Cool DVX Noise Reduction Technique

louis,
if you use mattebox ,FORMATT AND TIFFEN DO A 4x4 FILTER THAT IS CC70M(that is a little more of the combination of chris).
i will use it.
this will be my combination of filter:
ALWAYS FILTERS:
schneider DIGICON 1/2 grade.(always in fron of my lens.i like this filter)
FORMATT CC70M (always in front of my lens....except when i need more light)

NOT ALWAYS FILTERS
POLA
CC magenta filter to use outside instead of ND filter(reduce light and reduce a little more the blue channel during a blu day)


best
ivan
 
DO A 4x4 FILTER THAT IS CC70M(that is a little more of the combination of chris).

Be careful there...

I said 60 ccM for a reason, 3200K light can`t be shot under safely using filtering greater than 60 ccM. If you do, you will make your red channel clip before green, and thats a bad thing. You want equal clipping with the green channel. If you shoot under lower color temperatures, you can probably only get away with 30 ccM or lower, due to the change in color temperature, there is too much red/orange light.

Hope that helps!
 
Cardmaverick,

Do you really do all this? Does it really make a noticable difference?

In 25 years of shooting I've never seen anyone do anything like this.
I don't notice any particular problems with my green clipping earlier than other colors, but maybe I'm just not observant. Well actually I do seem to remember some color in the DVX100 clips but is that there in the HVX - I haven't noticed it?

What in the world is all of this work supposed to correct.
It doesn't help with HVX noise in the blacks does it?

Frankly if you need this level of control I think you should just shoot with a more expensive camera. Maybe it makes a difference with Andromeda and the DVX that sounds reasonable, but for us ordinary folks out there, it sounds like one way to lose a client as you go nuts doing things that no one else cares about.

- Lenny Levy
 
Last edited:
Lenny,

Yes, I really do use this technique, every day in fact, and it really does help. I actually do even more "esoteric" stuff. I have templates for the major focal lengths that I shoot with, which correct for the lenses lateral chromatic aberrations, AKA "color fringing". Correcting for the aberrations is VERY noticeable, and beneficial.

Getting back on track, the reason you don't see green clip first is because of the way your cameras DSP is setup. They purposefully DON'T let you see that this is happening. If you bought a camera and saw magenta highlights, you probably wouldn't be too thrilled (Red One has the magenta highlight issue, but they provide a user friendly way to correct it in Red Alert! and Red Cine).

The reason why you've never seen anyone do this probably has a lot to do with the fact that most shooters never knew about this in the first place. I suspect that until the Viper came around, most DP's were totally unaware of this.

This trick lowers gain in the red and blue channels, thus creating a cleaner overall image. It also increases usable highlight dynamic range. The easiest way to do this is simply put on a 60ccM filter, manually white balance and leave it on. It's super simple.

More expensive cameras have this same issue, even if they have bigger better sensors, you can still force them to perform better using the technique.

My clients don't care about what I do, they care about what I produce. Like anyone else, I have to make concessions from time to time, but by being highly educated about creating images, I can make better concessions.

Hope that helps!
 
cardmaverick - The easiest way to do this is simply put on a 60ccM filter, manually white balance and leave it on. It's super simple.

I'm really very much interested in this super-simple technique of yours. I'm still confused a lot getting the technical insight but I really want to apply this great trick to resolve magenta-higlight issue and increase the overall dynamic range of my DVX100B footage.

Please explain in easy Step-by-Step way in plain english for End-User like me :D. I have a bare DVX100B and no such filter what so ever. Do give me hint of that auxiliary 60ccM filter that you mentioned. Also, do i need any kind of CTB/CTO gels on the lighting etc.

Thanks in advance.
 
This is a curious technique. It would be nice to see some before and after examples (full-res stills that we can bring into Photoshop/AE and look at the color channels), before I drop some money on filters to experiment with this myself.
 
This little video illustrates what I'm talking about, watch as this blown out shot of some cloth suddenly fills with textural detail, and turns green. All I'm doing is going from a manual white balance, to a custom "unity balance" (all three channels balanced evenly). I shot this under 3200K light, no optical correction filters were used. Also note that this was shot on DV tape not the Andromeda....

http://www.chrisnuzzaco.com/special_uploads/dvrange.mov

NOTE:

I'm just starting to test the magenta filter "trick" while shooting to DV... Not ready to post too much on it yet, but it might take a while, as I now have to deal with that pesky DSP....
 
