What’s That? (59)

EX-S500

ƒ/2.7

6.2 mm

1/60

From the Puzzle Series: How did the door lattice image get on the front of his sweater arm? For this puzzle though, I’m not certain that I have the answer… Perhaps it’s a glitch in the Matrix…. =)

Context: this is a simple auto-flash photo of a kid jumping in the air. The kid is a 2-3 feet in front of the lattice window double door. There are no other lattice surfaces, or weird lights or shadows or mirrors or reflective surfaces. It’s just a normal room with no nearby walls to the left or right. It was not a planned shot in any way.

52 responses to “What’s That? (59)”

The Rocketeer

Dec 21, 2005 at 12:05 am

I believe the light from the flash is reflecting off the window onto the subject and because there isn’t as much reflection from the frame of the door, it leaves a darker band. You can see light on the back of his shirt also (supporting the theory).

Although your matrix glitch idea is pretty solid too…

Reply
angry shock

Dec 21, 2005 at 12:14 am

First: Hi to the young model posing for the unexpected puzzle. =)

I wonder if your camera lens works with interpolation of images? In that case, -I am no expert- I imagine that at the moment of making the juxtaposition something went wrong with the data. It looks like two pictures intending to be overlapped as one.

And I agree, the flash probably is the reason for it.

¿?

Reply
jurvetson

Dec 21, 2005 at 12:16 am

great. This is exactly where the homeowner and I started in talking about this as well…

Rocketeer: If so, then how does the reflection off the window get onto the front of the sweater? And why does it align vertically and horizontally with the background grid?

Alieness: I also was wondering if there are digital camera artifacts or interpolation bugs that can explain this (but it’s a really big effect as opposed to a few pixels…). I did not set up a double exposure…

Reply
The Rocketeer

Dec 21, 2005 at 12:22 am

It could be that the material that the shirt is made of reflects light differently depending on the incidence or angle of the light. A flash firing directly into the shirt isn’t reflecting as much light as light seemingly coming from behind him.

Alternatively, the boy’s quick movement in the frame was frozen by the flash, but the exposure was 1/60th of a second, giving him enough time to move within the frame during exposure. That means the framed windows were possibly already exposed onto the CCD prior to the flash going off and what you are seeing are artifacts of that exposure. (almost like a double exposure as Gisele mentioned)

To support my first idea… I would think that the light from the flash would need to come from a higher angle, possibly reflecting from the ceiling first and then onto the glass and then onto the shirt (minus the bars where the frame is on the door). It would be hard to duplicate, but I think it is mostly chance that the frame looks like it is continued on through the shirt. The horizontal portion on the shirt is the shadow of the frame above it, not the one directly behind it.

Reply
angry shock

Dec 21, 2005 at 12:25 am

But, I don´t see a simple reflection of the frame door… it rather looks like the boy´s arm is transparent(?!) to me.

Now, another idea… Did you see that for television studios setup they use a blue background which is the one the lens of the camera recognizes to replace with another picture or movie selected by the director? I know they use blue for a reason, it is not justa mere election… Could the blue sweater be inducing this camera craziness?

Reply
angry shock

Dec 21, 2005 at 12:26 am

-having just read Kev´s late comment-

Hahahaa… =D kev, we are quite on the same track… you say material, I say color =)

Reply
rtolmach

Dec 21, 2005 at 12:50 am

It’s actually a sweater with a gridded pattern. The question is how the grid got onto the single-pane glass door.

Reply
rtolmach

Dec 21, 2005 at 1:02 am

We’re not really seeing the grid shadow on the front of his sweater, but rather on upper and side surfaces. Look how rounded his arm is, how the upper surface is lit, how his side is also lit. Note how the grid lines disappear into the shadow as the arm curves away from the light source and on his side, just above his hip.

The light on the top of his arm and on his side indicate a light source from above and behind. So either there’s a strong light outside the door and the sweater lines are the shadow of the mullions, or – more likely — the flash is reflecting off the ceiling, off the glass door (except less so off the mullions), and back onto the upper arm and side. Since there are just two vertical mullions, the two vertical lines on his arm must be shadows of the mullions directly behind them. The horizontal shadow could be the shadow of a higher horizontal mullion.

