UFO:Alien Invasion
Development => Artwork => Topic started by: Psawhn on January 21, 2008, 10:56:29 pm
-
It's almost ready, I think.
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final60001_0425.avi
There are several things I want to fix:
Starfield: Overall I'm happy with the look and feel of the starfield I'm using. There are just some weird streaks that I suspect result from the post processing combined with adjacent pixels. I'll have to find a way to fix that, somehow.
The engine is lit by sunlight. It should stay dark until it lights up to fire.
Jaggy zoom boxes.
The atmosphere in the zoomed version should have a longer falloff. Overall I'm very happy with the look of the atmosphere. (Too bad it turns white as the camera turns away from the planet. I wonder how to fix that.)
There are black borders around stars that can be seen through the atmosphere.
I can also add graphs and charts and a bunch of numbers, to make it look more NASA-sciency.
So. Feedback? Direction? Criticisms?
-
It's almost ready, I think.
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final60001_0425.avi
There are several things I want to fix:
Starfield: Overall I'm happy with the look and feel of the starfield I'm using. There are just some weird streaks that I suspect result from the post processing combined with adjacent pixels. I'll have to find a way to fix that, somehow.
The engine is lit by sunlight. It should stay dark until it lights up to fire.
Jaggy zoom boxes.
The atmosphere in the zoomed version should have a longer falloff. Overall I'm very happy with the look of the atmosphere. (Too bad it turns white as the camera turns away from the planet. I wonder how to fix that.)
There are black borders around stars that can be seen through the atmosphere.
I can also add graphs and charts and a bunch of numbers, to make it look more NASA-sciency.
So. Feedback? Direction? Criticisms?
It's . . . not quite what I was expecting. Very different from what we discussed, and I don't think it works. The zoom box looks very cheap in this implementation, and it's horribly unclear what's going on with the whole video, which is why I suggested the double pass in the first place.
The camera moving all over the place just makes it all more confusing. At the resolution the video will be playing, I don't think the player would be able to get a clue what it's all about.
Sorry if I seem harsh, but that's what I'm getting away from it.
Regards,
Winter
-
It's . . . not quite what I was expecting. Very different from what we discussed, and I don't think it works. The zoom box looks very cheap in this implementation, and it's horribly unclear what's going on with the whole video, which is why I suggested the double pass in the first place.
The camera moving all over the place just makes it all more confusing. At the resolution the video will be playing, I don't think the player would be able to get a clue what it's all about.
Sorry if I seem harsh, but that's what I'm getting away from it.
Regards,
Winter
I don't mind harsh. :)
Here's what I had in mind for the camera movements. The satellite is controlled by momentum wheels, giving a gyroscopic effect to its movement. The telescope is also computer controlled, with a program that tries to focus on anything unusual. Because it has to react quickly to this sudden object, the gyroscopic effect of the momentum wheels adds a rotation while the camera tries to track the UFO.
The zoomed in version is done after the fact, and stabilized to the initial view.
I'll try to get another version done tonight.
-
Pff, took me long enough... ::)
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final70001_0833.avi
Went with a blinky zoom box instead of a picture-in-picture approach. There's a bunch of dressing and scientifikey whatzamahoozits that I I want to add to the image later.
What I'm worried about is when the animation starts playing for the second time, before the zoom, the player might think "oh, it just repeats" and not actually watch anything. It's easy enough to cut bits out if it doesn't work.
-
Hrm. Tough one. Here are my two comments:
1) I'm afraid I still don't like the camera movement. It makes it feel like the telescope is either anticipating the jump or tracking the Carrier after it jumps in, rather than a chance catch on a big wide-angle shot. It doesn't do anything to help the player understand what's going on in the video, either. I personally find it quite confusing, making it difficult to track what's supposed to be primary focus of the video, the Carrier, not the camera. Do you see what I mean?
2) In order to draw the two passes together, don't just do an immediate restart after the first pass. Try playing a second or so of static after the first pass, then pop up a freeze-frame of the beginning, followed by the zoom-in effect, and then resume play.
Also, I'm assuming the frame or two of blank cube at the beginning isn't supposed to be there. ;)
Regards,
Winter
-
Ah, now I get what you mean. Basically have no tracking on the camera at all, and the UFO just happens to jump within view and fly past. I can save my pretty Earth rig for other shots, too, instead of trying to squeeze it in. :P
I can try a second of static. That'll probably tie things together a bit better.
And, yeah, I accidentally overwrote the first frame of animation with that stupid cube. >:(
-
For the Zoom-in sequence, what might also work is the quick static, a super-fast rewind (like someone running a DVR backwards really fast), the zoom box frames in on the general area of the incident (jump-in). It then enhances the new framed area, frames an even tighter area, zooms and enhances again, then plays the "close up" of the carrier. Kind of a two-stage deal. Wide shot, static, frame, zoom & enchance, frame again, zoom & enhance again, close shot.
