For too long taking great photos was a chore: use a tripod, a shutter release, a gray card, and only shot landscapes at dawn or dusk. SSLR bring the fun back to photography. Just go out and enjoy taking photographs. Let SSLR’s computer vision wizardry take care of the rest.
Showing posts with label SSLR. Show all posts
Showing posts with label SSLR. Show all posts
Saturday, March 2, 2013
Friday, March 1, 2013
SSLR - The fun part
I love computer vision, and this projects oozes it. Doing super resolution imagery, HDR, and clever white-balancing all sounds like a lot of fun.
Thursday, February 28, 2013
SSLR - The growth potential
In 2015, total global revenue for digital cameras is expected to be $43 billion. Of that $43 billion, around $17 billion is expected to be digital SLRs. If SSLR was able to grab 10% of the market, that would mean it would have almost two billion dollars in revenue per year.
Wednesday, February 27, 2013
SSLR - The monetization
This is pretty straight forward. SSLR makes cameras and sells them to customers. SSLR would also likely bundle free software with the camera to make post processing easier.
Tuesday, February 26, 2013
SSLR - The idea (part 2)
Shutter speed
One effect that usually requires a tripod is taking very long exposure images, e.g., night shots of traffic or waterfalls. Like tripod use in general, this could be handled by taking many short exposure images and creating the blurring by using super resolution techniques (aligning and averaging).
Landscape Photography and HDR
Landscape photography can be amazingly beautiful. Unfortunately, it is also a huge pain in the ass. You basically have two times of the day where you can take good landscape photographs: dawn and dusk. The reason is that during the day, the sun casts harsh shadows on everything making them look pretty ugly. At dawn and dusk, there isn’t any direct sunlight and everything looks evenly lit.
Now, using software we can get almost the same effect during the day. During day shots, there are areas of the scene that are directly illuminated by the sun and others that are indirectly illuminated (and much much dimmer). Using widely different exposure times, it’s possible to capture all of the detail in the indirectly illuminated regions as well as the directly illuminated areas. Better yet, if a surface or texture is directly and indirectly illuminated then it’s possible to match the colors and make the entire surface look like it’s indirectly illuminated. The result is that everything looks like light is indirectly hitting it like it would be during dawn or dusk.
There are some caveats here though. The biggest one is that light that hits a surface can reflect onto other ones, e.g., wearing a green shirt can make your face look a little green. This technique cannot account for that.
This HDR photography is already available on current smart phones. So, it’s really not really a stretch. The difference between what’s used right now and my idea is that I hope to fully compensate for lighting not just make it look a little better.
White Balance
Many DSLRs today comes with GPS sensors, which means the cameras know where they are. If you are outside, the camera could know that and set the white balance accordingly. If you are inside, then it could know that too. Also, if someone else has been inside the same building at roughly the same location and taken a photo, all of that information could be used to determine the white balance. Lastly, if you take a picture of the same thing many times, say a friend who you drag along on your photo adventures or a hat, you can use how those objects look in different photos to help figure out the white balance.
One thing that all of these features have in common is processing. The SSLR camera will need to come with a beefy processor or graphics chip set.
One effect that usually requires a tripod is taking very long exposure images, e.g., night shots of traffic or waterfalls. Like tripod use in general, this could be handled by taking many short exposure images and creating the blurring by using super resolution techniques (aligning and averaging).
Landscape Photography and HDR
Landscape photography can be amazingly beautiful. Unfortunately, it is also a huge pain in the ass. You basically have two times of the day where you can take good landscape photographs: dawn and dusk. The reason is that during the day, the sun casts harsh shadows on everything making them look pretty ugly. At dawn and dusk, there isn’t any direct sunlight and everything looks evenly lit.
Now, using software we can get almost the same effect during the day. During day shots, there are areas of the scene that are directly illuminated by the sun and others that are indirectly illuminated (and much much dimmer). Using widely different exposure times, it’s possible to capture all of the detail in the indirectly illuminated regions as well as the directly illuminated areas. Better yet, if a surface or texture is directly and indirectly illuminated then it’s possible to match the colors and make the entire surface look like it’s indirectly illuminated. The result is that everything looks like light is indirectly hitting it like it would be during dawn or dusk.
There are some caveats here though. The biggest one is that light that hits a surface can reflect onto other ones, e.g., wearing a green shirt can make your face look a little green. This technique cannot account for that.
This HDR photography is already available on current smart phones. So, it’s really not really a stretch. The difference between what’s used right now and my idea is that I hope to fully compensate for lighting not just make it look a little better.
White Balance
Many DSLRs today comes with GPS sensors, which means the cameras know where they are. If you are outside, the camera could know that and set the white balance accordingly. If you are inside, then it could know that too. Also, if someone else has been inside the same building at roughly the same location and taken a photo, all of that information could be used to determine the white balance. Lastly, if you take a picture of the same thing many times, say a friend who you drag along on your photo adventures or a hat, you can use how those objects look in different photos to help figure out the white balance.
One thing that all of these features have in common is processing. The SSLR camera will need to come with a beefy processor or graphics chip set.
Monday, February 25, 2013
SSLR - The idea (part 1)
Most of the problems outlined above can be solved with technology. I’ll take them on one at a time. One thing that is important for most of these solutions is the mechanics of the digital SLR camera. Normal SLRs have a mirror that moves up and down to change where the light from the scene is going (either up to the viewfinder or down to the image sensor). This means that you can’t be looking at the scene and taking a picture at the same time. Also, it limits the rate at which you can do continuous shooting (if you still get to look at the viewfinder between shots). An example of SLR-like cameras that don’t have this problem are the new Sony cameras. The only moving part in these cameras is the shutter. Also any MILC camera could also work. That’ll be important very soon.
Tripods and shutter release
Tripods are considered critical for getting ‘tack sharp’ photographs. Every small movement can cause blurring, especially when using a zoom lense. If you are really serious, you use a shutter release to take the picture, even the act of gently pressing the button can cause blurring.
To take really sharp shots, you don’t need all that stuff. What you need is for the camera to take dozens of shots with a short exposure time, say 1/50 or 1/100 seconds, and then use super resolution techniques to align the images and produce a single ‘tack sharp image.’ Also, to improve the sharpness of individual image, one could put accelerometers on it and do de-convolution using the accelerometer data.
Tripods and shutter release
Tripods are considered critical for getting ‘tack sharp’ photographs. Every small movement can cause blurring, especially when using a zoom lense. If you are really serious, you use a shutter release to take the picture, even the act of gently pressing the button can cause blurring.
To take really sharp shots, you don’t need all that stuff. What you need is for the camera to take dozens of shots with a short exposure time, say 1/50 or 1/100 seconds, and then use super resolution techniques to align the images and produce a single ‘tack sharp image.’ Also, to improve the sharpness of individual image, one could put accelerometers on it and do de-convolution using the accelerometer data.
Sunday, February 24, 2013
SSLR - The motivation
Taking great photographs can be really hard. If you read Scott Kelby’s book, it’s hard not to come away a little depressed. You should use a tripod and a shutter release. Be sure to set ISO to 100. You should probably get a neutral grey card to get the white balance right. If you’re interested in landscape photography, you only have two times a day to shoot, dawn and dusk.
Does it really need to be this hard?
Does it really need to be this hard?
Subscribe to:
Posts (Atom)