Armory Sponsor
Posted: 4/21/2018 11:21:35 PM EDT
|
https://www.kickstarter.com/projects/189168434/aurora-worlds-1st-day-night-camera-with-true-night?ref=nav_search&result=project&term=night%20vision
Curious about this campaign, and the pricepoint. I have a PVS-14 already, but this one almost seems too good to be true. |
|
I never have participated in a crowdsourced effort by buying in. But I have watched a few disappear without a trace and others succeed.
The successful ones seem to disappear as suddenly as the failures. Were these successes acquired and then utterly shut down to eliminate competition to keep profits high in the big firms? Did the product get a re-brand and the price brought up to keep the acquiring company's preexisting prices high? I don't know. The FLIR One may have started just this way. In any event, there was a crowdsourced effort or two that were very similar to what FLIR One ended up looking and working like. Coincidence? I have no idea. I didn't buy in or follow the money. The reason I have never become too excited about any crowdsource funded development effort, is that the seed money goals usually strike me as being pretty meager. In this case, 50 thousand bucks (and 100 thousand realized) for a project run by a half dozen guys. If I were in business and needed a secretary for a side project such as handling correspondence for one like this, it would cost me that to hire her for a year, even if I paid her minimum wage. But I'm not in business, never have been in the business of developing projects, and don't know, for instance, what it costs to specify pre-existing components to be used in a specified manner, to an established production facility that is capable of engineering and making the enclosure and human interface elements, and have them made. I am always skeptical, and that's no reflection on anyone or any enterprise including this one, it's just how I think. Having said all that, If these hit the market and work as advertised, even if the battery life was short with options turned on, I'd cheerfully buy one at full retail. They look great to me. Very fun and useful for all but the most serious stuff and not so expensive that you would sweat blood if you handed it off to a kid or had to leave it in the trunk. I'd just prefer to wait for a review or two, post-production. This listing posts names and photos and some biographic info about the developers, which is something I have not seen in such detail before in similar offerings. If you choose to research the people, and if you find dead links associated, sites like the Internet Archive Wayback Machine might reveal archives of those old pages |
|
I participated in the HackRF kickstarter and it went as planned but that's my only experience with it so far. These guys appear to be ready for production. Their campaign is all just pricing incentive. If the thing doesn't exist it's usually obvious because they talk about beta products and research phases etc...
There was a similar product video going around a year or two ago (color night vision). I wonder if it's based on similar technology. |
|
Quoted:
I guess the sixty four dollar question is why can't they get conventional venture capital. If it's such a good idea then why aren't the big shooters getting in on it. |
|
It seems the FOV of this unit is somewhere around 43deg x 33deg (53deg diagonal) but may be smaller. That's a great FOV truly. I couldn't find good starlight photos of this unit, so I think it's rather safe to assume it's better than Gen1, but probably not that useful in dark conditions. Most (or maybe all, not 100% sure) of the sample pics & videos are from ambient lightning environments where a Gen3 tube monocular would go very low gain (with automatic gain control - MX10160 tubes) or even start autogating. Given the sensor has a color filter on top of it means it already loses somewhere around 2/3 of the light that's available, but this is assuming it has no dedicated no-filter-pixels in there, which could bring the light losses to a bit better level. In comparison to a full frame camera the light gathering ratio is around 1:7. Gathers 14.3% of what a full frame sensor would. For a reference video with a full frame camera with a rather fine objective lens: Night time footage from an Airsoft game with Sony A7S (already old model & sensor) - 50mm F/0.95 manual focus - ISO 104,000 (not even close what the Sony's are capable of) The SiOnyx camera sensor has a 812,000 ISO sensitivity while the Sony's have 409,600 so about half. That'd bring the capabilities to perhaps 30% of that of a Sony full frame. That is impressive if it really works, though still not there for starlight conditions I assume. Both with max ISO, Sony with F/0.95 lens and SiOnyx with the F/1.4, the percentage would be 13%. 