"Sensor Flares" - what's the truth?

Status
Not open for further replies.

Barry_Green

Moderator
Recently there's been a lot of buzz and discussion over the concept of "sensor flares" and, particularly, the AF100. A blog article claims that there's a big problem with the AF100 that causes unacceptable flares, and that article has been tweeted and retweeted over and over, and it's stirring up a lot of fear, uncertainty, and doubt. Links to the original article were posted here on DVXUser, but the OP's behavior in the thread led to the thread being deleted (the reasons for which, the OP kindly made public by screenshotting the report). It wasn't the subject matter that was a concern, it was the tone and the FUD that was the problem. Here's some of the phraseology that caused concern... the original blogger used terms including:
"obvious Panasonic design flaw"
"a source of HUGE disappointment and a deal breaker"
"This issue is very severe and will ruin a lot of otherwise beautiful shots"
"Panasonic, go back to the drawing board!"
"AF100 camera is a prototype!"
"released without thorough testing"
"It hurts that we can't use it with any lenses!"
"it is a DESIGN FLAW of the camera."

Wow. Strong words. You'd think the AF100 was out there destroying shots and kicking puppies. How simply horrible it must be, right?

Well, let's find out. I looked into the issue. My method of testing was to use an LED light, because when the flare shows up, it reflects the individual LED's and makes it very easy to see. Now, first of all, in order to test for this, you have to shine a very bright light directly into the lens -- the wisdom of which is suspect, but hey, I'm game, let's try it...

Well, there it is, in all its glory. Clear as day, reflected flare. So, should we execute the AF100 right now and be done with it? Seems like our friend the blogger thinks we should (after all, he says it's a prototype and needs to go back to the drawing board). But I wonder... is this an issue that's strictly limited to the AF100?

I looked around to see what other interchangeable-lens cameras I have. Hmm, there's the GH2, let's try that. Blare the light directly into it, and ...

ACK! There it is again! Oh no ... maybe all Panasonics are haunted! How terrible this day is, to find out this horrible truth, isn't it?

What else do I have that has interchangeable lenses? Well, look, there's my Super16 film camera, let's try that. Obviously that won't have the issue, because it's not made by Panasonic... okay, let's shine the light in and ...

Well, what do you know. There it is. The exact same lens flare. The same "sensor flare" is happening on a 16mm film camera. (No video to show it, because the viewfinder opening on the camera is about 1/4" in diameter and I couldn't figure out how to film the ground glass through that opening). But I assure you it's there and looks very similar to the AF100/GH2.

Clearly, this is a DESIGN FLAW of all movie cameras, right? Perhaps my 1970's-era movie camera is a prototype and needs to be sent back to the drawing board?

Fortunately for us, there are other brands we can turn to. For example, the Sony F3. That's a digital cinema camera, brand new, and costs 3x as much as the AF100. With such a higher pricetag (and none of those "incompetent" Panasonic engineers working on it) it'll be free from such an issue, right?

Right?

Oops. Have a look at Hello World Communications' gorgeous candlelight test of the F3.
Beautiful, rich, lush cinematography but WHAT -- wait -- at 00:45 to 00:52, what's that on the left? Yep, there it is -- the dreaded green ghosts! And Timur Civan, an F3 owner, has confirmed for me that he's had it happen on his F3 too.

Well, heavens to betsy and with much wringing hands, whatever are we to do? It must be a conspiracy by all those big Japanese companies, trying to stick us with inferior designs and "obvious flaws". I guess we have no choice but to turn to Red to free us from the dreaded "sensor flare"... so let's go look at some Red Epic footage...
Stephan Gray posted that Red Epic low-light footage, shot at 2000 ISO, and ... uh-oh. What's that at 00:32-00:35? It can't be! But it is -- it's the dreaded "sensor flare!"

Oh no! Recalls and lawsuits, whatever are we to do? Well, at least there's one place we can turn to. The blogger said that the flares were practically nonexistent and much-better-controlled on his 7D, so we can always just go back to using a DSLR, right?