Then again, it might be that matrix thing. Was I supposed to take the red pill or the blue? In any event, be sure to see http://www.themeatrix.com

Reply
jurvetson

Dec 21, 2005 at 1:02 am

it’s the question that drives us

the sweater is a normal, opaque one.

As for motion during the 1/60sec exposure, shouldn’t it appear on the sock as well? Looks like 3" of overlap… implying a movement rate of 15 ft./second or 10 miles per hour…. and it was a simple hop… =)

Rocketeer: I think you are on to a key point with your first guess – about coincidental grid alignment. I added a note on the arm about a troubling detail though….

Reply
angry shock

Dec 21, 2005 at 1:03 am

(to rtolmach) I *like* the way you think.

Reply
biotron

Dec 21, 2005 at 1:04 am

heh heh 🙂

i would be tempted to go for Rocketeer’s alternative explanation above, that of brief prior exposure of the frame… but even then, it ain’t common! hmmmmm

Reply
jurvetson

Dec 21, 2005 at 1:07 am

Oh… just saw that rtolmach and I posted at the same time… taking a closer look now…

Reply
angry shock

Dec 21, 2005 at 1:10 am

Steve, was there a moon or something alike out there? It was full moon that day.

Reply
jurvetson

Dec 21, 2005 at 1:11 am

nice new angle, so to speak…. (the EXIF data is off by one hour – photo was at 4:47pm)

For the "ceiling flash bounce" theory, do you think it is possible that a pocket camera’s weak flash can bounce off a white ceiling and the back glass and still show up superimposed on the sweater that is receiving a direct hit of the flash?

For the "backlight shadow" theory, wouldn’t we expect to see a vertical shadow on the sock as well?

Reply
jurvetson

Dec 21, 2005 at 1:23 am

Oh, and since I don’t know the answer, this will have to be a communal Bingo determination…

I’d nominate rtolmach’s "backlight shadow theory" as the post plausible so far… with the light source just out of view behind the outdoor awning.

Reply
Leino88

Dec 21, 2005 at 1:25 am

it’s how you’re camera balanced the ambient light with the flash output….. 1/60 sec is not enough to fully freeze Leif’s arm wich was moving faster than his body…(flapping his arms, or something)…hence it’s motion blur allowing some of the exposure off the door to "seem" like it’s coming through.
did that make any sense?
i see it all the time…different cicumstances, but common effect… anyways…
gotta run…my cd’s burnt, and i’m outta here to the lab! (i love the cryptic sound of that!)

happy holidays everyone!
ps steve….card’s in the mail!
😛

Reply
wallofhair

Dec 21, 2005 at 1:27 am

I keep thinking maybe it’s some kind of spot exposure autocorrection that the camera is trying to do; that it’s trying to keep bright spots of the image from blowing out/clipping when it fires the flash.

This explanation doesn’t fit perfectly: why is there a stripe on the kid’s hand where the darker door inset is, but nothing where the brighter part of the door is? Maybe the camera really does limit the spot adjustments to smaller regions.

Reply
biotron

Dec 21, 2005 at 1:29 am

there would have to have been direct opposition between camera and moon to result in that alignment of shadow, Gisela, surely?

any bright light source from outside would surely cast something visible on the floor from the frame, too…? so i don’t like the BS theory – it’s BS 😉

the only reason i can think of these shadows not appearing on the sock would be to do with the reflective potential of the fibres in the respective materials.

in response to the CFB theory question ( as it has quickly become the vogue…) surely the direct illumination from the flash would cancel any reflecting component?

i have to drive 600 miles tomorrow, beginning in 8 hours, and i’m in danger of not getting enough sleep! thanks! 🙂

Reply
The Rocketeer

Dec 21, 2005 at 1:31 am

Isn’t the backlight shadow theory what I said? I thought I was reading my own answer in rtolmach’s explanation.