20% if the Sony had a F/1.2 lens. Another comparison, with an APS-C size sensor (~3:8 ratio to this 1" sensor - 37% light gathering compared to APS-C) and a F/1.2 lens a 60hz video is not good starlight conditions. The marketed camera has a F/1.4 objective lens. Though the 800K ISO might offset the gap to favor the SiOnyx by a few multiples for most APS-C sensors. The Nikon D500 has an extended max ISO at 1,638,400 though the IQ is not pretty. In theory that would leave the SiOnyx at 18% of the max of D500 - also just by numbers the D500 would be ~50% more capable in low light than a Sony, but the IQ will be substantially worse. Sample video from Nikon D500 with F/4.0 lens at 1,638,400 ISO. F/4.0 would mean 9% of the light gathering of a F/1.2 so this video would look a heck more clearer with a proper lens. Of course assuming these F numbers really represent the T values of the lenses, but even if they were varying 0...20% the "results" still keep in the same order. Another thing I very strongly suspect is that the latency from sensor to viewfinder is more than 1 frame (17ms with 60hz) and will cause uncomfortable viewing head mounted unless you are stationary. It may very well combine frames for an effective shutter speed of more than 1/60 but that will introduce blur and if it is buffered before output will also lead to even more latency + due to that even more blur when turning your head. I know from experience that 60hz with low latency is blurry even with no frame combining and with no special tricks. One trick to eliminate most of motion blur would be to flash the backlight of the display panel so it is lit only for a fraction of that 1/60sec. At 60hz that becomes visible, it'd need to be closer to 100hz in order to be unnoticeable. Edit: they use an OLED, so they might be able to run it at 60+ hz and flicker it - in other words making the display lower in image persistence. Many modern devices (TV's, cell phones, vehicle LED lights, etc) do this to adjust brightness, but as many have noticed with 60hz cell phone displays you can see the flicker. In theory they could update the display at the same time the sensor is being read, racing the beam as some call it to essentially give a close to zero latency in the top part of the frame. Could be the sensor is read once and not in a line by line fashion so again in theory the latency might be low, but I bet it's at least one frame if not significantly more. More than one frame will certainly cause motion blur. All that said it is priced very competitively and if it works in moonlight it could very well be worth the money unless you plan on headmounting it (especially if it's not unity mag which I couldn't find from the specs). Quoted:
Quoted:
I guess the sixty four dollar question is why can't they get conventional venture capital. If it's such a good idea then why aren't the big shooters getting in on it. |
|
Quoted:
I have one on order, and I should have been clearer in stating a $500 CMOS monocular that was head mountable, such as this unit. Quoted:
Quoted: I have no personal experience with it, but the armasight spark core seems to fit that bill from people who review it. |
|
Quoted:
It seems the FOV of this unit is somewhere around 43deg x 33deg (53deg diagonal) but may be smaller. That's a great FOV truly. I couldn't find good starlight photos of this unit, so I think it's rather safe to assume it's better than Gen1, but probably not that useful in dark conditions. Most (or maybe all, not 100% sure) of the sample pics & videos are from ambient lightning environments where a Gen3 tube monocular would go very low gain (with automatic gain control - MX10160 tubes) or even start autogating. Given the sensor has a color filter on top of it means it already loses somewhere around 2/3 of the light that's available, but this is assuming it has no dedicated no-filter-pixels in there, which could bring the light losses to a bit better level. In comparison to a full frame camera the light gathering ratio is around 1:7. Gathers 14.3% of what a full frame sensor would. For a reference video with a full frame camera with a rather fine objective lens: Night time footage from an Airsoft game with Sony A7S (already old model & sensor) - 50mm F/0.95 manual focus - ISO 104,000 (not even close what the Sony's are capable of) The SiOnyx camera sensor has a 812,000 ISO sensitivity while the Sony's have 409,600 so about half. That'd bring the capabilities to perhaps 30% of that of a Sony full frame. That is impressive if it really works, though still not there for starlight conditions I assume. Both with max ISO, Sony with F/0.95 lens and SiOnyx with the F/1.4, the percentage would be 13%. 