Varicon & Panaflasher - analog noise reduction and curve control!

[Karim D. Ghantous]

Some time ago I recall reading about the Panavision Panaflasher - a light source placed in front of the lens to lift the base fog level, that works with both film and digital. It's like a low con but only for the shadows, and it leaves the mid tones and highlights effectively untouched. It's basically real-time flashing, as opposed to pre-flashing which would occasionally be done on film rolls before loading them.

I thought it sounded like a decent enough idea but didn't think of it much. That is, until I read a piece featured the DP for Little Women (2019), Yorick Le Saux:

https://www.kodak.com/AU/en/motion/Blog/Blog_Post/?contentId=4295015369

Little Women was shot on 35mm, Cooke primes and Angenieux zooms, and one of the tools Le Saux used was an ARRI Varicon:

For scenes taking place during earlier time periods in the script, Le Saux wanted to encourage a sense of pastel colors, subdue the contrast and reveal shadow details, appropriate to the film's retrospective mood. This saw him frequently flash the negative – a technique pioneered by such legendary cinematographers as Freddie Young OBE BSC and Vilmos Zsigmond HSC ASC. For this he deployed an ARRI Varicon at different intensities with a 1/2 CTB gel filter, sometimes with an 1/8 SFX or CS diffusion filter in addition. To further cultivate the effect of flashing via the Varicon, the footage was push-processed by one stop at the lab.

Okay, pretty straightforward stuff... but then I did a quick search for more information, suspecting that the Varicon was a similar tool to the Panaflasher, and found this blog by DP Rob Draper:

http://www.robdraperacs.com/2011/11/varicon-creativity-and-dynamic-range/

He writes:

I could never understand why more DP’s, especially those shooting for TV, did not use the Varicon. The advantages were obvious.

And as I read this blog post, it all sunk in: duh! This is a fantastic tool! I dare say, an underused one. And the great thing is that it's arguably more useful for digital than for film - how about that?

Think about it: you're using something like a Sony A7 or a Panasonic GH5, and you're shooting compressed codecs. And you want to protect your highlights, so you might underexpose a little bit. The Varicon encourages you to do that, while lifting the light from the shadows (not really, but stay with me) to compensate for your underexposure. You then don't have to lift the shadows in post because they're already lifted. In fact in some cases you might want to pull your shadows back down. The benefit: cleaner shadows, no matter what you do with them.

Whereas with film, you have the ability to overexpose by one or two stops without too much of a penalty in the highlights. You don't really want to do that with a compressed codec.

So basically, the Varicon does two things: modify your tone curve, giving a quasi-log profile to the image; and it reduces noise while introducing no artefacts. It also seems to look better than low con filters (which I never liked but could see use for) while introducing zero halation.

I would only disagree with Draper's statement that the Varicon gives "approx 1-2 stops more detail in the shadow areas." I don't quite think that it works that way, as there is no free lunch here. However, the appearance of more shadow detail makes sense, as the noise is mostly gone.

A final note: the Monstro is such a good sensor that you probably would never need something like a Varicon. Those shadows are clean as the Queen's teaspoons. But it would be a very useful tool for hybrid cameras like the GH5 mentioned above, as well as for cameras with the MX sensor.

If you've used one, do let me know what you think of it.

 

[David Mullen ASC]

It's an interesting issue, even if shooting film today -- flashing fogs the image and affects the shadows more quickly than the highlights. The question is always whether the lowering in contrast is simply due to the lifting of the blacks (which usually get lowered back to black in color-correction) or an increase in usable shadow detail.

I flashed four 35mm features back before D.I.s (in the time frame of 1998-2002) but this was in conjunction with silver retention printing, so the flashing was a way of compensating for the loss of shadow detail / increase in contrast from leaving silver in the print, plus it helped reduce color saturation. My theory was that flashing the negative was like adding white paint into colored paint, and silver retention in the print was like adding black paint to colored paint -- both reduced saturation but did opposite things to contrast.

