!!! SIGMA SD1 - 46Mpx !!!

[Oblidor]

I am considering a SD1, but like others need to see samples. I was not happy to hear they were using compact flash, why not SD cards? but can live with CF if they insist.

Isn't CF supposed to be faster than SD? If SD1 is writing 45Mb RAW files, it needs to have a extremely fast card not to use several seconds per photo to write.

 

[afshalders]

Yes but Nikon, Canon, Hasselblad, Mamiya and Pentax have some amazing lenses. I don't see any point on spending $6-7 K in a body that lacks in this subject.

I own some not so cheap Sigma lenses and besides the macro primes, they just can't deliver an image quality compatible with the SD1 price tag.

When you buy a Mamiya , Blad or Leica you know that the lenses are simply the best ones in the world.

 

[JOHNBLUE3]

Yes but Nikon, Canon, Hasselblad, Mamiya and Pentax have some amazing lenses. I don't see any point on spending $6-7 K in a body that lacks in this subject.

I own some not so cheap Sigma lenses and besides the macro primes, they just can't deliver an image quality compatible with the SD1 price tag.

When you buy a Mamiya , Blad or Leica you know that the lenses are simply the best ones in the world.

I think Sigma has many lenses already that are suitable to the SD1. The reason MF lenses are so much more expensive is that the lenses have to work into a much larger image plane. You may have noticed that lens mfgrs. specify their lenses in terms of their contrast and sharpness as a function of radial distance from the center or axis. The APS-C sensor(24mm on long axis) has a maxi distance that's shorter than FF sensors, even by a factor of 1.5. Even larger factor compared to MF sensors. So the lenses in this application can be seriously degraded at very large distances and still provide superb performance on the smaller sensor. If you look at the line pairs per mm on this 15MP/24mm sensor, it's ~64lp/mm. There are many many lenses that can resolve to this level so long as you are not too far off axis. To sum up, Sigma isin better stead on this point. Those SD15 users with good Sigma lenses won't have to go out and spend thousands on new lenses. As a case in point, Sigma, itself, thinks it has over 40 lenses that are SD1 ready.

 

[LUIS A GUEVARA]

I think Sigma has many lenses already that are suitable to the SD1. The reason MF lenses are so much more expensive is that the lenses have to work into a much larger image plane. You may have noticed that lens mfgrs. specify their lenses in terms of their contrast and sharpness as a function of radial distance from the center or axis. The APS-C sensor(24mm on long axis) has a maxi distance that's shorter than FF sensors, even by a factor of 1.5. Even larger factor compared to MF sensors. So the lenses in this application can be seriously degraded at very large distances and still provide superb performance on the smaller sensor. If you look at the line pairs per mm on this 15MP/24mm sensor, it's ~64lp/mm. There are many many lenses that can resolve to this level so long as you are not too far off axis. To sum up, Sigma isin better stead on this point. Those SD15 users with good Sigma lenses won't have to go out and spend thousands on new lenses. As a case in point, Sigma, itself, thinks it has over 40 lenses that are SD1 ready.

I do not know how you arrived at your ~64lp/mm. The new sensor has 4800 photosites (Pixels) over a Spatial distance of 24 Millimeters . This gives us 4800 divided by 24 equals 200 photosites per Millimeter . Since we need at leat 2 photosites to resolve a "Line" (One active photosite and one showing black), we are estimating a Calculated Resolving Power of 100 Lines per Millimeter . This is Leica-R territory.

We have not produced any mount replacement kit yet for the SD1 but we would like to make an offer to whomever can afford it and is willing to convert his SD1 camera to Leica-R:

If anyone is willing to collaborate with me in exploring the possibilities of making a Leica-R or Nikon F ,Mount replacement kit for the SD1 and is willing to remove and send me the original SD1 mount for analysis and measurements, I would make a mount replacement kit ,free of charge. For more info please get in touch with me at luis@sigmacumlaude.com Offer ends with the first person to accept it.

Sigmacumlaude is now also present in Facebook at:

SIGMA CUM LAUDE Group. (General photographic chat .)
Sigmacumlaude Atelier (Art , Black and White and tutorials)

Luis
SIGMA CUM LAUDE
Please click here to visit our new Forum
--
http://photo.net/photos/Luis-A-Guevara
http://www.pbase.com/luis_a_guevara/galleries
http://gallery.leica-users.org/v/LUIS+A+GUEVARA/
http://www.summiluxart.com/
http://www.sigmacumlaude.com/

 

[JOHNBLUE3]

Luis,

Please recall that the RGB photo sites are stacked vertically. That's the beauty of this thing. You get to increase the resolution dramatically without a proportional increase in lens reqirements. Visit the Sigma site if you wish. Clearly they are in the best position to know what is required in terms of lenses. They clearly think they have more than 40 appropriate lenses.

Also visit Carl Rytterfalk's site where he has posted several SD1 Raw files of photos which were taken with the well known 18-50mm/F2.8 lens. There are 5 photos zipped and pkg'd for download. Be aware thoguh that the pkg is 250MB in size. I think you will agree that stunning results are possible here without a lot of extra expense in lenses.
I know all of this can be confusing as there are a lot of myths about. But remember, the manufacturers know much more about their products than we mere photographers do.
Regards.

