A concept for a digital camera

Cameras these days have incredible intelligence algorithms. I think Sony or Samsung has a camera that not only recognizes human faces, but also can be programmed to take pictures when that face smiles. What if you could use that intelligence to interpolate and use high quality anti-aliasing to create super high quality pictures that don't necessarily have high megapixels by taking several pictures taken and grabbing the integral data and eliminating the noise?

There are lots of high speed cameras these days. Say with this new camera, when you hit the click button, it takes a picture, like you would expect it to. But really what happens is in that split second it takes 40 pictures and stores them all in the internal memory. It then uses one of those pictures as a model picture, for the very instant of time that that picture should represent. And then the software inside the camera looks at and analyzes all of the 39 other pictures, removes defects and noise that result from the CCD or CMOS sensors that the camera is using. It creates an idealization, sort of like what our brains do. Think about the incredible quality that could be created in low light conditions. After it makes that idealized picture, it then deletes all of the other pictures and then you're ready to make another picture.

That would be a badass camera. I want one. Someone make it, please.

 

Log in or Sign up to hide all adverts.

Then what would we need professional photographers for? Imagine a camera that does what you want and makes a "perfect" picture everytime you take a photo. What would the professionals do then? :shrug:

 

Log in or Sign up to hide all adverts.

Photography is more than taking pictures that are high quality. A shitty photographer with this camera would still make shitty pictures in terms of composition, although they would still be super high quality in terms of clarity and light.

 

Log in or Sign up to hide all adverts.

Your question is like "What's the point of buying a BMW? What would be the point of having good drivers?" Makes no sense.

 

But you can get that today with the 30 meg digital camers that are being sold. All you need do is just resize them to alter them in any way and then resize that to anything you care to print out.

 

Not really. There are all kinds of problems with the quality of pictures of cameras that are super high megapixels. You get serious diminishing returns when you increase megapixels. The photocells get smaller, they absorb less light. They introduce more imperfections, more noise. My concept is totally different.

 

I see. So instead if megs you use a new, improved algorithm to adjust for everything.

 

I agree.

I'm not a great photographer but I have taken some nice images with a crappy Casio just because I happened to snap a nice photo by sheer luck. Well, I used to live with a photographer so I had some hints about how to take a good pic.

 

Well, that would be ... impossible. Your "split second" can actually be 1/2500s, 1/4000s, or faster shutter speeds especially when photographing sports, race cars, hair blowing in the wind, etc.. The best pro SLRs today can probably achieve close to 10 frames/sec. The main key to low noise in low light conditions is LARGER SENSOR. Increasing sensitivity (ISO) on larger, more disperssed photocells will result in lower noise & better IQ (image quality). This fact unfortunately, is not known by the average consumer. What they're looking for is simply higher MP thanks to bullshit marketing hype. The IQ of some newer point & shoot cameras have actually suffered because of this (ex 12 mp on tiny sensor). The manufactures compensate by using software & firmware techniques in the camera to "adjust" the photo output (jpg) giving the impression of a good photo but not representing the true characteristics .
Real photographers & critics judge camera performance by the RAW output images (not processed). They can then, be downloaded into photoshop (ACR) or other software & adjusted to the photographer's preference to create the jpg.

Something possibly related to what your talking about (not to reduce noise) .. but to acheive an interesting effect is called High dynamic Range (HDR) processing. This is when auto-bracketing is used to taking pictures at different Exposure Values (norm, +EV, -EV), then downloaded on to a computer and processed together to acheive a higher dynamic range. This can only be done with a tripod and shooting non moving objects.

 

dsdsds, nothing is impossible.

What I'm talking about is a totally different kind of digital camera. While it may be true, like you say, that the best pro SLR cameras achieve no more than 10 frames per second, many other cameras can do a lot more than that. Many cameras these days can take pictures extremely quickly--some can shoot quickly enough to film bullets. There are many HD cameras that film at at least 30 FPS and in high def. The point I'm making is that each individual frame would not have to be high quality. The sensor itself could be one or two megapixels, and then in one second it could take 30 pictures or so. Then the software can take One of the pictures and then compare it to the others and use its intelligence algorithms to decide what is ideal and then draw a super high resolution high quality model from it. That model wouldn't be reality. It would be better.

 

Have you seen the quality of pictures taken from good photographers and from good cameras lately? They are much better than any high def quality you're talking about. Improved image quality will not come from algorithms & processing. Image quality is directly related to the following (in order of criticality):

1. experience & know-how of the photographer
2. lens quality
3. camera (most importantly - sensor technology)
4. Post processing (whether done in-camera or on pc)