But I felt that in actual measurable increase in shadow detail as opposed to lifted blacks, I was not seeing a big improvement -- maybe 1/3 of a stop. Sort of depends on how you define usable shadow detail. Certainly the flashing lifted shadow information above the threshold where the contrast of the print stock would normally crush it. But with a D.I. where you can scan the entire dynamic range of the film negative and set the black level to whatever you wanted, not to mention saturation and contrast, I wasn't sure if the work of flashing was worth it anymore.

Mark Woods wrote a technical article for ICG Magazine back in the day comparing the D-Min detail from flashing using the Panaflasher (which happens before or after the image capture depending on whether the unit is mounted to the rear or top magazine port) versus the Varicon (which happens simultaneous to image capture since it is mounted in front of the lens), and his tests showed that there was a little more actual shadow detail increase from the Varicon device, at least in terms of the zone that the detail was lifted to.

As you said, digital cameras tend to have more latitude in the shadows (because there is less noise compared to the size of film grain) whereas film negative has more latitude in the highlights, so most people aren't fighting for shadow detail in digital photography. However, perhaps a Varicon would be useful in terms of allowing the highlights to be underexposed more and the flashing used to preserve some shadow detail in digital. This is the principle behind the Schneider DigiCon filter -- shadows are lifted and highlights are suppressed.

But actual testing would have to be done to see if the improvement from using a Varicon was worth the hassle of using it. The classic ARRI VariCon fit into a 6x6 matte box, which may be a bit unwieldy for a GH5 user... I heard a rumor that a smaller VariCon was developed but I've never seen it.

 

[David Mullen ASC]

Panavision created their own version of the ARRI VariCon, again it fits into a 6x6 matte box:
https://www.youtube.com/watch?v=uw8YkpdSxrI

 

[Hrvoje Simic]

Think about it: you're using something like a Sony A7 or a Panasonic GH5, and you're shooting compressed codecs. And you want to protect your highlights, so you might underexpose a little bit. The Varicon encourages you to do that, while lifting the light from the shadows (not really, but stay with me) to compensate for your underexposure. You then don't have to lift the shadows in post because they're already lifted. In fact in some cases you might want to pull your shadows back down. The benefit: cleaner shadows, no matter what you do with them.

Lit, well exposed shadows = > you capture information in shadows

Lifted signal = > you get lifted shadows and loose contrast


Lifted shadow signal doesn't compensate for underexposure. Because you are not just exposing to any light, you are exposing to light bouncing of something you are shooting and if you don't capture that shadow detail, no filter or flash on top of the sensor is going to draw it back on.

 

[John Brawley]

And as I read this blog post, it all sunk in: duh! This is a fantastic tool! I dare say, an underused one. And the great thing is that it's arguably more useful for digital than for film - how about that?


Whereas with film, you have the ability to overexpose by one or two stops without too much of a penalty in the highlights. You don't really want to do that with a compressed codec.

So basically, the Varicon does two things: modify your tone curve, giving a quasi-log profile to the image; and it reduces noise while introducing no artefacts. It also seems to look better than low con filters (which I never liked but could see use for) while introducing zero halation.

I think the idea is seductive.

There was a theory that photochemically if you "activated" the photosensitive layers it was better all round, even the with the base density.

I think they are in reality like a low con filter. A slight "fogging" or lifting of the shadow areas, but it's not magic.

I tested the Panavisaion version of a varicon a few years ago (see the below link) which is LED based and variable colour. The original Arri varicon was huge. Basically a 6" production mattebox on steroids with a tungsten light source that has to be powered.

What I took away was, there wasn't anything that was happening in camera that I couldn't replicate in post AND it wasn't baked in. For a little while i thought it might be interesting to have the "look" for a certain part of the show be done with the Panafalsher.

At about 19:50 i start testing the Panafalsher, by dialing up the adjustments in shot.