Reply
biotron

Dec 21, 2005 at 1:33 am

it’s what you said first, yep! i agree with yer "alternatively" / Lamo’s MB theory…

Reply
AMagill

Dec 21, 2005 at 1:34 am

I’m with the Rocketeer’s second idea. The exposure is longer than the flash time.. and most cameras allow you to adjust exposure during the flash, so it must be doing some interpolation.

Reply
biotron

Dec 21, 2005 at 1:40 am

any chance it was rear curtain flash?

Reply
angry shock

Dec 21, 2005 at 1:41 am

The relativist in me asks: Why wouldn´t this be a "concoction" (I learned this word a little while ago and I am trying to use it, sorry =) of the several theories tried so far?

Hola Leino! =D !!!!

Reply
jurvetson

Dec 21, 2005 at 1:50 am

It was the little built in flash in the pocket Casio.

Rocketeer: since you win 70% of puzzles, you are held to a higher standard now… 😉 The CFB (ceiling flash bounce) theory is subtely different from the BS (backlit shadow) theory… I don’t think the ray trace can work and the intensity would be too low for a double-bounce.

I’m not sure if your MB (motion blur) theory is the same as Lamo’s light balance version.

I have a dumb question about MB; if we are seeing a bit of both the arm and the background, why are the areas of the white mullions relatively darker on the sweater? Shouldn’t they make the sweater more white?

Reply
Robbie1

Dec 21, 2005 at 1:54 am

Notice how the ‘shadow’ near his hand stops exactly where the colour changes from a colour which matches the background (his sweater)to skin tone. Seems like a processing artifact.

Reply
benjiman

Dec 21, 2005 at 1:58 am

i don’t think its a shadow from another light source.

the light on the arm seems too flat, too much like bounce light. if there was another source outside i think we’d see other indications of it.

on the other hand, the moonlight has been trmendous the last few days…

…still my gut tells me its a flash timing/exposure timing artifact.

Reply
jurvetson

Dec 21, 2005 at 2:03 am

I had originally titled this one Glitch in the Matrix matrix… =)

Reply
biotron

Dec 21, 2005 at 2:08 am

Ghost on the Shell Suit

(perhaps a little UK-orientated to be amusing…)

Reply
jurvetson

Dec 21, 2005 at 2:32 am

well, our jumper was quite anime-ted…

before we chalk it up to MB or artifacts, there is a troubling detail I noticed in the full size upload… The horizontal bar on the arm does not appear to be a perfect overlap with the background… That would support the shadow theories (which, if true, require an amazing alignment coincidence)

Now if the artifacts are just a burp in the multiverse, then we can unify this concoction of theories….

Reply
angry shock

Dec 21, 2005 at 2:45 am

What are ‘multiverse´s burps’ but the ad hoc patches par excellence to bugged concocted theories?

Others call them singularities, I think, but that´s for another puzzle.

Reply
drona

Dec 21, 2005 at 4:12 am

Any chance a second flash gun source that went off micro seconds after the first one from your camera’s flash could cause this effect? Remember there were two cameras in the room taking that picture.

Reply
drona

Dec 21, 2005 at 4:27 am

As Rocketeer has suggested I think there’s a double exposure going on here. The light from the setting sun is to the right of the picture. The camera is pointing due south. I’d choose the sun/ambient light over the "second flash" theory. Now the other question to be resolved is — why does the sock not have the pattern as well.

Reply
Gwênlyn

Dec 21, 2005 at 4:33 am

All I can say is that I have somewhere a photo of my daughter jumping in our living room, also with flash and showing the image behind her through her. Not finding it easily (too many photos!) but eventually it will show up…

Reply
Leino88

Dec 21, 2005 at 4:47 am

the more i look at this … the more i’m weirded out…
…i’m second guessing my own initial explanation….

…ah screw it…. i’m going home and taking the red pill.

😛
gnight

Reply
JKaljundi

Dec 21, 2005 at 7:40 am

I agree with what Rocketeer said, it is a question of the exposure being 1/60 and flash firing only 1/1000-1/10000 sec of that, and the grid exposure on the hand created by the outside backlight from behind the door. You do not see that on the body because during the whole jump that has blocked the light from the part of the door behind it. You see the grid on the shoulder and part of the body because these were at some point not covering the light from the back. Why there is no grid on the sock – because that part does not have a backlight, only dark background.

