I know that gimbals are really cool and make great things possible, but I just can't wait for them to die. They are heavy, draggy and add complexity. It would be great to just record everything and then scroll to the portion on video you want to see. If you are capturing video for a production, you can fix it in post. If you are doing surveillance, you could completely stabilize your view with software. Maybe this will get us one step closer:
http://bit.ly/1gV7n2c
Comments
Carles, that is the nature of any fisheye lens. To be precise, the actual lens can have great optical sharpness at the edges, but the areas that the sensor has to cover gets larger and larger as you get closer to the lens edge. Meaning that the spatial image resolution is much lower at the edges compared to the center.
Here is a nice application video..
http://www.carxpert-racing.com/fr/moto2/video-360o/
To bad they did not mount another one on the front!
I have asked the team about the length of the boom, on the shadow it looks about 40-50cm however this would greatly influence the drive ability so i suspect that the cam is a lot closer..
The reason software stabilization won't completely replace mechanical stabilization any time soon is the physics of light. It might work in bright sunlight, where there are lots of photons, and you can do fast exposures, but as soon as you have low lighting, there just aren't enough photons to have a fast shutter speed, so you're going to get motion blur, which is not impossible, but very difficult to correct in software. And, it doesn't matter how sensitive your sensor is - the limitation in low light shooting is the number of photons. Of course, you could slap a giant lens onto the camera to capture more photons, but then you end up back at square one with a very heavy rig.
John, you say that the fish eye 360 deg are only sharp on the center. What makes you think so?
I believe that this is the future of gimbals, simpler, light, cheaper than gimbals, and requires much less prossesing power than camera arrays.
well, that's brave of you :) cinematography is there for more than hundred years and yet most of the stuff is still done manually, we saw very little improvement in last 10 years.. it may change - also with help close range aerial copters - but it won't go away :)
i offer you a bet that in 5 years gimbals will be much more affordable, much more effective, but there will be gimbal and there will be camera; and they will be separate pieces of brilliant technology :) and most of aerial video will still be done using (almost perfect) gimbals :)
practical usability of such things falls into very far future, i'd say.. and even if technical issues are all solved i still am a bit sceptical on person's ability to see and digest 360deg video - we're not even able to see everything on our existing 1080p and "discover" details when watching movies again; now with full-round video you wont notice even more..
maybe some day when reality will be mixed (virtual reality and real reality) it might make some sense, but this is very very far from "dying gimbals" :)
This is one of those nice idea deals that sadly has lots of technological limitations.
There is no magic way to get "perfect" electronic/sensor based stabilization. To get a sharp stabilized image, you also need sharp raw data from the sensor. But with lots of motion/vibrations you get motion blur. To compensate you need a low shutter time to get sharp results, and to get low shutter times you need lots of light. So it only works i bright sunlight, and even then it has problems, especially with vibrations. Another problem is that a fish-eye/360 lenses are only sharp at the center, and degrade quickly as you get closer to the edges. To get something resembling a nice HD video after 360 correction you would need a VERY high resolution sensor. But that again means degraded low light performance and increased price, so it's a bad circle.
But it's a great solution for hobbyist that only need "youtube" video quality.
Horizon would still move around...
I agree it would be nice to have a hobby version of the Google Street View camera array or the Gorgon Stare, but I don't think this camera is it, yet. The sensor is only claimed to be 1500×1500 px, and the geometry of projecting a 360° view onto a square sensor means that many of those pixels will be useless. The best no-fisheye crop of the resulting image will probably be only 640×480, which is just about right for live FPV but nowhere near the HD recording that everyone gets from their gimbals and GoPros now.
-
1
-
2
-
3
of 3 Next