From memory I tried colours and white, along with the OG varicon but didn't see any real advantage in terms of noise, extra DR etc. You'll notice in the "contrast only" versions below where it's "white" light, there does appear to be a variable fill affect, but it's deceptive in that it's very very hard to do it consistently and there wasn't any meaningful difference to what I could do in the grade with more control.

It's in the name. Vari CON, as in CONTRAST. Not DR.

https://vimeo.com/199521554

JB

 

[Hrvoje Simic]

What I took away was, there wasn't anything that was happening in camera that I couldn't replicate in post AND it wasn't baked in.

If that look is desired it is technically supeirior to achieve it optically.

Optically - lifted shadows, more bits used, no data manipulation needed

Post - less bits used for quantization, blacks closer to sensor's noise floor, additional image manipulation needed, on less data for the same effect

 

[John Brawley]

If that look is desired it is technically supeirior to achieve it optically.

Yes in theory.

But the problem is that it's impossible to do this shot to shot and be consistent. It's highly variable, depending on the lighting situation and content. It also changes when you move the camera around.

It's a little like ETTR. Great idea on a static non dynamic frame that stands on it's own. Once you have a dynamic moving camera in changig lighting conditions that has to match to other shots in a sequence exposed in different settings it becomes a nightmare to make it all match in a sequence.

Hence, my belief that there is little *meaningful* advantage to doing this in camera unless you can always control the conditions you're in. I usually don't have that luxury.

JB

 

[Hrvoje Simic]

Yes in theory.

But the problem is that it's impossible to do this shot to shot and be consistent. It's highly variable, depending on the lighting situation and content. It also changes when you move the camera around.

It's a little like ETTR. Great idea on a static non dynamic frame that stands on it's own. Once you have a dynamic moving camera in changig lighting conditions that has to match to other shots in a sequence exposed in different settings it becomes a nightmare to make it all match in a sequence.

Hence, my belief that there is little *meaningful* advantage to doing this in camera unless you can always control the conditions you're in. I usually don't have that luxury.

JB

Got it.
Good point.

 

[Ron Steinberg]

Matthew Libatique used this on A Star is Born, similar to the Arri Varicon but you can add colour to the shadow regions:

https://www.camtec.tv/new-specialty/2017/7/20/camtec-color-con

Looks like there's a 456 version.

 

[Karim D. Ghantous]

This is the principle behind the Schneider DigiCon filter -- shadows are lifted and highlights are suppressed.

Now that's interesting. I wonder how a passive filter can suppress highlights. It's easy to understand how it can lift shadows, as it uses the incoming light to cause that.

Panavision created their own version of the ARRI VariCon, again it fits into a 6x6 matte box:
https://www.youtube.com/watch?v=uw8YkpdSxrI

Now, that was interesting. Are all these things so darned big?

What I took away was, there wasn't anything that was happening in camera that I couldn't replicate in post AND it wasn't baked in.

Funny, but the way I see it, even after reading your take, I still like the idea! I think it's because I have to do less in post than otherwise. I'm not quite sure how RGB lights would help, but I'm sure someone has a good use for them.

but it's deceptive in that it's very very hard to do it consistently and there wasn't any meaningful difference to what I could do in the grade with more control.

Would you not set the device to an EV value? Let's say you measure T8 with a 180 degree shutter, which is roughly 1/50. So your EV is roughly 12, for the sake of argument. And let's say that you want the flasher to kick in when shadows go down two stops from the reading. So in that case, you'd set the flasher to T8 and 1/250, which is EV 14 (at least I am sure that's how you'd do it). Is that how you use it? Is there an EV setting? [Edit: I messed that up. It should have been T8 and 1/12 which is EV 10, which is two stops less than T8 and 1/50].

Thanks for the video, too, that is very helpful.

Matthew Libatique used this on A Star is Born, similar to the Arri Varicon but you can add colour to the shadow regions:

https://www.camtec.tv/new-specialty/2017/7/20/camtec-color-con

Looks like there's a 456 version.