Er, not so fast. Hold your horses and hang onto your lunch. Philip Johnston of HD Warrior posted a (much-shorter) rebuttal to the original blogger's article, and showed just how much of a sensor flare his Canon 5D Mark II has:
Green-ghost.jpg


Interestingly Philip said that before he got this ghost shot on the 5D, he also shot the same thing with a bunch of other cameras, including a Sony MC50, Canon G12, Panasonic G2 and Sony HX5V. All of them delivered the same "Green Ghost" issue.

So -- where do we stand? Here's where we stand -- it happens. It is not an "AF100 problem" or a "DESIGN FLAW" or anything of the sort. It's an artifact that happens when you point a very bright overexposed light directly into the lens. Of course, this is old hat to any shooter who's been around a while; we all know that you're not supposed to purposely point a bright light directly into the lens or you'll get flares (including flares off the back element). For those who are panicking about it, it makes me wonder if they've never shot on a CCD camera? In the older days, there was another (far worse) reason to avoid a bright light directly in your shot, and that's the vertical smear that a CCD would exhibit:

300px-Vertical_smear.jpg

(hotlinked from www.wikipedia.org's article on CCD cameras).

That artifact was far more glaring and noticeable and inescapable than a ghosted reflection, but it taught us not to point the camera into extremely bright light sources (and not to point extremely bright light sources into the lens!) With CMOS, the vertical smear doesn't happen anymore, so now maybe people are more inclined to point directly at a bright light source? I don't know, but it's just common cameraman technique to not do that, or -- if you do decide to do that, you accept the consequences: lens flares.

Look -- here's the bottom line -- if you want to go out of your way to find a flaw in a camera, you can do that. Every camera, every lens, every recording format has limitations. Is it good to know those limitations? Sure -- that way you won't run into the problem. But EVERY camera has its limitations.

Is it helpful to find a limitation and call it a "DESIGN FLAW" and a "dealbreaker" and then re-tweet it to everyone on earth trying to stir up pandemonium and scare people from using a valid product? Of course not. To do so is the very definition of FUD -- it's spreading Fear, Uncertainty, and Doubt. We're not big fans of FUD here. We are focused on exploring what these cameras do and then teaching people what to do, and what not to do, to get the best results (and here's the lesson for today, folks -- don't go blaring a bright light directly into your lens!)

So let's turn to practical considerations -- can you actually shoot with an AF100, with a bright light in the shot, without getting "sensor flares"? Of Course You Can. Here's a little example shot I threw together, it's handheld, it's not meant to be beautiful, but it's a very low light scenario (shot at f/1.4, lit with one 55w fluorescent tube) and there's an incandescent lightbulb in the background that is so hotly exposed that it's clearly clipping.
See any horrible ghosts? Neither do I. It looks fine, just like it's supposed to. And if I was shooting this for real, I would have put a dimmer on the background light to bring it down so it wasn't clipping so ugly. And every bit that you bring that background light down, lessens the chance of it flaring and causing an issue. But even in this scenario, where it's obviously too bright and is clearly clipping hard, it still worked just fine.

PostScript: by the way, something can be absolutely true and still be FUD -- it's all in the way it's presented. For an excellent example, I would like to point you to the following warning site about the supposedly "dangerous chemical" dihydrogen monoxide:
http://www.dhmo.org



That website is for the Dihydrogen Monoxide Research Organization. Read the site. See if that chemical doesn't scare you to death. Here are some of the things they say about it:
  • Some call Dihydrogen Monoxide the "Invisible Killer"
  • Others think dihydrogen monoxide should be Banned
  • Dihydrogen Monoxide is linked to gun violence
  • Dihydrogen monoxide was found at every recent school shooting
  • Athletes use DIHYDROGEN MONOXIDE, or DHMO, to enhance performance
  • Dihydrogen Monoxide has been found in our rivers, lakes, oceans and streams
  • Dihydrogen Monoxide is a major component of acid rain
  • Thousands die each year after inhaling dihydrogen monoxide
  • Dihydrogen Monoxide can be deadly
Is any of it true? Yes, it's ALL true. 100% true. Should it be banned? Well, seeing as dihydrogen monoxide is ... well, it's two hydrogen, and one oxygen, hmm, that's H2O... yep, dihydrogen monoxide is WATER. Yes, if you inhale it you'll die. Yes it's found in our rivers, lakes, oceans and streams. Yes it's "invisible". Yes it's a major component of acid rain. Yes it's all true. But the way the dmho portrays it, it spreads fear, uncertainty, and doubt.