20% if the Sony had a F/1.2 lens. Another comparison, with an APS-C size sensor (~3:8 ratio to this 1" sensor - 37% light gathering compared to APS-C) and a F/1.2 lens a 60hz video is not good starlight conditions. The marketed camera has a F/1.4 objective lens. Though the 800K ISO might offset the gap to favor the SiOnyx by a few multiples for most APS-C sensors. The Nikon D500 has an extended max ISO at 1,638,400 though the IQ is not pretty. In theory that would leave the SiOnyx at 18% of the max of D500 - also just by numbers the D500 would be ~50% more capable in low light than a Sony, but the IQ will be substantially worse. Sample video from Nikon D500 with F/4.0 lens at 1,638,400 ISO. F/4.0 would mean 9% of the light gathering of a F/1.2 so this video would look a heck more clearer with a proper lens. Of course assuming these F numbers really represent the T values of the lenses, but even if they were varying 0...20% the "results" still keep in the same order. Another thing I very strongly suspect is that the latency from sensor to viewfinder is more than 1 frame (17ms with 60hz) and will cause uncomfortable viewing head mounted unless you are stationary. It may very well combine frames for an effective shutter speed of more than 1/60 but that will introduce blur and if it is buffered before output will also lead to even more latency + due to that even more blur when turning your head. I know from experience that 60hz with low latency is blurry even with no frame combining and with no special tricks. One trick to eliminate most of motion blur would be to flash the backlight of the display panel so it is lit only for a fraction of that 1/60sec. At 60hz that becomes visible, it'd need to be closer to 100hz in order to be unnoticeable. Edit: they use an OLED, so they might be able to run it at 60+ hz and flicker it - in other words making the display lower in image persistence. Many modern devices (TV's, cell phones, vehicle LED lights, etc) do this to adjust brightness, but as many have noticed with 60hz cell phone displays you can see the flicker. In theory they could update the display at the same time the sensor is being read, racing the beam as some call it to essentially give a close to zero latency in the top part of the frame. Could be the sensor is read once and not in a line by line fashion so again in theory the latency might be low, but I bet it's at least one frame if not significantly more. More than one frame will certainly cause motion blur. All that said it is priced very competitively and if it works in moonlight it could very well be worth the money unless you plan on headmounting it (especially if it's not unity mag which I couldn't find from the specs). I bet they have a finished product already hitting the production lines as they "promise" a delivery in a few months. They probably have already all the capital needed, this is just for free publicity - which seems to have worked pretty good. With that 50k goal they had you sure would not develop a product like this unless you worked alone and knew people to give good discounts for everything, for example the tooling for the molds for the plastic parts is already taking a good chunk of that money. |
| In April 2013 SiOnyx announced their line of sensors that offered "unprecedented performance advantages in infrared imaging with sensitivity enhancements as high as 10x incumbent solutions". Their intent was to replace image intensifier tubes with something much better, much more modern. I suppose that resorting to Kickstarter indicates that they may have a product of some sort (digital night vision) but image intensifier tubes are still here to stay. |
|
SiOnyx is the "Black Silicon" company... Just a note - look at the sensor pictured. Does it look black to you? Nope... That's not Black Silicon.
If I had to take a guess, it's probably the unmodified sensor that they then convert to black silicon for military application. If that was a true black silicon sensor, then it would be dirt cheap at that price. So it's probably just a back-lit sensor, which are still pretty good in dim light, but it's not true night vision. As a point, notice in the video that the image struggles a little with the night color image of the bridge. Gen1 would be blooming like crazy in the same environment, so it's probably somewhere around, but not quite up to, Gen1 performance in color. I'd guess it's slightly past Gen1 in Monochrome. So might still be worth a look, especially if the video quality of recording is worthwhile and if it's still sensitive to IR light at all. I'd be interested to know how they do that if it's a color element - but the kickstarter page is 99% marketting, and 1% incomplete facts. But it would probably lose even to a Spark CORE. One thing I can be absolutely sure about - anyone who has anything even remotely military useful doesn't put it on kickstarter, they sell direct... And SiOnyx does make military technology, so they have all the background and experience they need. Probably it's not SiOnyx direct - just one of their sensors. David. |
|
Their website says the sensor comes with either a Bayer filter or monochrome, and now that I read the comment section Compasscall had asked this and they answered it's the filter version.
