| Snapdragon 8 Gen 2 will deliver professional quality photos and videos expect these professional quality |
| cameras and smartphones in 2023 all right |
| so yeah expect this sensor out in 2023 thanks very much |
| okay let's talk about even another image sensor technology |
| prophecy is is a company that has developed a really unique image sensor |
| called an event-based Vision sensor and it's really unique because |
| our world does not actually operate in 30 frames per second or 60 frames per |
| second with a scanning raster it doesn't do that um our eyes our visual system perceives |
| things much more uh detailed with a lot better temporal resolution and so these |
| image sensors are quite unique in that they don't they're not fixed to a certain frame rate for example so and there's |
| some benefits to that so the prophecy image sensor has a lot of really unique characteristics the first is that it can |
| run at an equivalency of 10 000 frames per second because this image sensor like I said is not locked to like 30 or |
| 60 frames it outputs data only when there's an event or a change in the |
| scene so the essence of the scene is captured only the changes are captured this can |
| lead to much less bandwidth coming out of the sensor because if you have the camera pointed at a static scene why |
| output any data you know why why waste battery life doing that if the scene hasn't changed so this sensor actually |
| does that it also can work in extreme lighting conditions this sensor can be programmed |
| to work at you know 120 DB of dynamic range huge amount of HDR dynamic range and |
| finally it works at really really low power so it's a quite unique technology we're working with prophecy to bring |
| this technology to smartphones and so I'd like to play a video for you from Luca who is their CEO the video was in |
| French but it has subtitles so please take a look |
| there is a |
| at the risk like this is |
| foreign |
| really interesting and unique technology um and it works a lot like the human eye |
| which I think Luca mentioned here but you might ask yourself why wait what's |
| what's the purpose of this why would this be important for a mobile application so let me show you |
| you see here that we're using an RGB camera a standard camera on the left to capture someone playing tennis and on |
| the right is the image that comes out of the prophecy event-based Vision sensor and so like I said the one on the left |
| the RGB camera operates at a fixed frame rate 30 or 60 frames per second |
| the prophecy image sensor operates 1000 times faster than that so the prophecy |
| image sensor can actually output only pixels that change and they can output them really really quickly so |
| as far as when the person swings the racket it's sampling 10 000 times more data than the RGB sensor is so the the |
| movement the fine the the really tiny movements are are captured so much better and again why is this important |
| for smartphones the reason why this is important is because if you take an action sports |
| shot and you use a standard image sensor that's locked to a certain frame rate you get an image like that where the uh |
| the actual uh static image has motion blur so the one on the left where the racket is in motion it's completely |
| blurred and there's no easy way to get around that unless you up the frame rate |
| but then that lowers your light and it's you know becomes a real problem but if you can intelligently combine the data |
| from the RGB sensor with the event based sensor from prophecy fuse those together what we can do is we can apply that that |
| tons of motion data from the proxy sensor to the RGB side and then get rid of the motion so this will get rid of |
| all the motion blur in in your shots and so the good news is that we're working |
| really closely with prophecy to bring this motion reduction technology for static images into smartphones so you'll |
| see this really soon really cool technology so let me shift gears a bit and talk |
| about bokeh portrait video and portrait photography last year on 8gen one we |
| talked about how we have a hardware-based bokeh engine and that engine uh was capable of creating these |
| blur shapes completely in Hardware at low power and it worked for video |
| now on Snapdragon 8 Gen 2 we've taken this engine and we've taken it a step |
| further so we said to ourselves why not modify the bokeh engine so that it could |
| simulate different F numbers of lenses so something from F14 all the way down to F 1.4 and when you do that when you |
| when you simulate a larger aperture what you end up with is bouquet shapes that are a lot bigger a lot more intense |
| so that's one of the cool things we can do live with this technology that's now in |
| Hardware in Snapdragon 8 gen 2. and so we actually oh this is first a video |
| here intensity of out of focus sources of light in the background or change the effect entirely with radical new bouquet |
| shapes that are Hardware accelerated to save power now did you see that not only can we |
| change the bokeh size based upon the F number a simulated enough number we can change the shape so this is really going |
| to be cool for people who want to do fun you know videos or selfie photography where they want to change the the |
| standard kind of blur circle into a diamond into a heart Etc so to show you |
| that this is real I have my esteemed colleague Jeff here and we're going to show this actually |
| live so take a look at the screens here so you see behind him we have simulated |
| Christmas lights behind him and you can see here that he can change the shape of those blurred lights |
| from circles to Diamonds to hearts or he can change the intensity of the of |
| the bouquet so back to the standard with no bouquet at all he can sweep and go right back between you know kind of a |
| very subtle effect to a very extreme effect with a really fast lens |
| and again you see this in the viewfinder this is happening in Hardware in real time 60 frames per second and can be |
| shot with video or photo so pretty cool stuff and you can only do |
| this in Hardware thanks Jeff |
| okay let's also talk about ultra low light in the past I've talked about our |
| low light engine and how we've been able to really reduce noise efficiently to get really really good performance in |
| low light |