fredmiranda.com Login
This topic is in the 2010 archive. You can read it and like its posts here. It is read only.

Archive 2010Sigma SD1

  
 
Alf Beharie
Offline
Details
p.10 #1Sigma SD1
Mike V wrote:
I don't get the resolution argument?



The SD1 will have equivalent resolution to a 30mp Bayer sensor of similar dimensions but of course with no AA filter it will be even sharper....Thats all you need to know.

Oct 01, 2010 at 01:31 PM
Daniel Heineck
Offline
Details
p.10 #2Sigma SD1
Ummm.... no.

This might help Alf

Oct 01, 2010 at 01:44 PM
Alf Beharie
Offline
Details
p.10 #3Sigma SD1
Jorgen Udvang wrote:
Alf Beharie wrote:
I did'nt publish a conversion tutorial but basically you swap the C/Y mounting plate for a Sigma SA plate...There is no machining to do except the usual drilling securing holes...The C/Y mounting plate on the 85/1.4 is exactly the same diameter as an SA mounting plate so the SA plate fits on nicely. Then using the C'Y plate as a template you drill four new holes in the SA plate and contersink them, add the original Contax screws and your in business.



Jorgen Udvang wrote:
That sounds almost too good to be true. What about the register distance? Will that be correct as well??
.

Alf Beharie wrote:
Well SA mount reg' is 44mm and C/Y is 45.5mm, so it will focus a tad past infinty on a Sigma camera and therefore you should'nt trust the lenses distance scale.


Jorgen Udvang wrote:
Does Sigma sell mounting plates?


Alf Beharie wrote:
They have done in the past for a few users that said their lenses mount was damaged, but I dought they will now, though you could try asking them.



Jorgen Udvang wrote:
or will I have to cannibalize a lens?


Alf Beharie wrote:
Thats how I got some of my SA mounting plates. Others I bought from a fellow Sigma user who just happended to have some for sale...I bought four of them for £15 each as I recall.


Jorgen Udvang wrote:
Will it work on other lenses too, like the 35-70 or the 135mm f/2.8?


Alf Beharie wrote:
It may do but I can only confirm it will with the lenses I have already converted. I know another Sigma user converted the 50/1.4 Planar T* to M42 mount and another C/Y lens I cant remember now so I dont see why not.
Oct 01, 2010 at 02:06 PM
Alf Beharie
Offline
Details
p.10 #4Sigma SD1
Daniel Heineck wrote:
Ummm.... no.

This might help Alf



Ummm...Yes! This may help you...But it seems you totally ignored it before:


kwalsh wrote:
Daniel Heineck wroteTo make sure we're clear on this--from a spatial resolution standpoint, this camera should behave like a 16 mpixel. That's what you care about when thinking from a lens standpoint.
No, actually if you are comparing to Bayer cameras its resolving power will be more akin to about 25 to 30MP - see the earlier link posted regarding the resolving power of the older 4.5MP sensor compared to a Bayer.

So yes, this sensor will likely out resolve most glass.

Ken


Oct 01, 2010 at 05:08 PM
Alf Beharie
Offline
Details
p.10 #5Sigma SD1
Spyro P. wrote:
Ever since I tried a DP1 I am a Foveon believer.
I'd love to try this thing with some good legacy glass



Thats very simple...Buy an SD14 or SD15 instead then you can use loads of legacy glass with the same sensor.
Oct 01, 2010 at 05:12 PM
Alf Beharie
Offline
Details
p.10 #6Sigma SD1
garydavidjones wrote:
Owned several Sigma cameras--SD10, 13, and 14.


You owned the SD13?



Oct 01, 2010 at 05:28 PM
kwalsh
Offline
Details
p.10 #7Sigma SD1
Alf Beharie wrote:
Ummm...Yes! This may help you...But it seems you totally ignored it before:

kwalsh wroteNo, actually if you are comparing to Bayer cameras its resolving power will be more akin to about 25 to 30MP - see the earlier link posted regarding the resolving power of the older 4.5MP sensor compared to a Bayer.


