*warning* I managed to sound unbelievably pretentious and arrogant. Please just ignore that
Also, I failed in keeping this short. I failed
miserably. Very miserably. You may want to skip to the paragraph starting with "Either way", that's pretty much my opinion on the whole video.
Then, of course, I expect you to be so icredibly intrigued that you actually want to read this horde of words, and then vote me for mayor. Or at least
not make a smoothie out of my family. I don't know. *warning*
I'm not entirely sure that the video in the OP is real, but I'm more sure it's real than that it's CGI. Like most of you, I have about a decade
and a half of experience in the latest games and 3d packages, but I doubt 'credentials' matter in this case. What does matter is what we're looking
at, and though a lot of comparing between videos and techniques has been done, a proper analysis has been lacking. A proper, incredibly long and
boring analysis.
Of course I won't pretend to be able to make a proper analysis, but at least I hope to provide a few pivotal points to the discussion. Sure, it's a
boring and a tedious process, but in stark contrast to simply repeating one's differing opinions, it is actually a process. The goal is first to
describe the video itself, then to lay out what aspects of the video could be used to deduce what's real and what's CGI.
- Shot setup and angle: We're looking at
ne long shot from apparently the right side of a plane, just behind the right wing.
- Foreground objects: we can see the sides of the window we're looking through (first the left side, then after panning the right side), and the
right wing. There's no immediately noticeable amount of dirt or damage on any part of the plane - it seems to be flawless. However, at 00:40 a drop
of water is visible on the window, and if you follow the drop while the camera moves around (not panning, but bobbing), you can see what appears to be
a faint layer of dirt on the window.
- Background objects: landscape mainly consisting of mountains; the only layer of clouds between the plane and the ground is at the height of the
mountains; a very thick layer that blocks the light (we can't see the landscape) for the most part, though the clouds thin out at the edges of the
layer until faded away. No layers higher than that are visible, apart from the layer of clouds at the horizon, which is seemingly at the same height
as the plane. Apart from possibly being just an visual effect of the sunlight leaving the atmosphere (? no clue), the layer in the distance seems to
be very thin.
+ Foreground lighting:
- Though the reflection on the nozzle-thingies (terminology fail, apologies) shows the contrast of the dark underside of the wing to the light sky,
the only part of the foreground receiving direct (sun)light is the tip of the wing. Since there is no shadow from the (vertical part of the) tip of
the wing, the light is coming from behind.
- The shadow on the wing can only be caused by the body of the plane. Assuming normal proportions of the body of the plane, the large surface area
on which a shadow is cast suggests that the sun is relatively low - ready to set or risen not long ago. However, since the horizontal part of the tip
of the wing does receive light - i.e., the light shines down on the wing - the sun can't be lower than the wing (as seen from the plane). It's
probably just above the wings.
- The top edge of the shadow runs diagonally, with the right side of the shadow being closer to the body of the plane than the left side of the
shadow. This means that whatever is causing the shadow tapers down to the right (lacking in vocabulary here, I'm sorry). Assuming the plane has an
ordinary body shape (no bulges on the body), that object would have to be the back of the plane, meaning that the sunlight is coming from the right as
well as behind. The shadow on the wing is sharp/crisp, suggesting a focused light - i.e., there is no layer of clouds between the sun and the plane.
- Background lighting: The sky is clear blue, apart from forementioned layers of clouds. The ridges of the mountains in the background cast shadows to
the right. No long shadows are visible, and most angles of the mountains seem to be receiving direct light, suggesting either a high sun or a sun more
or less directly behind the camera, casting the shadows on the backside of the mountains filmed.
- Changes over time: The camera moves around very lightly (easily visible by looking at the edge of the window to the left), then pans over slightly
to the right, exposing the right side of the window; and after a while zooms in to the background. When zoomed in, no part of the plane is visible.
So far the analysis of the video itself - on to the analysis of the role of specific aspects in the interpretation of the video. By this, I mean that
whether you think the video is real or CGI, your opinion is formed largely if not solely by observations of aspects of the video itself. Arguments
such as "the plane number proves this is not a real plane" are left out, since - though quite plausible - it relates to semantic interpretations of
the contents of the video as opposed to analysis of the visual aspects, and it would just convolute the analysis. In other words, it's a good
argument, but has no place in a visual analysis. I've tried to filter out all the arguments presented so far on both sides as well as provided some
of my own, but please forgive me if I've missed anything
..
