Blue Noise in Shadows

David Cunningham · May 6, 2013

What causes them? I only see it in underexposed negative films, but I've seen it on multiple scanners. What it is and why does it happen? Let me know if you need me to post example images showing what I mean, but it's so prevalent in all scans I've had done, it must be a known issue with underexposed film. I'm just really curious what causes it and how best to deal with it.

Robert Houllahan · May 6, 2013

Blue information tends to be the most noisy and least sensitive on both film and digital capture. I think it shows up a bit less on digital because a Bayer mask only has 1/4 of all of it's recorded information with blue filters.

With film stock the blue layer tends to get grainiest the fastest, and the tendency when coloring it is to stretch the underexposed blue layer out to balance color which reveals allot more of the D-Min grain. With a telecine (YUV Color) there is a bit of crosstalk which makes stretching the blue harder. With a data (RGB) scan you can easily do a bit of custom curve work on the blue and individual channel NR to suppress underexposed grain without messing up color balance.

-ROb-

David Cunningham · May 6, 2013

Blue information tends to be the most noisy and least sensitive on both film and digital capture. I think it shows up a bit less on digital because a Bayer mask only has 1/4 of all of it's recorded information with blue filters.

With film stock the blue layer tends to get grainiest the fastest, and the tendency when coloring it is to stretch the underexposed blue layer out to balance color which reveals allot more of the D-Min grain. With a telecine (YUV Color) there is a bit of crosstalk which makes stretching the blue harder. With a data (RGB) scan you can easily do a bit of custom curve work on the blue and individual channel NR to suppress underexposed grain without messing up color balance.

-ROb-

Hey Rob,

The most recent scan you did for me is a prime example. Not sure if you still have access to it. But, the darker highlight areas are full of blue lightning like lines. I went back to some other scans I have from underexposed 500T and it looks like I have the same thing on the LaserGraphics ScanStation (a 2K area CCD) and Y-Front. The only ones I cannot see this same issue with are the Spirit at Spectra Film and Video and the Millennium II at Pro8mm.

I have other underexposed negative scans on other stocks... but they do not have this noise. For example, I have some Fuji 64D (from the Birthday part scan you just sent me back). Neither your scanner, the Spirit or the LaserGraphics Director show this noise in the underexposed highlight areas.

So, it looks like it's something to do with underexposed Vision3 500T with area CCD sensors. I wonder if it would be the same deal with an Arriscan or LaserGraphics Director.

Robert Houllahan · May 6, 2013

Huh, was that Super-8mm? I don't think I have that file anymore but if you want me to try to rescan it and go a bit stronger on the NR feel free to send it back, we won't be doing 8mm data untill the servo based Xena is running, the 3.4K CCD is on it;s way though...

-Rob-

David Cunningham · May 6, 2013

Hi Rob,

The scan you did was 16mm 500T on your Xena.

All the other scans I see this on are Super 8 500T done at other locations.

Here is a zoomed in frame from your Xena 500T scan of an underexposed dark area. It's not as obvious when they aren't moving. But, you can see the little blue lines. This is similar to (but no where near as drastic as) the old home movie reversals you did for me a while back.

Robert Houllahan · May 6, 2013

I still have those files, it may be that the stock was a bit funny, I can do another pass from the DPX frames to ProRes444 and this time add a curve and some noise reduction, probably can get it out tomorrow.

-Rob-

David Cunningham · May 6, 2013

Hey Rob,

Not really a complaint. Don't want you to think it that way. I'm mostly asking the question because I see it frequently and I wonder what it is. I only ever see it with underexposed 500T for some reason.

Those films weren't really good enough to go through the trouble of grading, etc. The only clips I'm actually going to use don't have the problem because they are correctly exposed anyhow. So, no worries about doing another grading pass. If I have the noise in any of my clips I choose to use I'm sure I'll be able to "de-noise" it out.

I'll keep you posted if I change my mind.

But, why it happens is what I'm most interested in.

Dave

David Cunningham · May 6, 2013

Here is an example of 500T from the LaserGraphics ScanStation with the same problem:

http://skipper.mecnet.net/~macgyver/blah/blue2.tiff

David Cunningham · May 6, 2013

And I'm just curious why I've never seen this sort of thing with a multi-line-CCD system like the Spirit.