I realize it's likely a lot of work, but the actual time watching-wise would only be a matter of ~10 seconds.
Just another idea to consider...
-
For the Zoom-in sequence, what might also work is the quick static, a super-fast rewind (like someone running a DVR backwards really fast), the zoom box frames in on the general area of the incident (jump-in). It then enhances the new framed area, frames an even tighter area, zooms and enhances again, then plays the "close up" of the carrier. Kind of a two-stage deal. Wide shot, static, frame, zoom & enchance, frame again, zoom & enhance again, close shot.
I realize it's likely a lot of work, but the actual time watching-wise would only be a matter of ~10 seconds.
Just another idea to consider...
I like that idea. A quick-reverse inside the static would help make the point of the vid clearer. The double-zoom is also interesting, but maybe unnecessary.
Psawhn, what do you think?
Regards,
Winter
-
That's an idea I'm willing to try. I didn't think of rewinding combined with light static, though, which I think will seem better than just a raw rewind. I agree, though, that the double-zoom might not be needed.
Another idea I had was a 'scan' that goes through and enhances the resolution, however I think that would be lost with only a few stars in the field of view.
I have to re-render the first sequence to get rid of the rollercoaster camera, and while I'm at it I might as well re-render the second sequence to take away a few of those glitches. Once I finish rendering, though, I save the files so it's easy enough to chance post-processing. It's also easy enough to go and change parts of the animation, as long as the frames match up before and after. (So we could experiment with alternative jump-in flashes and graphics.)
-
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final110001_0850.avi
Just finished rendering another pass. There's no post-processing (zoom effects, static, rewinding) etc... on it just yet. This is just to test the lack of camera shake and stuff. When I get time later today, there'll be some static and rewinds.
-
I hate triple posting, but here's the previous rendered sequence with the static/rewind added.
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final110001_0940.avi
-
I like it, it's great. The zoom box is perfect, the effect is great, it's finally clear what's happening. Maybe you have time for one or two final tweaks:
1. The static on the rewind section looks quite weird, and not recognisably like static that you might see on rewinding something. Anything you can do?
2. I don't think the second pass zooms in quite far enough. Just a wee bit more to bring the Carrier more in focus would be nice. Remember that the video is going to be shrunk to play in the UFOpaedia.
3. Could you add a timer in one corner? This would be great to help bring the point home, and having it run very quickly would help show that the footage is time-compressed.
Regards,
Winter
-
Here's the newest version. I still have to figure out how to fix those weird star shapes, but overall it's probably almost ready.
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final13_0001_0920.avi
The weird static is actually caused by the xvid compression. The raw files look more like this:
(http://img.photobucket.com/albums/v388/Psawhn/UFO-AI/static_shot.jpg)
I actually kinda like the distorted look of the compressed version, but I admit I don't know what rewind-static would actually look like with digital video.
I also just realized I made the timer compressed during the fast-rewind, but I forgot to reverse the timer! :P I'll fix that right away.
Edit: Fixed timer:
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final13b_0001_0940.avi
-
Is that grid effect in the latest version intentional? 'Cause it kind-of detracts from what's going on on the screen.
I think it's great, my only remaining request would be to make the timer faster, I want to see the first two digits go up a few points from 50.
Regards,
Winter
-
The grid effect was an experiment. I guess it didn't go over too well. Here's a try with the grid set to only 10% opacity.
The timer was also originally in realtime. (Useless trivia: the camera orbits the Earth in realtime, too.) I sped it up x3 as you asked.
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final14_0001_0940.avi
You can still go ahead and be really picky if you want. ;) For instance, there's a weird star shape in the second sequence (that can be called a nebula). and the clouds seem to disappear for some reason.
-
The grid effect was an experiment. I guess it didn't go over too well. Here's a try with the grid set to only 10% opacity.
The timer was also originally in realtime. (Useless trivia: the camera orbits the Earth in realtime, too.) I sped it up x3 as you asked.
The camera may orbit in real time, but I find it very hard to believe that a Carrier could make turnover and disappear on the other side of the Earth in a matter of seconds, or even minutes. It would have to be going a good fraction of lightspeed in order to do that, and decelerate at hundreds of Gs so as not to overshoot Earth orbit. Even antimatter engines can't do that without getting backed up by some magitech, which we don't want.
It should probably be sped up by about 10x more.
The grid looks better at this opacity, we can keep it, but I still don't like the big circle around the Carrier in the zoomed-in run. Can you get rid of that? We're already zooming in on the thing and tracking it as it moves, we don't need any more emphasis to draw the eye to the middle of the screen.