Ain't gotta be a lie, to be FUD. FUD is FUD.

Bottom line -- all these cameras work just fine, and if you blare a harsh light into them all of them will give reflected flares. So don't do that. Keep direct light from shining onto your lens (which is what flags and matteboxes are all about). And if you have to have a light shining directly into the lens, tone it down as much as you can, using ND gels and stopping the iris down -- the more overexposed the light is, the more likely it is to cause the flaring. Really, this is Camera Basics 101, but I guess with new tools comes new techniques and maybe not everyone knows these things, so it bears repeating.

We welcome constructive discussion and when an issue is found, we welcome a constructive and productive dialogue towards how to overcome or avoid such an issue. Thanks to all those over the years who have given of their time and talent to explain things to our membership.
 
Last edited:
Barry,

Nice work, and nice that so many other people added to your research. In summary lens flares are the responsibility of the DP, and thus would be considered his job description to only allow the flares he wants into the frame. That isn't a design flaw, it is a job description.

Thanks for digging into this. Hope to offer my D700 flare frame tomorrow.

Best,

Jan
 
Well done Barry

Well done Barry

I have to say far too much homework Barry... the chap was only seeing what we all see under normal circumstances, it's the REVENGE of the DSLR brigade and the best they could come up with was a "haunted AF100 camera" with a green ghost. PS Just before I shot my 5D pic I tried a Sony MC50, Canon G12, Panasonic G2 and Sony HX5V all suffered the same "GREEN GHOST". We should start a competition to see if anyone can produce a "BARRY GREEN GHOST" from their DSLR/Camcorder !
 
Hi Barry, thanks.

I guess the issue with information being so quickly disseminated there is nothing to distinguish it from being verified.

Good read
 
I'm an editor, and even I know that is not a sensor issue. It's a reflection - clearly from within the lens elements. LEDs will be especially bad because they are high-intensity and have a narrow focal point with even higher intensity. Look directly at an LED - they are very bright, a little off axis and the brightness is much reduced.

No doubt the AF100 has real issues (what doesn't?) but this isn't one of them.
 
Yeah, but the girl in the F3 footage is way prettier than the mannequin in your's. And presentation matters.

Seriously, nice work Barry!!
 
Last edited:
What else do I have that has interchangeable lenses? Well, look, there's my Super16 film camera, let's try that. Obviously that won't have the issue, because it's not made by Panasonic... okay, let's shine the light in and ...

Well, what do you know. There it is. The exact same lens flare. The same "sensor flare" is happening on a 16mm film camera. (No video to show it, because the viewfinder opening on the camera is about 1/4" in diameter and I couldn't figure out how to film the ground glass through that opening). But I assure you it's there and looks very similar to the AF100/GH2.