Reply
oldcola_

Dec 21, 2005 at 2:51 pm

Hello everybody,
that’s a nice one Steve.

I’ll try to give a new direction, not completely new as Gi talked about it first, but nobody gone digging that way. Interpolation. Not during capture but during compression. You use jpeg compression on your camera [EXIFF said me so ;-)].

If you look at the frontiers where the ‘illusion’ starts, you see that the compression created an artifact, a darker-then-everything-around zone. And the ghost is the gradient created from this zone, ’till the surrounding color ‘digest’ it. [1]

It’s interesting that near the boy’s hand you clearly see that this isn’t the lattices ‘projection – reflection’ but the lighter to darker transition, even from the wall! [2]
The flesh is darker under the white wall [4] then under the darker shirt [3].

Reply
aegdss

Dec 21, 2005 at 4:21 pm

I’ll add my crackpot theory to the mix.

As the grid on the sweater is darker, It doesn’t seem like
MB or double exposure is the cause, plus the hair doesn’t show any MB or double exposure effects.

Also, the effect show very good alignment with the door grid, no significant angular displacement, and seems to only occur in the dark background region of the sweater, not on the hand for example.

Going by the rule that “if it’s not on fire it’s a software problem”, my 1 cent is on a
the interaction between the CFA demosaicing and color balancing the camera is doing.

The grid lines are pretty orthogonal, and most CFA demosaicing algorithms tend
to weight edges to improve sharpness and reduce color artifacts. The grid on the door is almost white, while the sweater is almost black. Perhaps the cumulative effect of the demosaic and color correction along the white – black transition of the door grid – sweater gets the camera’s brain a bit confused.

Although the large area distance over which the effect occurs speaks against pixel processing, which is why I invoke the voodoo of unintended cumulative effects.

Any chance of seeing a camera raw file?

Reply
jurvetson

Dec 21, 2005 at 5:49 pm

Ahh, the plot thickens. I am now pinging a camera imager designer in Korea….

no raw file. Sorry.

PepsiVieux posted a great CSI analysis of the area in question… worth a peek to see the details:

Reply
T a k

Dec 21, 2005 at 7:01 pm

This is an awesome little puzzle.

I’d have to go with the double exposure theory. I must object to it being called the "BS" theory. 😛

The reason that the mullions appear darker on the sweater is because, without the flash, I think the dominant light source in the photo would have been the sky, visible in the top half of the doorway.

This is also the reason the pattern doesn’t appear on the sock – because there’s no sky visible behind it, and it would have been almost as dark as the mullions themselves.

See the shadow in the sky underneath the arm? This would be the silhouette cast by the arm against the sky during the non-flash exposure. And to address the speed issue, I think a swinging arm can easily reach 15 ft/sec.

Well, that’d be my guess anyway. I have a feeling most of this has already been said in slightly different language.

Reply
jurvetson

Dec 21, 2005 at 8:45 pm

Tak: I think you nailed it. But it’s the Rocketeer MB theory (BS was for backlit shadow)… And there would be some poetic justice in this outcome since "Rocketeer is always right"

The clincher detail, IMHO, is the arm shadow that you describe. Compare it to the smaller, sharper shadow cast by the hand. So the arm is flapping up (count on Lamo to spot the kid dynamics in flight… =).

The wide, diffuse shadow on the backlit glass shows the range of arm motion during the exposure (matching the amount of grid-overlap on the other side of the arm. You can see a diffuse "ungrided arm + shadow" that is about the width of the arm itself).

Then the flash fires, casting a sharp shadow from the hand, but wiping out the hand’s earlier diffuse shadow on the bright white wall. From this angle, the glass/outdoors is not as reflective, and the flash does not wipe out the earlier shadows. But it does fill in the white of the unobscured mullions.

I wonder if there is some white balance effect going on during in-camera processing. In a non-flash photo of the same area, the mullions look fairly bright in the pink domain and the backlighting from outdoors is quite blue. (maybe camera processing can also explain the non-perfect alignment of the horizontal line that crosses the sweater (which can be seen more easily on OldCola’s version)). But I think these issues are in the noise, so to speak… The key features in the photo are all explained parsimoniously by the Tak-Lamo-Rocketeer hypothesis.