Allow me to be the devil's advocate here...

Adding a strogn Magenta filter does two things. Mainly, it significantly cuts green light. Also, as a side effect, it does cut overall light. Looking at the available CC60M filters, they have a "filter factor" of 1 to 1 1/3 stops, so they cut all light somewhat, though the intent of your research is to cut green light.

I would think that the effect of shooting through such a filter would be not to decrease the channel gain on red and blue, but rather, for the same shot and exposure, significantly increase the green channel gain. In fact, due to the strong filter factor, even the red and blue channel gain is raised somewhat, while green is significantly raised.

Am I way off base here? Is this not absolute, but rather relative gain, so that given the magenta filter, the camera splits the difference, raising green channel halfway and lowering red and blue channel gain halfway, thus splitting the difference?

The fact that simply switching from a manual white balance to the unity balance shows up more detail without having to alter the exposure would seem to indicate that the red and blue channel gain is lowered, as you hypothesize. Yet, it doesn't make much sense. Is the green channel gain perhaps constant, and only the red and blue channel gain is adjusted relative to it?

Altering the green channel gain in any way seems to perhaps have another side effect, since the Luma porition of the DV image is primarily derived from the green channel, could this result in increased Luma noise? I guess only so if the green channel is not constant.

All of this is very interesting to me. Unfortunately, it looks like a CC60M filter would run me about $150-300. A little too costly to play around with.
 
The green channel is constant and never changes. This was one of the first things I verified using the Andromeda histogram.

The reason why the image looks green is because there is more green content in real world images. Its hard to for people to twist there head around this concept, it took me a long time to understand this. Because there is an excess in green, you need to pull it back. Yes, any filter reduces light in each channel, but a colored filter will reduce even more light in some channels and not others. Remember, you're not dealing with an ND filter here. If you are shooting with the unity balance, you will also notice that once proper filtering is in place, the camera is balanced for 12,000 Kelvin, so from there you need additional CTB filters.

This technique is probably most useful for outdoor shoots. Even if you can only use the partial filtering technique (which is what I do), you still benefit from it, as for how the cameras DSP reacts to this, I'm still looking into that, but clearly from a dynamic range point of view, it looks to be beneficial.
 
One more point I should add...

I know that my channels are at unity balance because they all clip at the same point on the histogram, and this can be verified by seeing fixed pattern noise in each channel (which only appears when a channel truly blows out or "saturates fully"). The interesting thing with this though, they don't all blow out at the same instant in time... Green is always first, so this supports the theory that there is more green content in the scene. Color temperature also affects this as well, but so far I haven't seen the green channel not blow out first.
 
Fascinating.

You still need CTB filters even when shooting through the Magenta filter using unity gain? CTB on the lens or on the lights?

Is auto-exposure still accurate when using unity gain? I don't leave iris on AUTO, but I do like to use auto to get in the ballpark, then adjust to taste.
 
You would put the CTB on the lens, but you could also gel the lights. Its balance is pretty darn high up there, 12,000 Kelvin, so even under typical daylight, you need to raise the color temperature in some way. I think using an on camera filter would be the easiest way to go. Also, the higher your color temperature, the more magenta you can use. Getting the camera to be totally "optimal" is a bit crazy filter wise, as you need about 150 ccM plus "x" number of CTB (depending on the color temperature your working under). This is why I use my partial filtering technique.

After thinking about this for a while, I can't help but wonder if higher end cameras like the Red One have special ND filters over the green photosites.... It certainly helps that the camera is balanced for a more user friendly color temperature (5000K). Just a theory!
 
I'm sorry, is it 60 or 150 ccM to achieve unity gain?

So, if one CTB cools 2400k (difference between 3200k and 5600k), you need 2.5 to 3 CTB under daylight, and an extra under tungsten lights. Really doesn't matter if you put it on the lights of camera, the stop loss is the same either way. You need a ton of light for this.

Man, the glass filters are expensive for this stuff, but lighting gels are cheap. You can get a Double CTB and 3-stop Magenta sheet for $12 and make your own filter. :)
 
Back
Top