Reading mode: Light Dark

Quattro and AFE

Go to last post
mike earussi said:
You guys lost me in the discussion. But coming from Sigma I just take it to mean the implementation of the AFE is similar to what was done on the SD15 as opposed to the SD14 and Merrills, hence the easily overexposed highlight problem similar to the SD15.
Click to expand...
I assume so. But ... the highlight clipping problem should only be a problem for ISO 400 and above.
 
xpatUSA said:
xpatUSA said:
The interesting stuff here is that Quattro amplifies the signal when changing ISO. And an important question is if we can show that the amplification is analog. It could be digital for all we know.
Click to expand...
Roland, you say this after I presented a block diagram and a link to a highly credible reference? But I do realize that your expertise is in software, so please allow me to explain:

The amplifier is in front of the ADC; therefore it's output must be analog. the amplifier receives an analog signal from a Foveon channel output which is also analog. Therefore the amplifier is analog, not digital.
Click to expand...
 
xpatUSA said:
Roland, you say this after I presented a block diagram and a link to a highly credible reference? But I do realize that your expertise is in software, so please allow me to explain:

The amplifier is in front of the ADC; therefore it's output must be analog. the amplifier receives an analog signal from a Foveon channel output which is also analog. Therefore the amplifier is analog, not digital.
Click to expand...
I am quite able to understand hardware :P

Correct me if I am wrong. The block diagrams you presented were general block diagrams from some AFE chips - or was they the actual chip used in the Quattro?
 
Last edited:
Roland Karlsson said:
xpatUSA said:
Roland, you say this after I presented a block diagram and a link to a highly credible reference? But I do realize that your expertise is in software, so please allow me to explain:

The amplifier is in front of the ADC; therefore it's output must be analog. the amplifier receives an analog signal from a Foveon channel output which is also analog. Therefore the amplifier is analog, not digital.
Click to expand...
I am quite able to understand hardware :P
Click to expand...
Glad to hear it, I was beginning to wonder ;-)
Roland Karlsson said:
Correct me if I am wrong. The block diagrams you presented were general block diagrams from some AFE chips . .
Click to expand...
I detect "spin" :-) Actually, they were specific block diagrams for specific AFE chips.

To refresh your memory:

fig05.GIF


AD9814 and AD9822 are chip model numbers which, as one with an understanding hardware, you will have realized already. They appear to be for scanners:

"The 14-bit AD9814 and AD9822 , operating at 6 and 30 MSPS, are at the high end of an Analog Devices family of three-channel AFEs with 10- to14-bit resolution and sampling rates from 6 MSPS to 30 MSPS."
Roland Karlsson said:
or was they the actual chip used in the Quattro?
Click to expand...
No. Someone needs to disassemble their Quattro and publish some good pictures of the main board. Then our discussion will be over.

--
Cheers,
Ted
 
Last edited:
the bman said:
Scottelly said:
Did you read the French article about how ISO 125 is best for shooting blue stuff on bright sunny days, because it makes the least noise and captures the greatest dynamic range, but ISO 250 is best for dynamic range and noise in the green channel?
Click to expand...
How do you choose ISO 125 or 250?

--
Regards,
Bharat
-------------------------------------------------------------------------------------
http://be.net/bharatdarji
http://www.facebook.com/bharatphoto
http://www.flickr.com/photos/28113257@N04/
Click to expand...
Just look at this noise graph from that French article:

Noise graph of the various color channels/layers of the Quattro (from the French article by Jean Pierre).

Noise graph of the various color channels/layers of the Quattro (from the French article by Jean Pierre).

Here is the article translated by Google:

Loading…

translate.google.com

Here is the original . . . someone posted it near the beginning of this thread:

Loading…

sigmadp2quattro.blogspot.fr

Now you can see that the noise at 125 is pretty low in the blue channel. Take a look at the dynamic range chart now:

Dynamic range of Quattro sensor (from the French article by Jean Pierre).

Dynamic range of Quattro sensor (from the French article by Jean Pierre).