It is not a regular lens flare but the flare in which the sensor's surface plays the critical role. The sensor reflects back some rays of light, acts like a mirror in a way. It is really not a big discovery, this phenomenon has been known for years. You mixed all kinds of flares all together here which only adds confusion (or uncertainty if you prefer).
There is a misunderstanding of what causes this particular phenomenon and the quote from your article above proves it.
You can't assume that Super 16mm film camera (or any other film camera) exhibits the same issue (as stated in your article) simply by looking through its viewfinder. What you see through its viewfinder does not include involvement of the surface of the emulsion ( which in case of a film camera is an equivalent of the sensor) instead it shows set of issues introduced by the viewfinder's optics itself combined with any possible lens issues. Even if the emulsion's surface was actually reflective enough to cause the problem, construction of the film cameras simply does not allow you to see what happens while the film is being exposed - whether it is a non-reflex or a reflex camera. It is completely different in most video cameras - you actually witness that moment "live". In a non-reflex camera we look through an independent optical viewfinder system which is not connected with actual lens-to-emulsion light path, it is obvious. In a reflex camera what we see through the viewfinder is not what is being exposed but its counter phase - the shutter is in a viewing position and closes the lens-to-emulsion light path: no light hits the emulsion at the moment of viewing (given that you keep your eye on the viewfinder and not allow any rays of light enter it "back door"). Furthermore, in reality what you see in the viewfinder of a reflex camera is not 100% the same as the image that is actually being captured by the camera- certain things seen in it are simply introduced by the viewfinder's optical system (i.e viewfinder's chromatic aberration, some additional flare, and obviously any frame markings). Yes, certain film movie cameras have issues with internal reflections (i.e ARRI 2C has a flicker problem, reflection coming from the ground glass, etc.), however, usually you can't evaluate them simply by looking into the viewfinder (often you would see some issues that are not being captured and vice versa). Once again, mixing this all together in your article means that you are talking about flare issues in general, not this particular type. It simply needs little bit more attention.
 
Last edited:
Clearly the film flare is an extreme example, although I do think it would be comical if such a flare actually did show up on film. I can't think of any examples, but I wouldn't be surprised, because film isn't a perfectly matte surface and if you're overloading such an incredibly bright light directly onto it, it may very well reflect back. I'll try to dig out some ground glass to see if anything shows up when removing the viewfinder and mirror from the equation. And, I'm testing a new S16 camera soon, so I'll run the LED hyper-bright flare test on it.

Regardless of whether it happens on film or not, the fact remains: all interchangeable-lens video and digital cinema cameras are going to be subject to this same type of flare. It isn't a design flaw, it's the way video cameras are made, and furthermore, it really isn't a big deal. The original blog tried to paint it as an AF100-specific issue, and that is patently false. Had the article been more along the lines of "interchangeable-lens video cameras may have more flaring than you're used to on a DSLR or fixed-lens video camera", that would have been fine; it was the incendiary, inflammatory nature of the article, designed to scare people, that caused me to delve into the level of research I did for this article. Reflected flares are nothing new. They are not the exact same as typical lens flares, no -- but the remedy for both is the same. People have had to deal with flares of all types for a long time. There is nothing new here, there is no camera fault, there is no design flaw, it will happen on any interchangeable-lens video camera if a bright-enough light is shined directly into the lens, period.

From what I understand, Fujinon and Angenieux have actually developed new coatings on the rear elements of their newest digital cinema zoom lenses to reduce this effect, specifically because it can and does happen on digi-cine cameras like the Red, the Epic, the F3, and the AF100. If you really had a circumstance where you absolutely had to shine a massively-overexposed light directly into the lens, and simultaneously not see a reflected flare, you should probably be looking at renting a lens that's made to handle that task.

For the rest of us, a little exposure control and a little composition control are enough, or more than enough, to render the issue moot. I give you an example of someone who just shot the AF100 entirely at night, doing nothing but pointing the camera directly into lights. This was a largely uncontrolled test, and I dare say it was pretty much a worst-case scenario.

Look at how many shots there are, and count how many times the "sensor flare" shows up.
First shot, buildings, black with bright lights, no flare/ghost.
Second shot, extremely bright headlights, yep it's there.
Third shot, lots and lots of bright windows, and an extremely bright spotlight on the top of the building. There's a little bit of the ghost as he pans right, but they likely most audiences would never notice it, and if they did, they'd probably chalk it up to "flare" because, when you point a light that bright into a lens, you've always gotten flare -- whether it's typical or "ghost", doesn't really matter, in that shot it just looks like a flare.
Fourth shot, lots of headlights, no flare, no problem.
Fifth shot, big spotlight, lots of windows, no flare, no problem.
Sixth shot, lots of bright headlights pointed directly into the lens, no flaring, no ghosting, no problem.
Seventh shot, "Mother Annas" sign, a big bright white light right in the lens, no problem, no flare.
Eighth shot, the clock tower -- no flaring.
Ninth shot, closeup of the clock tower, lots of bright lights -- no ghosting, no flaring.
Tenth shot, major headlights, super-bright streetlights, shop lights, all pointed in the lens -- no big ghosting at all. There's a little bit of halation around some of the super-bright streetlights, and there's a few little tiny specks of purple reflected flare, but -- this has got to be one of the worst-case scenarios ever, and I dare say no rational person would even notice it.
Last shot - mega streetlights and headlights, everywhere. Looks just like I would expect it to. Lots of halation and flare, but nothing out of the ordinary.