Regards,
Winter
-
I figured that the carrier was already heading towards the horizon as soon as it jumped in. After disappearing behind the horizon, it still had a good 30-60 minutes of deceleration time. But there's already the problem that in the wide-angle shot, the carrier's apparent size is simply too large for the craft to actually disappear behind the earth instead of plunging straight into the atmosphere.
...actually, I just realized the best solution might be to make the carrier move parallel to the horizon, instead of heading towards it. That definitely helps solve the problems of apparent size vs. realistic distances. The video can also still be realtime, because relative velocities at orbital speeds emphasize the luck and ranges of the encounter, without making it unrealistic. (It also takes it away from that stupid weird star shape near the horizon)
I can definitely get rid of the big circle, too. Its original intent was just another way to distinguish the zoomed-in shot from the wide-angle shot.
-
Not many changes, actually. The big circle was removed, and I adjusted the times. Now it goes from 02:08 to 03:57. I had to develop a new method of making the numbers (Blender doesn't yet have an easy way of doing things like that), but now I can pretty much pop up any range of numbers, using HH:MM:SS convention quickly, for any length of time, for any range of values.
I thought of adding a date, but Alien Origins aren't researched at any set date.
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final15_0001_0940.avi
-
The zoom-in shows up really really blurry for me. Is that intentional? After all, this is a space telescope supposedly capable of photographing distant galaxies, it should be able to provide a clear enough picture between the Earth and Moon.
Regards,
Winter
-
There shouldn't be anything different between this zoom in and the previous ones. The shot's blurry because I had to enlarge the picture to be seen more easily by smaller resolutions - the rendered image itself is enlarged 4x. The camera's zoomed to Blender's max, so I would have to rerender the entire shot at a higher resolution, probably putting my laptop out for a day or two, or move the camera physically closer, which may introduce some awkward movements and further reduce the correlation between the two shots.
I'm trying to maintain a balance between technical realism, astronomical realism, and artistic necessities, and the last one tends not to mix well with the previous two. :)
Without rerendering anything, the loss in resolution may be explained by the fact that both the lit side of the Earth and the Moon actually are in the picture at all, and in space those things are bright, which would probably need 2080s techonology for the telescope to even operate in those conditions. The filters and settings allowing it to take a shot without the image being hopelessly overexposed might reduce the resolution. Combined with that, this is a wide-angle shot, not a narrow-field shot used to take pictures of galaxies.
Truthfully, I think that if the telescope were looking at the settings needed to view planets and galaxies, I think the carrier would show up only as a very long streak in a single image. :)
-
There is also the thought that the telescope was not designed to shoot objects so close to itself.
It was designed to focus on stars and objects light years away and that is based off of light waves entering the telescope lens. Not a literal physical object in such close proximity to itself.
-
How is focusing on an object close to itself NOT based on light waves entering the telescope lens?
-
I was thinking about that, but even though the space telescope can only focus at infinity, objects would still have to be exceedingly close (within several kilometres) before the object would be visibly blurry, I think.
-
Lets put it into context of the human eye.
The eye has the ability to shift its focus very quickly on objects at various distances. This is what gives us our telescopic vision.
Now, in terms of a space telescope, it is designed for one purpose...to focus on small points of light at an EXTREME distance based on different light wave frequencies. The Hubble Space telescope sees in Infrared, Gamma and UV light frequencies only. It is only after computer analysis of these images and compositing do we get the full color images we all have known to enjoy viewing.
So, for a space telescope to 'suddenly' focus in on an object that is only a few hundred thousand miles away from it and moving close to MACH 10 in speed, is nearly if not impossible. Thus blurred images, let alone images that aren't even in a visible light range to the human eye.
-
I know what you're saying, and I think that even to a space telescope, a distance of several hundred km is as good as infinity as far as focusing is concerned.
The part I'm skeptical about is not focusing on the UFO, but the UFO staying in the shot longer than a fraction of a second. Relative velocities even in LEO can reach past 14 kilometres per second (Mach 10, which doesn't apply in space as there's no air or sound in space, is only about 3.4 km/s). At any distance, if the camera can zoom in to visibly see the UFO, the apparent velocity (speed the object moves across the sky) will be too high to see more than a streak in a single frame of video. The camera would have to somehow be rotating at the right speed to track the UFO, then the UFO miraculously jumps in front of the camera.
Actually, maybe the best way to fix this is to not have the tracking camera be a space telescope, but some other, lesser, camera with a wide field of view. Considering the Hubble can see the moon at this level (http://www.nasa.gov/vision/universe/solarsystem/hubble_moon.html) (top image), even my zoomed-in shot is way too wide of an angle to be of any scientific value in a photography mission.