Pointing towards CGI:
1. Lighting seems unrealistic
a - The wing looks weird, the wing tip has no shadow but the rest of the wing is occluded from direct light, yet is still quite bright
b - The reflections on the nozzle-thingies look too smooth and neat
c - Incongruency of light sources: The wing receives light from the right & behind; the ridges in the background however seem to be casting shadows
to the right (suggesting a light source on the left).
2. Camera steadiness seems unrealistic
- The lack of movement - it's presumably shot manually, but initially, apart from a slow pan & roll, there seems to be no kind of shaking and
fumbling one usually sees with common hand-held cameras.
3. Foreground seems unrealistically flawless
a - The immaculate foreground objects; the absence of visible damage (dents, scratches) and dirt (specks, rust) on both window and wing
b. - Absence of physical effects one expects to occurr at a wing when flying; there is no condensation nor liquid visible at any point of the wing,
so no movement of air is visible around the wing
c. - Absence of physical effects of light hitting the camera: the shot is clear with no reflection of the interior of the plane.
Pointing towards real:
4. The background seems flawleslly realistic
a - The dynamics of the lighting are believable, as are both layers of clouds (the one at the mountains and the one at the horizon)
b - Detail in the clouds in the background seems flawless, particularly in the way it peters out at the ends in unique streaks, and in the way the
clouds rotate as the plane moves.
5. Flawless sync between foreground & backgroUnd
- The movement of the foreground as the camera pans and rotates seems to be completely congruent with the movement of the background.
6. Foreground: there's a rain drop visible near the bottom of the window at 00:40, which looks and moves very realistic.
Now you may have noticed that I included "unrealistic lighting on the wing" as one of the aspects of the video that points towards CGI, and more
importantly, that noone has brought it up specifically (right?) so far. It may seem like I'm supplying silly arguments for one side as a setup for a
straw man argument, especially since I "explained" the shadows on the wing before I listed this
argument. In reality I wanted to include that one specifically because it's one of those things that doesn"t necessarily register cognitively, but
definitely influences how real the plane looks to you. What I mean is that you might think it looks fake, take a second look, decide that it's
possible, so you want to disregard it - but subconsciouisly, it will still keep looking fake, even though you think it's because of other things.
I've had this happen plenty of times in real life, when for example a creek looked creepily much like a crappy 3d render - but it being real life, of
course I had to disregard any thoughts trying to explain what it was that made it look fake; because any reason I could come up with, was easily
countered with the fact that it was a real scene. After all, reality does not depend on my definition of what looks fake, but the exact reverse should
be true; what I think looks fake should depend on reality. So if reality looks fake, I should revise what I view as "looking fake" - try to adapt to
reality and change what I think (and feel!) looks real. But this is way easier said than done; we can easily choose what we think but not how we see
(referring to the vast amount of processes involved in the transformation from the light on the receptors in the retina to electric signals all the
way down to the recognition of objects and motion, initial aesthetic appreciation, et cetera). Sometimes reality just looks a little fake
..
Presenting the part of my post that may actually contribute to the conversation: discussion of the observations!
Pointing towards CGI:
1a + 1b. As described in a fashion too elaborate for any more words, the lighting looks weird and out of place - fake - even though it's completely
possible for a wing to receive light like that. It's important to remember that you will probably never see lighting like that on the ground, and you
probably won't see it on a plane often. You can't see it on the ground partly because the sunlight is coming from just above the wing, meaning that
the sun would be too low in most surroundings for its light to be seen so clearly; and partly because even when you have a clear blue sky, you'll be
surrounded by the ground for 180 degrees, meaning that all the light is coming from above. When you're in the sky though, a clear blue sky becomes
even more clear; the quality of direct light you get up in the air with a clear blue sky on clean metal is surreal, but the indirect light (coming
from the sky as a whole) is also the strangest there is. To me, such things often look fake, but I realise they're not. Of course, since many of you
fly regularly, you're not so easily tricked
..
1c. I've got me there. The observation of the incongruency of light is of course only valid if my initial observation of light sources is correct,
which once again are as follows; foreground receives light from the right as deduced from the angle of the shadow on the wing in relation to the
normal shape of a plane; background receives light from the light as deduced by the shadows of the mountain ridges.