To be fair, my estimates there are probably a bit high considering current AA filters and RAW converters. Back when that 4.5MP test was done against the 5D the AA filters were stronger and the RAW converters not quite as good. These days with RAW shooting a Bayer will get good luminance response without color moire up to around 85% of Nyquist. That'd make a "Bayer equivalent" number of about 22MP for the SD1. So my numbers in that post are probably to the high side unless you insist on shooting JPEG.

Ken
Oct 01, 2010 at 05:37 PM
Tariq Gibran
Offline
Details
p.10 #8Sigma SD1
kwalsh wrote:
Alf Beharie wroteUmmm...Yes! This may help you...But it seems you totally ignored it before:

kwalsh wroteNo, actually if you are comparing to Bayer cameras its resolving power will be more akin to about 25 to 30MP - see the earlier link posted regarding the resolving power of the older 4.5MP sensor compared to a Bayer.
To be fair, my estimates there are probably a bit high considering current AA filters and RAW converters. Back when that 4.5MP test was done against the 5D the AA filters were stronger and the RAW converters not quite as good. These days with RAW shooting a Bayer will get good luminance response without color moire up to around 85% of Nyquist. That'd make a "Bayer equivalent" number of about 22MP for the SD1. So my numbers in that post are probably to the high side unless you insist on shooting JPEG.

Ken


But actually, the 5D has a very weak AA filter.
Oct 01, 2010 at 05:39 PM
Yakim Peled
Offline
Details
p.10 #9Sigma SD1
Rob Riley wrote:
another way to do it is to junk an EOS body by removing the mount, and resoldering the electrical connections onto the the Sigma body (which are in a different order).


From: http://www.foto.nordjylland.biz/SD14/SD14-NonSigmaLenses.htm

It is however a pretty complicated conversion which requiers to open the camera.



Happy shooting,
Yakim.
Oct 01, 2010 at 06:04 PM
Alf Beharie
Offline
Details
p.10 #10Sigma SD1
Lotusm50 wrote:
garydavidjones wroteWith 46 MP crammed into APS-C
and

j.liam wroteNo knock on Sigma lenses because some of them are fine optics, but at a stated 46MP very few, if any can deliver quality optical information to do the sensor justice.
OK. Let's be clear on this, as it appears that Sigma marketing is marking some headway in perceptions. They are no cramming 46 mp in to APS-c. They are cramming 16mp on to APS-c. (each of these 16mp has a stack of 3 color receptors). So rather than having to resolve 46mp in an APS-c space, lenses really only have to resolve 16mp on an APS-c space.



I dont understand why this keeps cropping up because the way Foveon X3 sensor pixels are counted is actually exactly the same as Bayer sensor pixels are counted, ie, they count the total number of photodiodes on the sensor.
Sigma are not being inacurate in saying the SD1 has 46mp, because it does actually have 46 million pixels, being equal amounts of Red, Green and Blue over three layers.
However, Bayer sensors do not have equal amounts of red, green and blue pixels..50% are green, 25% are red and 25% are blue.
A fictional 46mp Bayer sensor would share the same percentages and so that translates as 23mp green, 11.5mp red and 11.5mp blue.
This fictional Bayer sensor only has higher resolution than the SD1 in the green channel but its has somewhat lower resolution than the SD1 in the red and blue channels...This gives a colour resolution advantage to the SD1.
The consumer is'nt told this fact when they buy a camera using a Bayer sensor so if anyones being inacurate...It is'nt Sigma/Foveon.
Oct 01, 2010 at 06:12 PM
Rob Riley
Offline
Details
p.10 #11Sigma SD1
like I said right off the bat
its spatially 15.3Mp with the detail information of 46Mp
Oct 01, 2010 at 08:49 PM
Daniel Heineck
Offline
Details
p.10 #12Sigma SD1
Sigh--kids, read page 4 and 5 again, perhaps you missed the response thesuede and I had for Ken. The three of us have a grasp on what's going on in the sampling differences between the two architectures, as does rhameed. You're making some very simple (and, honestly easy to do) mistakes on sampling theory. I mean no insult by that--we all have different specialties.

