Will the Quattro Embarrass Sigma into dropping their camera line?

Started Jul 21, 2014 | Discussions thread
OP maxotics Regular Member • Posts: 320
Re: My Conclusion

I'm no video expert, so maybe I shouldn't be commenting about this . . . but doesn't compression happen on the fly with current video capable cameras? I mean my Sony A55 would compress 1080i60 video to pretty small file sizes. I think it was creating video that could be played in a 16mbit stream. That's VERY low . . . only 2 MB/sec. You're talking about 100 times that. With good, fast, lossless compression that should not be necessary. Even if it is . . . we have what we need:

I'd like to say you sound like an idiot (which you do in the video realm), but I've been there myself These are very difficult concepts to get straight, at least they were for me.

The biggest blunder I made was not understanding that each pixels on a bayer camera only reads one color of light (roughly), red, green or blue. So bayer cameras already use a pseudo compression scheme by only recording 1/3rd the color value for every pixel. INCOMPLETE color information is exactly what makes bayer cameras inferior to Foveon cameras. However, getting 3 colors on the Foveon extracts a very high price. Each layers cuts down on the amount of light so in the end, you need a lot of light, which makes them limited to low ISO work.

So to match up a Foven to a bayer camera we'd take 1 color from each pixel, borrow 2 of the other colors form neighboring pixels, and construct our RGB value for each pixel. But if we do that we lose all the advantage of Foveon (full color values for each pixel) and gain none of the high-ISO advantages of the bayer. So NO point in that.

Therefore, to get the most out of the Foveon sensor we need to combine all 3 color values at each pixel into an RGB value, then compress those into a video stream. I believe you are right, there is NOTHING THEORETICALLY preventing Sigma from creating a good video stream from the sensor. It just needs a processor 3-times as strong. If we simplify. Again, remember, bayers work with 1/3rd the color information as Sigmas.

I don't know the answer to why Sigma can't create video. What I do know, is that they CANNOT quickly create accurate RAW RGB values from the sensor. That is the FIRST step. Once you push the shutter, the camera seems to save each pixels separate channel values but then must run various algorithms to figure out what the separate RGB values are; that is, Foveon needs a lot more RAW processing than bayers between the sensor read-out and digital saving. I know part of the problem is that the red layer, having received the least amount of light, needs to be smartly "amplified" to bring it in line with the green and blue values.

The closest analogy I can think of it when you move your eye from something brightly exposed to dimly exposed. Not only does your iris open or close, your brain starts to "fill in" missing information. Sigma cameras must do that with each layer. The Foveon may get a good "blue" value, but the green and red values are dirty and must be cleaned. More computation, more power, more battery needed, more heat created.

Again, I don't know the specifics, only what seems to be the case. It isn't video compression that would hang up Sigma (there are algos galore). It's the speed in which Sigma can process each RAW frame/image from the sensor. (What you see on your LCD is just a quick, low FPS guess that wouldn't look good blown up).

Perhaps others can continue...this stuff is exhausting to explain

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow