What the imager has

Started Feb 11, 2014 | Discussions thread
Raist3d Forum Pro • Posts: 37,908
Re: Color Resolution Zen

Kendall Helmstetter Gelner wrote:

Raist3d wrote:

Kendall Helmstetter Gelner wrote:


But color bleeding is a case were you end up with a made up color. There's also color accuracy (for areas of color that have a higher frequency that the sensor can sample).

Color bleeding is not a made-up color. Color bleeding is color from one area possibly slightly going into another. It's a real color from the image, shifted.

What makes it "made up" is if the color does not represent the image aligned with the detail. Picture a case for example of an object that ends in detail at X/Y position, but the color being lower resolution overspill by one X position to the right. In that case, as far as the quality aspect for the image goes it will be "made up."

But again, this is only a problem at the high resolution, not in the low raw.

In a bayer sensor you get red/green/blue pixels out of the blue, in areas they never were - all because of simple patterns. That is made up.

You get on average blur but not a spill of this kind where the detail (luminance) ands and one pixel over carries. This is what I mean by "made up"- though different, because the final scene shouldn't have that over spill.

But because of the top layers also getting some values the lower layers get, you can eliminate a lot of color bleeding kinds of effects simply by having a clear picture of where the boundaries are.

You won't be able to make that determination like that due to the lower sampling


Refer to Dicks' post where he in no uncertain terms talks about lower chroma resolution. Now, there are strategies to try work this out that I would imagine Sigma will try. They won't work all the time but they can help.

, though there are certainly strategies that Sigma could purse based on human perception and pattern recognition (several Bayer cameras are already doing stuff like that).

Double that last statement, since what you can do is nothing like what they are doing.

If what they are doing is the blend with the luminance as Dick explained by pointing out to the Photoshop blend of two layers with the lower color rez, this can apply. BTW, I am not saying they are doing that in that paragraph just above, I am saying they could pursue some different options here.

In the end the result should be way closer to 19MP of color resolution than 5MP.

For color I don't think so

, but for perception of detail yes. There are strategies that can be pursued but they can't make up data that doesn't exist.

Since the data is all captured it need make up nothing, it just has to uncompress it.

Dick mentioned in no uncertain terms what I have said already- you have 1/4th chroma color for the most part (taking out whatever color wavelength bias the first layer has). Vs a traditional foveon sensor, not all the spatial data per layer is captured. You capture averages in the lower two.

This is why Dick like what I pointed out also, says that the chroma resolution is less. He mentioned it's a good tradeoff, and I also mentioned I think it will be a good tradeoff.

If I add four numbers together, then subtract three of the original numbers, I can still get the fourth value out.

Yup but that is not what is going on here. You have 4 numbers from the top layer in a spatial area, and 1 number from the bottom layers. You can do all the subtractions you want, but the numbers you get that involve math from the bottom layers, involve a sample of one number, which represents the average value for that area that in the top is 2x2 (per layer, per spatial area of the 2x2 above of the top layer).

It's at that point (capture/sampling) you already lost color resolution, because what you will be subtracting from each of the 4 values of the top layer the very same average value *FOR ALL 4*, not individual numbers for each.

In the previous Foveon design, you would have gotten 4 different numbers to subtract from the top layer, not one like now.  But the previous Foveon design also had some real world implementations that rendered some of that data unreliable, which is why this new design is pursued.

In fact the more original numbers I subtract the closer I am to the exact real value of the remaining numbers.

But as established above, you have less original numbers in this design.

I agree it should still look pretty good, but depending on subject/light/resolution of subject in those colors, it may not look as sharp.

The may is more like "rarely".

No more technical responses for me for some time, this is a waste of time.

Yes, apparently you can't understand the simple fact that this is loss of color resolution in two layers, which Dick pretty much stated. It baffles me that we are both saying the same thing and you just can't see it.

In fact what you are saying goes *against* what he said.

-- hide signature --

Raist3d/Ricardo (Photographer, software dev.)- I photograph black cats in coal mines at night...
“The further a society drifts from truth the more it will hate those who speak it.” - George Orwell

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