The newer Foveon sensor has 15.3 megapixels--each with 3 separate color readings at each pixel. A RAW decoding then requires RGB color reconstruction at each pixel site. Yes--that's a big deal and worth bragging about.

Ken and I may disagree on the *exact* differences in effective resolution, but that's no biggie, and it gives a range of expectations. This will all be hashed out when the product comes to market. It's not going to behave like a 46 Mp sensor in any way, shape or form. Full stop. If you want to argue, a bayer format would need to be derated in mpix equivalent, not the foveon promoted, and in either case, not by much, unless one wants to get into doing synthetic tests not based in reality.

An equivalent Bayer sensor has 15.3 megapixels, 1/2 of which are green, 1/4 of which are blue, and 1/4 of which are red. A raw file then requires a reconstruction at each site for both luminance and chrominance via some rather sophisticated interpolation algorithms. Yes, each bayer pixel contains less color information, but (generally) does its luminance job better than a foveon pixel. Chrominance information within a scene also varies much slower than luminance information, except in cases where moire is likely--textiles and the like. Have a look at the A and B channels of a LAB-converted file versus the L channel to understand.

Foveon has registered color, which is really cool, but it's not all lolipops and roses. Like for example, color fidelity (not resolution), noise, and leakage problems. Life is full of compromises--we'll see how it turns out. Scenes where color varies rapidly are to the Foveon's advantage. I hope, and look forward to a cool new option. As in, I hope it succeeds.

That said, it would take me a lot, lot, lot more work to go into the details of what's going on, but I don't get paid to be here
Oct 02, 2010 at 01:46 AM
kwalsh
Offline
Details
p.10 #13Sigma SD1

First off, everyone should listen to Daniel on this, it is an entirely accurate summary.

Second, one question/comment for Daniel.

Daniel Heineck wrote:
Yes, each bayer pixel contains less color information, but (generally) does its luminance job better than a foveon pixel.


I'm not sure in what context you meant the luminance comment in the above statement, maybe you could expand? Maybe the whole "what is a pixel" is the problem here. To be clear, per photodiode the Bayer does much better than the Foveon in spatial luminance sampling because each photodiode is at a separate spatial site in the Bayer array. Per spatial sampling site the Foveon does somewhat better than the Bayer because the Foveon doesn't have an AA filter or CFA to deal with.

Ken
Oct 02, 2010 at 07:06 AM
carstenw
Offline
Details
p.10 #14Sigma SD1
I don't know why this is such a hard topic: a pixel is a location on a screen. Whether there are three or one photosites below is irrelevant. The Sigma has 15MP, period. It has 46 million photosites, and if Sigma would just write that, there would be no controversy.
Oct 02, 2010 at 08:00 AM
Daniel Heineck
Offline
Details
p.10 #15Sigma SD1
kwalsh wrote:
First off, everyone should listen to Daniel on this, it is an entirely accurate summary.

Second, one question/comment for Daniel.

Daniel Heineck wroteYes, each bayer pixel contains less color information, but (generally) does its luminance job better than a foveon pixel.
I'm not sure in what context you meant the luminance comment in the above statement, maybe you could expand? Maybe the whole "what is a pixel" is the problem here. To be clear, per photodiode the Bayer does much better than the Foveon in spatial luminance sampling because each photodiode is at a separate spatial site in the Bayer array. Per spatial sampling site the Foveon does somewhat better than the Bayer because the Foveon doesn't have an AA filter or CFA
...Show more →

Ken, you caught a simplification I wrote. Well spotted

I agree with what your saying entirely. My point was more to the end that a bayer photodiode behaves a lot "nicer" because we aren't trying to extract 3 different color ranges from it. The photodiode well is less sensitive to incident angle, and generally larger in volume. It doesn't have the same level of cross-talk or noise that any one of the 3 photodiodes in a foveon architecture has as well. So whether it's reading red, green or blue, it's generally a much higher quality measurement on a per pixel basis, but loses that advantage in the demosaic process, as you well mention.

Hope that clears it up.
Oct 02, 2010 at 11:03 AM
kwalsh
Offline
Details
p.10 #16Sigma SD1
Daniel,

Understand what you mean now, spot on. One would expect the individual photodiodes in the stacked pixel to be at a performance disadvantage to the single layer photodiodes.