I reiterate -- this whole issue has been massively overblown, there's absolutely nothing out of the ordinary here, and flare control is something that any competent shooter would need to be aware of and manage, and you absolutely can get bright lights in your shot without seeing some big scary ghost.
 
Last edited:
I'll try to dig out some ground glass to see if anything shows up directly on the film.

If I understand correctly you want to put a ground glass on the film plane (in the place of the film) and look through it with a magnifying glass, correct? I appreciate the effort but this won't prove anything either because then it will be the ground glass' surface reflecting the light and causing another problem (it has a different reflective property than film's emulsion). The only way to test it is to shoot the actual test on film (I think I kinda know the results - no issue , but who knows).
 
Last edited:
And I will do so... it just takes a week to get film back, there's no labs nearby. I was saying that the ground glass test would eliminate the viewfinder from the equation. I suspect that with a bright enough light, it's possible that it will show a reflection. The film surface would have to be absolutely nonreflective in order for it to prevent any sort of flare, and just looking at film, it's clearly reflecting light, so it's possible. I'm not saying for sure that it will happen, but if it does happen, I'll find out.
 
since when did this lens flare actually became an issue?
the guy's first post was that article and he made a blog post specifically for the lens flare. its obvioys he was spreading FUD and looking for attention
 
Thanks Barry, I just bought 2 AF 100s and have been looking for a nice assortment of lens' when I came across this thread. I have been using the Panasonic 2000, HVX 200, 200A for years. No real problem with lens flare/ghosting etc. I just shot my first test this morning of a sunrise on the lakefront with the AF 100/Canon EF 16-35 & Kipon mount. I will be looking at it on a Panasonic 17" monitor tomorrow with my DP. Now, back to my search for a nice lens package. P.S. I was really upset when I read the post that started this thread. Thanks for the thorough follow up. GE
 
If I understand correctly you want to put a ground glass on the film plane (in the place of the film) and look through it with a magnifying glass, correct? I appreciate the effort but this won't prove anything either because then it will be the ground glass' surface reflecting the light and causing another problem (it has a different reflective property than film's emulsion). The only way to test it is to shoot the actual test on film (I think I kinda know the results - no issue , but who knows).

Regardless, take the film component out of it for a moment and weather it happens or not means little as it does happen on Red (I shot a heavily flared MV in Tokyo in January - flares were the point and its there) and F3 for example and a few quick tests show me I can get it easily on the 5DMKII. That being the case, and the fact we agree its there - whats the point you're trying to make?
 
Regardless, take the film component out of it for a moment and weather it happens or not means little as it does happen on Red (I shot a heavily flared MV in Tokyo in January - flares were the point and its there) and F3 for example and a few quick tests show me I can get it easily on the 5DMKII. That being the case, and the fact we agree its there - whats the point you're trying to make?

My point was that because Barry made reference to a film camera in his article it proved that he was mixing different types of flares all together and that he was talking about flare issues in general, not this particular type and that it needed more attention.
 
Last edited:
Watch any movie from the '70s and you will see lens flares galore. Vilmos Zsigmond turned them into an art form, and won an Academy Award for Best Achievement in Cinematography for a movie where lens flares were added to special effects shots to increase their realism ("Close Encounters of the Third Kind").
 
Status
Not open for further replies.
Back
Top