


Slight blue push - same magenta push as ND2


Ever so slight more magenta and blue push than ND4.


Original image again - ND0


Very slight blue push and lifted shadows


Slightly elevated red and less blue than ND0. Again lifted shadows.


Slight green push and lifted shadows.


Significant green push and less blue - I had to drop to f4 to match exposure, so this may have contributed but the ND8 was significantly less accurate than any of the other filters.

Overall I'd say on my Pocket 6K Pro the internal ND is more accurate and easier to correct than the Firecrest IRND, and when you take into account the ease of use it's a no brainer.

I agree and I’m pleasantly surprised by how good the built-in IRND look. For the money, excellent value. A set of external IRND filters that are better will cost as much or more than the price of the entire BMPCC6K Pro.

With my NiSi (8 stops) and SLR Magic (10 stops), both have a greenish tint around either 6 or so stops but then get better again. Your Firecrest ND8 may be unacceptable for some shooters.

I tried at least a half dozen 5 and 6 stop 4X4 Firecrests and never found a copy that didn't have a green or cyan shift. The stacked 2 and 4 delivered consistently better results.

I'm only subtracting about 10 Magenta, either in camera or in the RAW tab, to match the 6 stops internal ND to the clear setting. Maybe also taking out 100K of blue from the 6 stop images

I'm seeing roughly the same in this footage uploaded in the other thread, the magenta shift tracks linearly with only a little blue flaring out the blacks. The P6KP NDs look far more consistent than any Firecrest 6 stop I've personally tested.

Good Luck

irnd.jpg (67.18 KiB) Viewed 7741 times

I did some testing as well, pictures can be found here:

https://drive.google.com/drive/folders/ ... sp=sharing

jpegs as well as a zip file with uncompressed images. Those images were taken with consistent aperture and changing shutter speed / frame rate to match exposure over 30 seconds or less on a pretty overcast day, so lighting didn't change much.
I applied the same Gen5 to Extended Video LUT by Blackmagic to all of them, nothing more is done to any.

I found my results to be pretty similar to yours, Ian. Really not bad at all for a camera in this price segment.

Regarding the title of this thread, I didn't do a specific test for IR contamination. Unfortunately my unit suffered from some dust on the 2stop ND glass and I had so send it back today so I can't do those tests anytime soon.

Sascha, great to see those tests and hear your results, though I'm sorry to also hear you had to return your camera.

And Ian and Howard, great to see you're also confirming what I think some of us have always known about how neutral ND's really are(nt).

JB

I did some testing as well, pictures can be found here:
I found my results to be pretty similar to yours, Ian. Really not bad at all for a camera in this price segment.

Regarding the title of this thread, I didn't do a specific test for IR contamination. Unfortunately my unit suffered from some dust on the 2stop ND glass and I had so send it back today so I can't do those tests anytime soon.

Like you, I didn't find much colour shift between the different stops of internal ND. Also, like you I had to send my unit back because I had some particles on the glass on clear and 4 stops. I'm hoping to get a replacement unit soon.