Ken
Oct 02, 2010 at 04:21 PM
phidong
Offline
Details
p.10 #17Sigma SD1
Do foveon sensors have any advantage correcting WB in post compared to bayer sensors?
Oct 02, 2010 at 05:31 PM
theSuede
Offline
Details
p.10 #18Sigma SD1
Not really. Actually, when you go out of the "comfort zone" of good D-quality light (or sunlight) the Foveon sensor behaves quite erratically from a colour perspective. It is in fact a physical impossibility to fully correct the colour in >6500K and <3500K light with the previous Foveons.

Active cell count is not the same as resolution. But that's pretty well established by now I would hope.

What you're REALLY after is spatial resolution. And the MTF70 of the Foveon 15MP sensor coupled with an optically "perfect" lens will most probably be around 1400lp/ph - giving a REAL resolution of about 12MP. Probably exactly the same (or a very few percent more) than the Canon 18MP sensors and the slew of coming 16MP solutions (Sony, Nikon, Pentax). The difference will be bigger with some static targets, namely dominant blue or red resolution test-targets. Those do however very rarely occur in real-life situations. Unfortunately (for Sigma) this advantage is only valid for <ISO400 and with non-spiky light spectras. As soon as you get to ISO400-800 all of the resolution advantage is gone.

Positives for the Foveon:
-Much nicer noise pattern, chromatic noise is perfectly scaled to pixel size (until you start correcting for the abysmal colour hue resolution, then it's actually worse than Bayers).
-Less (but not totally removed) risk of colour moire.

Negatives for the Foveon:
-Low sensitivity - noise performance is pretty poor
-Low colour hue sensitivity
-VERY high metamerism errors (bad for difficult lights like fluorescents and such)
-Very sensitive to light incident angles.

Oct 02, 2010 at 06:05 PM
ntotrr
Offline
Details
p.10 #19Sigma SD1
Yakim Peled wrote:
phidong wroteIf I recall correctly, Sigma uses similar electrical contacts to the Canon cameras as well. I've seen some Canon lenses converted to Sigma with full AF as well as some brave people converting some of the older SD cameras to have a Canon bayonet as well as other alt lens mounts.

http://sigmacumlaude.com/products.html

I wonder how much the SD1 is going to be.
This is incredible. I never new that such a thing exists. That is even further fortifies my question: Why don't they do it themselves?

Happy shooting,
Yakim.


Sigma is interested in developing their own system. They want to sell lenses, the profit is in lenses, not camera bodies. There won't be any SD cameras in any mount other than the Sigma SA mount. They don't license the Nikon, Canon, Pentax, etc. mounts for their lenses and there may be licensing issues if they try to make a camera with one of those mounts. Or maybe not, I really don't know anything about licensing issues.
Oct 03, 2010 at 07:51 PM
ntotrr
Offline
Details
p.10 #20Sigma SD1
Lotusm50 wrote:
Finally Sigma got around to increasing their sensor size. Still, they should have taken it up to full frame. 16mp is much better than where they were, and while not sector-leading, is finally enough for many. Not sure why they even bothered producing the SD15 -- a complete waste of time. Let's see if they have been able to increase high ISO performance. And, as implied above, Sigma's biggest problem is their lens mount. You are locked into a lens system that overall appeals to very few, that in spite of a couple interesting lenses is lackluster overall. But Sigma,
...Show more →

Sigma has far more than a "couple" of interesting lenses, far more like the 50mm EX DG, 70mm EX DG, 105mm EX DG, 17-50mm EX DG, 50-500mm, etc., etc.

Talk is that the SD1 will be priced to compete with the Canon 7D, not the D3/1D.

I don't see where the development of the SD and DP cameras has diverted attention and management from their lens line. Look at all of the new lenses they've introduced and the addition of OS to others. Their lens development is in full-swing.
Oct 03, 2010 at 08:02 PM
FM ForumsLeica & Alternative GearJoin Upload & Sell
  
 

You are not logged inLog inRegister