2a. Camera steadiness - there seems to be a lack of movement, but in my opinion this could easily be caused by a motion stabilizer - not the
post-processing type like VirtualDub's Deshaker, but an internal type of stabilizer. I have an ordinary photocamera which can make HD-movies with an
- in my opinion - amazing automatic stabilization, such that I've though about shooting short movies with it (if only I had an artistic side
). I
have absolutely no problem believing that there are normal people out there with a camera meant to shoot videos which can do at least the same or
better than my camera meant for photography. Of course this only applies to the smallest of movements - shaking and trembling of the hands; there are
many more types of movements you'd expect a hand-held camera to show. Those do appear to be missing in the start of the video, but it is absolutely
possible that the camera operator is more capable than the average joe (not everyone shoots crappy videos (no offense to anyone, I do!)). When the
camera zooms in (and just before), there is plenty of jerky movement - again, not the small-scale movement that can be filtered out by automatic
stabilization, but the short pans to and fro. To me, it is typically the type of movement you get when handling an object with two hands, holding it
with one and pushing or pulling a button or such - probably difficult to replicate with a headset that registers your movement, as well as with
complete cgi (try moving your mouse like that - never mind keyboard, it's even less sensitive). Animating it would be possible, but if you've ever
tried those things, you probably noticed that it is really easy to make the wrong kind of movement, suddenly making the whole sequence look fake.
3a. Admittedly, the wings don't look just clean, they look immaculate - "as though" rendered with a metal material without bump maps and normal
maps, something you commonly only see in the world of CGI. However, there are some minor streaks visible on the wing, close to the body, so the wing
is not actually as immaculate as it seems. For the window, it's hard to see particular specks of dust, but notice that on 00:40, you can see a drop
of rain on the window; if you follow the movement of that drop of rain, you can see a very faint layer of dust on the window. Since the camera has a
constant focus on the distant clouds and mountains, everything on the window is completely out of focus; even though the rain drop is quite thick,
it's not easy to see when it's not rolling down. Little specks of dust could easily be so much out of focus that you wouldn't see them at all. As
for the wings, one has to keep in mind that metal objects receiving light from a very flat angle with more than 200 degrees of blue sky all around
will look both surreal and immaculate, since both the angle and quality of light make for a type of reflection basically only seen on planes in flight
(though you can get similar effects on water). The airplane is supposedly 22 years old, but that only means we know for sure it has gotten at least
one paint-job after it's initial one (I assume they don't just let it decay). Also, it's not at all unlikely for a plane to be squeaky-clean;
planes do get cleaned once in a while, after all. It would just imply that the video was shot very short after bathtime, which may not be particularly
common, but isn't unlikely either. Of course, the same goes for the windows - cleaning the wings seems like something you'd combine with cleaning
the rest of the plane. Uncommon - yes, unlikely - no.
The lack of damage is also not unlikely - do you remember any dents on the wings of the planes you flew on? I don't, and if I had, I might've
panicked a little .. you know, aerodynamics and all
.
3b. Lack of condensation - condensation takes place in specific circumstances (the air has to be cool and moist enough for it's hydrogen molecules to
spontaneously condensate as soon as it touches the surface of the wing cutting through it), and even then normally only gets visible quite a distance
behind the wing. Disregarding the issue of distance, we might expect some condensation of the nose of the plane - but this being a clear blue sky with
practically no clouds, it's very reasonable to assume that the sky is too dry to condensate.
3c. Lack of light effects inside the plane - the complete lack of reflections of the interior of the plane seems suspicious. Once again though, I
believe the strange lighting is to blame; since the sun is shining from behind, the front of the camera-man and the camera itself are receiving no
direct light. The exterior though is very bright, so it's likely that any reflections of the dark interior would be to weak to show up in a
reflection.