Robert Houllahan · May 6, 2013

Because they are not as good at resolving what is on the film? The LG Scan Station is a color camera, where the Xena is a B&W camera (Like the Arri or the Director) so the image is made from a parked frame of film (on pins) and the exact same imager (the area sensor) the only thing which is different is that the LED lamp fires different colors (and IR) to make RGB. I will ask the guys in LA about this and see what they say.

-ROb-

David Cunningham · May 7, 2013

So. Reviewing more of my films I've found that not only does the Millenium and Yfront have the same problem but its way worse!

Also found some light examples of it with the Spirit.

In all cases it's 500T. I never see it with 50D and only very lightly and occasionally with 250D.

I don't see it at all in my Fuji 64D clips even though some have some very dark highlights. That's been scanned in your Zena, Metropost Director and Spectra spirit. All look about the same with the edge probably going to the director at this point. But. That might just be the raw vs graded status.

Frank Vine · May 10, 2013

If you would like me to scan some frames with my adjustable lighting system and want to send me just a short strip, then drop me a PM.

David Mullen ASC · May 10, 2013

Well, 500T is not only very fast and thus grainy, but also has the fastest (grainiest) blue layer compared to a daylight-balanced stock. Add to that, it's very sensitive so it ages faster and accumulates fog damage from x-rays over a shorter time.

David Cunningham · May 11, 2013

Well, 500T is not only very fast and thus grainy, but also has the fastest (grainiest) blue layer compared to a daylight-balanced stock. Add to that, it's very sensitive so it ages faster and accumulates fog damage from x-rays over a shorter time.

All that is true. But why do these blue artifacts show up in dark scans with 500T? Is that "data" really what's there? I've reviewed all my 500T scans and its at least noticeable if not extreme at least once on every roll with every scanner I've used. It becomes rare with 250D and I have yet to see it on 50D (Vision 2 or 3) or Fuji 64D.

Robert Houllahan · May 11, 2013

This is a roll of 7219 that I had bought fresh from Kodak two days before I shot this and developed it the day after the shoot.

I would say that the data is definitely there I cannot see how the scanner would introduce that as an artifact, the Xena has a 1-tap area sensor with 10micron pixels and the LEDs used in the illumination system are the very best available currently. I think that it may be a component of what makes Kodak's 500t stocks fast, and a data scan does show everything on the film but can be noise reduced in a post step to clean up the grain a bit.

-Rob-

David Cunningham · May 12, 2013

This is a roll of 7219 that I had bought fresh from Kodak two days before I shot this and developed it the day after the shoot.

I would say that the data is definitely there I cannot see how the scanner would introduce that as an artifact, the Xena has a 1-tap area sensor with 10micron pixels and the LEDs used in the illumination system are the very best available currently. I think that it may be a component of what makes Kodak's 500t stocks fast, and a data scan does show everything on the film but can be noise reduced in a post step to clean up the grain a bit.

-Rob-

Hey Rob,

Yeah, it's definitely more subtle in your 500T footage, but it's definitely there and it's definitely apparent in every scan of 500T I've ever done. I'm just not sure why I never noticed it before.... or maybe I did and didn't think much of it. I just find it interesting that it's pretty much non-existent in the 50D and 64D I've had scanned. There must be something about the un/under exposed blue layer in 500T.

Dave

David Cunningham · May 12, 2013

Also, I believe the blue layer is the top most layer on the emulsion side. So, assuming your viewing/scanning the emulsion side, flashing light through the celluloid side would mean that all the light passes through the blue layer last. If it's black or near black then the negative is going to be clear or almost clear... sooo... I don't know what I'm getting at, but there ya go. :)

Dave

Chris Burke · May 13, 2013

I don't think it is on the negative, it may just be that the scanner is attempting to "see" into very dark areas, where there isn't much detail and finds an odd bit or two that is blue. With a proper grade, it can disappear completely. I have found it on most 500T I use when it is dark or underexposed. 7219 the least of all. Like I said before, with the proper settings, it goes away. I have shot a good deal of Eterna 500T outdoors this winter and never any blue in the shadows, indoors in a dark scene, plenty, but nothing that couldn't be corrected. I really wouldn't worry about it. I will encode some I shot indoors April 15th and post it in a bit.