What is the sRAW in camera process

[AndrewRodney]

Originally Posted by Noel_Carboni

The one additional thing I can think of in this discussion that makes sRAW interesting is that it may well hold more latent dynamic range than any in-camera JPEG can deliver.

I suspect that is quite possible depending on the in camera JPEG from raw processing. But then, might as well shoot raw and have full rendering control. Seems like sRaw is splitting the baby in two with questionable advantages. That is, if you are concerned that the in camera raw to JPEG processing affects the range, among other aspects, why not just go full bore and work with the raw data?

People tout the benefits of Foveon cameras. Here we may have a well-kept secret in our own bodies... Foveon-like performance and/or reduced noise (with a downsized result, of course), as more sensor readings go into a given pixel.

How do you figure? The data captured from such CMOS chips and the Foveon are vastly different. The later appears to capture true RGB data unless I’m missing something.

[Noel_Carboni]

Originally Posted by AndrewRodney

How do you figure? The data captured from such CMOS chips and the Foveon are vastly different. The later appears to capture true RGB data unless I’m missing something.

I was building on Doug's point that true measurements of all the colors go into each pixel with the small version of sRAW. That makes it more Foveon-like, and has been pointed out in this very thread, the image quality is improved.

I can see using sRAW when you want the full benefit of the camera's dynamic range at the same time as not using up so much room on a card, when you don't need the increased resolution. Essentially it could make one of these "megapixel race" cameras a bit more like the lower pixel-count, faster sports-oriented cameras.

P.S., One additional point: I see no reason to believe Raw conversion SOFTWARE authors have done a better job than the people who invented the on-chip processing of Raw files. As was pointed out visually at the top of this thread, the quality the chips are delivering is hard to beat.

-Noel

[AndrewRodney]

Originally Posted by Noel_Carboni

I was building on Doug's point that true measurements of all the colors go into each pixel with the small version of sRAW.

The measurement of RGB data or luminance data from each pixel? I can’t see how it would be the former, the chip, unlike the Foveon, doesn’t capture such data or again, am I missing something?

[Noel_Carboni]

Yes, you are missing something.

Recall what the 's' in sRaw stands for.

Imagine that you have set the camera for the smallest sRAW setting. Here the word "smallest" stands for a reduction of pixels in the x and y directions.

Now imagine how many sensor photosite measurements, behind which bayer-patterned filters, went into creating each sRaw pixel.

- Noise is reduced

- More accurate information goes into each pixel.

- Ergo, image quality is improved for uses not requiring maximum resolution.

-Noel

[AndrewRodney]

Originally Posted by Noel_Carboni

Yes, you are missing something.

OK if you say so

Recall what the 's' in sRaw stands for.
Imagine that you have set the camera for the smallest sRAW setting. Here the word "smallest" stands for a reduction of pixels in the x and y directions.

Got that. Nothing earth shatteringly difficult to understand there.

Now imagine how many sensor photosite measurements, behind which bayer-patterned filters, went into creating each sRaw pixel.
- Noise is reduced
- More accurate information goes into each pixel.
- Ergo, image quality is improved for uses not requiring maximum resolution.

I would agree that if you take more initial data from the raw and sample it down, you get less noise due to the sampling. That would be true with a rendered image too. That’s quite clear.

What I don’t understand is the idea that this is anything like the way the Foveon chip captures its data (as true RGB, not interpolated from “Grayscale” data). Where is the “true measurements of all the colors go into each pixel“ coming from?

Lastly, from what you quoted above in terms of an advantage, how would any of this be different from working from the raw data with the added benefit of the user having control over the rendering?

[Noel_Carboni]

What kind of measurement do you call it then? False measurement? Fake Measurement? No measurement at all? Are we just imagining the images then?

One could point out that Foveon sensors could be considered to measure "true" color even less directly than a bayer-patterned sensor, in that the color from a Foveon sensor must be derived from measurements of RGB, RG, and R, while those "grayscales" you mention measured by a bayer-patterned sensor are each measured through "true" R, G, and B microfilters. Yes, I know that in some cameras they're RG, RB, and GB filters.

Please stop being argumentative. I am not interested in debating your interpretation of terminology vs. what a more typical individual would read from what I wrote.

-Noel

[AndrewRodney]

Originally Posted by Noel_Carboni

What kind of measurement do you call it then? False measurement? Fake Measurement? No measurement at all? Are we just imagining the images then?

Wouldn’t call them measurements at all but that’s just semantics.

The CMOS sensor is a photon counting device. The sensor with its bayer array is providing such a count with a single color filter, and then some assumption about a color space in the demosaicing process. Its not at all like the Foveon sensor.

One could point out that Foveon sensors could be considered to measure "true" color even less directly than a bayer-patterned sensor, in that the color from a Foveon sensor must be derived from measurements of RGB, RG, and R, while those "grayscales" you mention measured by a bayer-patterned sensor are each measured through "true" R, G, and B microfilters. Yes, I know that in some cameras they're RG, RB, and GB filters.

That’s exactly the point I’m asking.

Please stop being argumentative. I am not interested in debating your interpretation of terminology vs. what a more typical individual would read from what I wrote.

Take a stress pill Dave... I’m not being argumentative at all. I’ve simply asked some questions in general. There’s nothing argumentative about it. You are the one using the above language that’s being argumentative or lets just say hostile to questions that are not necessarily directed to you. Chill. I’m simply asking (again), for clarification of the statement
true measurements of all the colors go into each pixel with the small version of sRAW. What makes it true? Or any truer than the raw itself after rendering in any raw converter?