Pointing towards real:
4a. It's true, both the mountains and the wingtip are receiving kind of a 'diffuse' light, as though the sun is shining through some kind of haze -
but not thick enough to make the light seem dull. The mountains do seem a bit darker, but the metal reflects a lot more light. (Just talking about the
'depth' of light here, the dynamic range)
4b, This for me is the pincher for the background; the clouds look absolutely real. One thing is sure; those are not 2d planes with alpha textures
stacked to resemble a 3d cloud (such as
in this post), the method that has
been used for almost all games up to the present. Real-time voxel manipulation is on it's way to our games, but as far as I have seen, it has not yet
been implemented in any game; it is
not just a matter of tweaking some settings and mods, it is a radically different type of representing
particles in games than has thusfar been used. If voxel clouds had been implemented in any flight simulator, you can bet your ass they would have
launched a campaign around it (as it would allow you to actually fly through clouds the way it looks in real life). Even outside the realm of
real-time games, it will take a lot of work to make clouds look that good; in the shot, you can see how the ridges of the mountain divide the clouds
in different streaks, and all of those streaks are highly detailed. Apart from the work it takes to get it to render right in combination with the
mountains, light source and sky, it takes a hell of a long time to render in such detail. It would be way easier to just insert some random background
footage and make it fit.
5a. There are plenty of open source (free) software packages that enable you to track a video fragment, so that you can import the camera data to a 3d
scene and - basically just start inserting 3d objects. If the video tracking went right, your objects will be seemingly inserted just fine in the
video. Often, you can still see some jerks and glitches; I couldn't spot any of that, but that doesn't mean it's impossible that the foreground was
added to the background through video tracking.
6. Raindrops like this are virtual non-existent in games so far. There are video-cards with awesome chips that just simulate the physics of particles
- and it hopefully won't take long till you can enjoy them in the cheaper videocards as well - but a single raindrop rendering and moving like that
in-game is just not happening right now. It could be done with 3d fluid simulation (which a lot of 3d software can perform), but doing it like this
would be a terribly nice job.. It doesn't draw attention, it refracts beautifully, it moves just like one would expect - once again, I don't think
someone went through the trouble of doing
Bringing this monstrosity to a halt, my conclusion: the foreground initially looks fake, but for reasons that could occurr in reality as well,
rendering them inconclusive; the background initially looks quite real, possible to fake but very difficult to make it look this good.
If only the background is fake, then the CGI-artist went through the pain of taking footage from a real plane and creating a scene to replace the
background with. Probably the least efficient way to hoax this.
If both background and foreground are fake, then the CGI-artist went through the same pain of making a background scene, but additionally made a plane
to go with. Both with very realistic physics effects - the clouds in the background, the raindrop on the window (@00:40) on the foreground. It's not
absolutely impossible, but I do think it's highly unlikely.
Either way, I don't think the background is fake, and though the foreground looks fake in first instance, it could be real. Of course, having written
argument 1c, I have to say that at least the wing on the foreground is fake; unless I am wrong in the analysis of the sources of light in either
background or foreground, the lighting is incongruent - which of course is impossible with just one light source. You can expect some differences in
how tall shadows are, since the angle at which the light hits is a lot lower on a plane than on the ground; but it is impossible for shadows to be
cast only to the right on one location and only to the left on another. This also means that this can not possibly be captured in one shot from a game
or 3d rendering scene, as computers do not make the mistake of shining a point of light in directions that depend on the locations they're shining on
(trust me, I've tried ;D). One thing we can be sure of if my observations are correct: the wing and background were combined, and was thus not
acquired in a single 3d environment (such as a game or a single 3d scene).
Though I do not think the background itself is fake, it is reasonable to assume that if the wing on the foreground is fake, the UFO must be fake as
well. After all, it makes no sense when you have a real UFO video to "tamper with the evidence" by inserting a plane. In addition, the process of
inserting a UFO is the same as inserting the wing, and this UFO in particular would not have taken a lot of work to insert into the background
footage.
All in all, I suspect the wing and UFO have been inserted on genuine background footage on ground of the incongruency in lighting. The shaking, the
edges of the window, the raindrop on the window, the very faint layer of dust you can see when looking at the raindrop while the camera moves, the
clouds and sky - those are all real. The only question arising with this conclusion is: if the shaking is real and if the shaking is caused by zooming
in and if the UFO is fake, then why did the cameraman zoom in? A possible answer: to get that beautiful shot of the clouds hanging between the ridges
of the mountains.
Thanks for enduring! Hope you enjoyed the bumpy ride..
P.S. Regarding all those things of which I said there's no game yet with such features (even when modded & videocards with teh awesome chipsets,
drivers, etc.) - I should probably prove that in some way, but I hope we can all agree it's a good thing I didn't try to do that in this post; it's
lengthy enough as it is. Also, I may not have a lot of evidence beside my experience and slight insight in game technology.