Now I want to see a set up shot like this with longer exposure and perhaps faster motion to get a complete grid overlay!

Reply
T a k

Dec 21, 2005 at 9:23 pm

Whoa – my name has been hyphenated with the infallible Rocketeer!

Well. What more is there for me to do in this life?

Reply
angry shock

Dec 21, 2005 at 9:30 pm

My today´s aphorism:

A man with an analogic cam doesn´t get that effect.

The moral is up to you.

😛

Reply
Gwênlyn

Dec 23, 2005 at 4:47 am

Here’s the photo I mentioned earlier:

(Click to enlarge.)

I only blurred her face and raised the contrast a bit. The rest is unmodified.
Notice how her right leg is "transparent", as is part of her head and right hand/arm. In general, her right side (left on the image) tends to be "transparent". On the other hand (that is, foot), her left sock is seen blurred forward with the motion. In case it isn’t obvious: she’s jumping forward and down from a sofa, so the blurred pink streak in front of her sock is the direction of movement.

Reply
Uqbar is back

Dec 23, 2005 at 10:49 pm

Maybe CCD warming has a role as well. And the red-eye reduction mechanism’s pre-flashes might worsen its effect.

Reply
prolix

Dec 23, 2005 at 11:57 pm

I realize that it’s a bit late in the game to chime in (especially with redundant opinions), but I’d like to agree with a composite between Rocketeer and JKaljundi.

The details have already been described, so all I can add is argot: this looks like a more rigorous, metric artifact of "shutter drag" (movement plus disparity between flash and shutter intervals).

Reply
prolix

Dec 24, 2005 at 12:04 am

Sorry, I just want to add that on P&S digitals, this feature is typically called "slow sync".

However, Steve has already said that this was captured at 1/60. Still, the arm shadow below the arm, superimposition, and exposure of the outdoor light would seem to bear out similar effects, perhaps because of the speed of the subject’s movement relative to the shutter.

Reply
evanescent memory

Dec 24, 2005 at 12:46 am

Whatever it is: It’s a nice illusion! 😉

Reply
rtolmach

Dec 24, 2005 at 3:45 am

Wow, did you see TheAlieness wrote, "(to rtolmach) I *like* the way you think."
Made my day! (I love her photos SO much!)

Reply
angry shock

Dec 24, 2005 at 8:52 am

Panda! Are you ok? ’tis the season? ;;;-)

Robert, you are making mine with your words. Thank you. *hugs* =)

Reply
nhr

Dec 25, 2005 at 10:03 pm

Steve’s example picture had a shutter speed of 1/60th. seconds.

Picture a large, matte, opaque, black sheet of paper moving quickily before a background.

During the 1/60th s. during which the camera’s shutter is open and the CCD is accumulating photons, some parts of the background will first be entirely visible, then covered by the moving black sheet of paper (for the remainder of the 1/60th. s).
The black sheet of paper will thus act as a second camera shutter, shielding the dark, faint image that was recorded on the CCD by the ambient light before the flash fired, thus "protecting" it against overwriting by the photons emitted by the flash.
Also note that those parts that remain visible longest before being covered / hidden by the moving object will accumulate the most photons on the CCD, and thus appear brightest. This is consistent with the ghost picture’s brightness gradient on the kid’s arm.

I thus believe that the artifacts we’re seeing on Steve’s and GustavoG’s pictures are simply the dark picture that would have been recorded of the background, had the flash not fired and the shutter speed been higher and there wasn’t any moving object. Also note that in both examples, the mobile masking object is fairly dark-colored and opaque, thus acting as an effective shutter.

I suspect the mullions on Steve’s pictures look black because, without the flash’s light, the glass panes would have looked relatively brighter than the mullions, either because some light was shining through these panes from behind, or due to specular reflection on these panes of a light source — e.g. a bright white ceiling, or another window from across the room.

Reply

What’s That? (59)

52 responses to “What’s That? (59)”

Leave a Reply to The Rocketeer Cancel reply