You can see that the peak dynamic range is at ISO 125 and 250 (until you get to high ISO settings, where the noise is very high, in comparison to ISO 125 and 250). The relationship between dynamic range and noise in various color channels is what makes me say that the Quattro is best for blue things on bright sunny days (flocks of birds, airplanes, power lines, trees, and skylines, etc. against clear blue skies, blue flowers, blue pools with blue tiles, etc.), and forests (lots of green stuff at ISO 250). Forests are dark, so ISO 250 is better for that than ISO 125. You might think ISO 400 would be better, because there is so little noise, but the dynamic range is not so good at ISO 400, as you can see by the dynamic range chart (that is a low point in dynamic range).
 
xpatUSA said:
Roland Karlsson said:
or was they the actual chip used in the Quattro?
Click to expand...
No. Someone needs to disassemble their Quattro and publish some good pictures of the main board. Then our discussion will be over.
Click to expand...
hehe

Thought so.

We can guess ... and we can do educated guesses. The most reasonable guess is that Sigma uses some kind of AFE chip and that they do change the actual analog amplification in the chip whenever the level changes in the stored data in the X3F file. Why should they do otherwise? Still ... it is a guess.

We have one (stupid?) example where this guess would be wrong.

Many (most?) cameras do not add any analog amplification over a certain ISO. So - maybe they do amplify from ISO 100-6400, but for ISO 12800-51200 they simply do digital amplification. That they do the digital amplification is probably to simplify life. Then you do not have to save any meta data telling a multiplication factor. Something Sigma has to do to support the non amplified data.

Now, doing digital amplification is just plain destructive. You gain nothing, but lose highlight headroom. Totally unnecessary.

Maybe they simply did forget to add a multiplication factor when designing the RAW format (and DNG?).
 
Roland Karlsson said:
We can guess ... and we can do educated guesses. The most reasonable guess is that Sigma uses some kind of AFE chip and that they do change the actual analog amplification in the chip whenever the level changes in the stored data in the X3F file. Why should they do otherwise? Still ... it is a guess.
Click to expand...
 
jande9 said:
Roland Karlsson said:
We can guess ... and we can do educated guesses. The most reasonable guess is that Sigma uses some kind of AFE chip and that they do change the actual analog amplification in the chip whenever the level changes in the stored data in the X3F file. Why should they do otherwise? Still ... it is a guess.
Click to expand...
Click to expand...
 
Roland Karlsson said:
We, or at least I, have assumed that Quattro has no AFE.
That is, AFE in the meaning of Sigma vocabulary, i.e. that
the data is not amplified before stored digital.

Now - this is not the case!

I have made measurements at ISO 100, 200, 400, 800 and 1600.

I used the same aperture and exposure time and only changed ISO.

ISO 100 and 200 got exactly the same values, let us call that the base amplification.

ISO 400 had 2x base amplification, i.e. the values were twice as high.

ISO 800 and 1600 had 4x base amplification.

Somewhat confusing IMHO.

In my test image this means that the data that was intact in ISO 100-400 were actually clipped for ISO 800 and 1600.
Click to expand...
 
Johan Borg said:
BTW, the binary distribution of x3ftools isn't updated with Quattro support, yet, is it? ;-)
Click to expand...
Not sure, what you tried to do, but extracting JPEGs out of Quattro RAW worked fine for me (unix binary)!
 
Johan Borg said:
Did you try 125, 160, 250, 320 yet, to see if variable AFE really only kicks in from ISO 400?

BTW, the binary distribution of x3ftools isn't updated with Quattro support, yet, is it? ;-)
Click to expand...
No ... it isn't.

Major reason is because it can't read meta data. Sigma has changed the meta data encoding again.

But, ill fix binaries. Stay tuned!
 
Hi,

I used some time testing the AFE in the Quattro.
This is my conclusions.

1x: ISO 100-320
2x: ISO 400-640
4x: ISO 800-6400

I also did a sanity check of my Merrill and it did not have any AFE, everything from ISO 100 to ISO 6400 is 1x amplification.
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top