Also the answer whether it is unity mag or not is in there in the comments section too. Someone asked how things look with the digital 4x zoom and the answer was double size regarding natural vision, so I gather it's 0.5x mag then. Not head mountable very comfortably then - even assuming all other aspects of head mount requirements were met. Running at 2X digital zoom, 360p resolution and half the FOV, may bring it closer to 1X. |
|
Quoted:
snip In researching Sionyx I discovered Sony may have already worked with Sionyx and their black silicon tech for their image sensors: http://image-sensors-world.blogspot.com/2016/06/rumor-sony-experimenting-with-black.html |
|
Quoted: murtis, great breakdown with the Aurora/Sony comparison--which is the first thing I thought of when I saw this, given how impressive the Sony mirrorless cameras have been with low light performance (especially something like the A7S II w/ IR filter removed). I wonder if an IR-converted NEX5 or A6000 would give us similar or better low light video to the Aurora in an equally or more affordable (though admittedly larger) package? In researching Sionyx I discovered Sony may have already worked with Sionyx and their black silicon tech for their image sensors: http://image-sensors-world.blogspot.com/2016/06/rumor-sony-experimenting-with-black.html Maybe if you removed both the Bayer matrix & IR filter you'd get a good boost and gain nice monochrome low light abilities. I have no knowledge about how much sensitivity which sensor has at NIR ranges. SiOnyx 800K ISO with F/1.4 vs APS-C 51200 F/0.95 still leaves SiOnyx, at least in theory by just the given numbers, at 2.5x advantage. If the APS-C could go higher in usable ISO or have the bayer matrix removed they'd be close or the scale would start to turn. Edit: never did an IR modification to a camera nor have read in detail about it, but indeed it seems some cameras benefit hugely by removing it. Maybe almost 10x difference in ISO under certain lightning conditions if my eyes see it right. Looking at the video in Photoshop the IR removed pic is about 5 stops brighter - ~30 times "more light coming in" assuming my math holds here. That would certainly bring APS-C cameras closer & beyond the SiOnyx even with "low" ISO values - APS-C with F/1.2 @ 51200 ISO vs SiOnyx, APS-C brighter by about 3 stops. Taking out the bayer matrix would bring a hefty boost, in theory 3x or even a bit more if the matrix blocks "wrong" wavelengths good and is lossy in the "correct" wavelengths. I have no idea how feasible that is. On some sensors it's doable but who knows how it's with the A6XXX sensors. |
|
Quoted:
snip In any case, here is Sionyx's own answer to a backer question about how the Aurora compares to the unaided human eye in deep low light: ..."camera is better than your eye down to 2-3 mlux (.002 - .003 lux). Below that, if there is no IR light in the scene, once your eye adjusts after 5-10 min then your eye is slightly better. So right around moonless starlight is the transition. Anything higher than that (say quarter moon), the camera is better." |
|
Another interesting tidbit was their answer to a question about latency between the Aurora camera and its viewer: "The view finder is running at 60fps. Latency between camera and viewer is sub 1 frame." Does this mean that navigation via helmet mounted Aurora camera is possible? As UNV and N-Vision infer, probably not. But, we are getting close. Plus you have to consider the price point is a mere $800 ($559 on kstart ATM). A Canis Latran PVS-14 clone this ain't! Actually it seems a little more along the lines of an affordable consumer action camera version of the Rochester/Photonis CNOD. Makes you wonder what technology like this applied to a NVG product at the $5k to $10k or even $20k price-point can get you.
So is the future of night vision digital, with something like the Aurora or the Sony A7S II in goggle form? There are nontrivial hurdles, but maybe. Just maybe... I'm still not selling my DTNVGs. 
|
|
Quoted:
Another interesting tidbit was their answer to a question about latency between the Aurora camera and its viewer: "The view finder is running at 60fps. Latency between camera and viewer is sub 1 frame." Does this mean that navigation via helmet mounted Aurora camera is possible? As UNV and N-Vision infer, probably not. But, we are getting close. Plus you have to consider the price point is a mere $800 ($559 on kstart ATM). A Canis Latran PVS-14 clone this ain't! Actually it seems a little more along the lines of an affordable consumer action camera version of the Rochester/Photonis CNOD. Makes you wonder what technology like this applied to a NVG product at the $5k to $10k or even $20k price-point can get you. So is the future of night vision digital, with something like the Aurora or the Sony A7S II in goggle form? There are nontrivial hurdles, but maybe. Just maybe... I'm still not selling my DTNVGs. Surely the future is digital, but for a long time it might mean a sensor with a GaAs photocathode to get anywhere near starlight performance. My guess is color NV will be done with some other tech than a simple bayer matrix, something with which you lose almost no light. Like in an extreme case having 3 or 4 sensors, and splitting the light from suitable bands to each. |
|
Quoted:
I guess the sixty four dollar question is why can't they get conventional venture capital. If it's such a good idea then why aren't the big shooters getting in on it. If I had a business that I actually believed in and wanted to get going, the absolute last option would be venture capital. Considering how low interest rates are, an equity loan would be far more sensible |
|
Quoted: In many situations, it's not that they cant get venture capital....it's that its frequently a huge ripoff. This is essentially a preorder, and the company doesnt have to give away a massive percentage of the company. If I had a business that I actually believed in and wanted to get going, the absolute last option would be venture capital. Considering how low interest rates are, an equity loan would be far more sensible |
Armory Sponsor