-
I know what you're saying, and I think that even to a space telescope, a distance of several hundred km is as good as infinity as far as focusing is concerned.
The part I'm skeptical about is not focusing on the UFO, but the UFO staying in the shot longer than a fraction of a second. Relative velocities even in LEO can reach past 14 kilometres per second (Mach 10, which doesn't apply in space as there's no air or sound in space, is only about 3.4 km/s). At any distance, if the camera can zoom in to visibly see the UFO, the apparent velocity (speed the object moves across the sky) will be too high to see more than a streak in a single frame of video. The camera would have to somehow be rotating at the right speed to track the UFO, then the UFO miraculously jumps in front of the camera.
Actually, maybe the best way to fix this is to not have the tracking camera be a space telescope, but some other, lesser, camera with a wide field of view. Considering the Hubble can see the moon at this level (http://www.nasa.gov/vision/universe/solarsystem/hubble_moon.html) (top image), even my zoomed-in shot is way too wide of an angle to be of any scientific value in a photography mission.
Remember that the zooming-in is a digital enhancement of the image, not an actual refocusing of the telescope lens. The telescope itself was part of a program that reconfigured (nearly) all human space monitoring equipment to give as much coverage of the solar system as possible, so it's entirely plausible that the wide-angle setting of the telescope is part of that.
And if you consider how many terabytes of data a telescope in 2084 could store and transmit every day, there's really no restrictions to how much visual data it could absorb.
Regards,
Winter
-
In the end, Psawhn is correct. Image focusing would be a secondary issue with the shear velocity of the object entering orbit being the first.
-
I had to cheat a bit to increase the resolution. I actually increased the size of the carrier and the moon by four. See if you can spot the visual blooper this causes. :)
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final16_0001_0955.avi
-
To be honest with you, I can't tell. I much prefer this version to the previous one, though!
Regards,
Winter
-
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final17_0001_0955.avi
I fixed that little blooper, as well as getting rid of some annoying stars in the first shot, and also increasing the bloominess of the jump-flare.
If you have no objections, this will be the final version of the movie.
The visual blooper was the inconsistent location of the UFO relative to the moon at the point it jumped in. (The old video's still up if you want to take a look). In the wide-angle shot, the UFO appeared well to the right of the moon. In the zoomed-in shot, the UFO appeared while in front of the moon. This was because in the zoomed-in shot, the moon (and UFO) were actually physically enlarged, and the moon sort of 'reached out' to grab the UFO :).
-
https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_final17_0001_0955.avi
I fixed that little blooper, as well as getting rid of some annoying stars in the first shot, and also increasing the bloominess of the jump-flare.
If you have no objections, this will be the final version of the movie.
The visual blooper was the inconsistent location of the UFO relative to the moon at the point it jumped in. (The old video's still up if you want to take a look). In the wide-angle shot, the UFO appeared well to the right of the moon. In the zoomed-in shot, the UFO appeared while in front of the moon. This was because in the zoomed-in shot, the moon (and UFO) were actually physically enlarged, and the moon sort of 'reached out' to grab the UFO.
Speaking for myself, I think this looks very nice.
If you have any trouble making a RoQ out of it, give me a shout (PM) and I'll help you out (or I could do it for you).
-
Yeah, I don't think I want to fiddle with trying to compress it into ROQ, after hearing of your tales of adventure and horror trying to fiddle with it. :)
If you want a version without the xvid compression: https://webdisk.ucalgary.ca/~djetowns/public_html/misc_files/UFO_AI/MIMIR_sequenced7.zip
(You may have to wait a while until the file completely uploads. At the time of writing this, it's at 30MB/88MB at 57KB/s)
Those are the final composited results, before the compression into a the xvid avi. It's a sequence of 955 .png files, about as lossless as you can get. :)
Half of the effects in the staticky-rewind are actually from the compression algorithm not knowing what to do with the static, so the ROQ video may look considerably different in that part.
-
Here you go:
http://roq.destructavator.com/
If anyone else needs a roq made, let me know and send it my way and I'll put it on the page.
Regarding the multiple versions, the 512x512 would probably be better for fullscreen, although the 256x256 would probably be better for part of the screen, such as in the UFOPedia, and if I end up making any tiny ones, I'll render it in 128x128 also.
Seeing how many links to files I provided in the past, I've decided to organize them onto these pages on my site so that no one has to hunt through the forum via a search to find everything I offer to the project.
If any of the links don't work, send a message and I'll fix them ASAP.
-
Oops! I goofed on the aspect ratio, there's black at the top and bottom during playback within the game.
I'll fix this within the next couple of days, I'm about out of time right now.
- Dave
EDIT: Just got them fixed, they should be uploaded about 20 min or so from the time of this post...
EDIT: OK, done! If any links or playback doesn't work, let me know.