 

[LUIS A GUEVARA]

Luis,

Please recall that the RGB photo sites are stacked vertically. That's the beauty of this thing. You get to increase the resolution dramatically without a proportional increase in lens reqirements. Visit the Sigma site if you wish. Clearly they are in the best position to know what is required in terms of lenses. They clearly think they have more than 40 appropriate lenses.

Also visit Carl Rytterfalk's site where he has posted several SD1 Raw files of photos which were taken with the well known 18-50mm/F2.8 lens. There are 5 photos zipped and pkg'd for download. Be aware thoguh that the pkg is 250MB in size. I think you will agree that stunning results are possible here without a lot of extra expense in lenses.
I know all of this can be confusing as there are a lot of myths about. But remember, the manufacturers know much more about their products than we mere photographers do.
Regards.

Let's see:
Sigma SD9 had 2268 photosites over the same 24 mm giving a 94.5 photosites per MM or 47,25 Lines per Millimeter, so a Typical Sigma lens having around 65 line pairs /mm was more than adequate .

Sigma SD14-15 instead has 2640 photoreceptors over the same 24 mm equaling 110 photoreceptors per mm or 55 Line pairs/mm , this also fits comfortably with Sigma lenses capable of delivering about 65 Line pairs/mm.

The new sensor found in the SD1 has 4800 photoreceptors over 24 mm giving us 200 photosites per Millimeter or a Calculated Resolving Power of 100 Lines per Millimeter that WOULD EXEED THE SIGMA'S LENS 65 LP/mm . Like I said before this is Leica-R lens territory.

When using Sigma lenses of about 65 LP/mm on the SD1 there will certainly be a Perceptible improvement in Image Quality ,as a 100 LP/mm Sensor will use the Sigma lens to its limit of 65 LP/mm , something that the previous models couldn't do. In fact this new sensor will become Lens Limited , while the previous models were Sensor Limited.

If you examine an SD1 image area of fine detail this limitation will become clearly visible when some areas will fuse into one area of solid tone with no detail differentiation .

For example this impressive image published by Sigma itself at http://www.sigma-sd.com/SD1/sample-photo/ :

View attachment 2199

If you examine this portion of the image at full size and look inside the green rectangles you will see where the new sensor reached the limit of the lens and the bushes are showing as a BLUR:
View attachment 2200

You can examine this image full size here http://www.sigma-sd.com/SD1/sample-photo/img/Kendall-Gelner-SD1-SDIM0021r.jpg





This second image of a close up might show this effect better. The lens is able to image a single strand of hair , but in the regions where the same hair is PAIRED , it exceeds the ability of the lens to SEPARATE THEM , instead fusing two or more hairs into one undifferentiated mass .

View attachment 2202

The green rectangle area clearly shows were the low lens resolving power (65 LP/mm)is limiting the sensor (100 LP/mm).

View attachment 2201

You can examine this image full size here and judge by yourself:
http://www.sigma-sd.com/SD1/sample-photo/img/SDIM8118AA.jpg

If you had a Leica-R lens in the camera that area would be fully resolved however , under careful scrutiny , new areas of undiferenciated tone will appear WHERE THE SENSOR ITSELF REACHES ITS MAXIMUM RESOLVING POWER , UNABLE TO RESOLVE WHAT THE LEICA LENS IS PROJECTING.

This will happen at 100 LP/mm or above , showing considerably more detail , but not as much as film or a LARGER sensor or SMALLER PHOTOSITES would .

I think it is convenient to remember here that common FILM was able to resolve 160 LP/mm , while Fine Grain film could achieve 600 LP/mm or more..

Clearly this camera is quite an improvement on Resolving Power over the previous models , but it is also clear that BETTER LENSES ARE NEEDED.

We have not produced any mount replacement kit yet for the SD1 but we would like to make an offer to whomever can afford it and is willing to convert his SD1 camera to Leica-R or Nikon F for Zeiss ZF lenses:

If anyone is willing to collaborate with me in exploring the possibilities of making a Leica-R or Nikon F ,Mount replacement kit for the SD1 , and is willing to remove and send me the original SD1 mount for analysis and measurements, I would make a mount replacement kit ,free of charge. For more info please get in touch with me at luis@sigmacumlaude.com Offer ends with the first person to accept it.

Sigmacumlaude is now also present in Facebook at:

SIGMA CUM LAUDE Group. (General photographic chat .)
Sigmacumlaude Atelier (Art , Black and White and tutorials)

Luis
SIGMA CUM LAUDE
Please click here to visit our new Forum
--
http://photo.net/photos/Luis-A-Guevara
http://www.pbase.com/luis_a_guevara/galleries
http://gallery.leica-users.org/v/LUIS+A+GUEVARA/
http://www.summiluxart.com/
http://www.sigmacumlaude.com/

 

[JOHNBLUE3]

I did not mean to start out by getting involved in a big debate on the SD1. But I would suggest that you down load the Sd1 Raw files directly and view them using SPP5. Again, Carl has a set of five Raws that you can download. Look at those photos and show me the resolution shortfalls.
Now regarding the resolujtion requirements:
There are 15MP on the fron of the APS-c sensor with 1.5 aspact ratio. This means that, the number of lines along the x axis is given by,

(Nx)^2 = 15,000,000/1.5, ao that Nx = 3162 lines

When this is divided by 24mm, that's 132 lines per mm or 66 line pairs/mm

It's as simple as that. Now I am sure you know that just showing jpgs on the Web is no substitue for the real Raws, particularly when viewed on an excellent and well calibrated monitor. Please examine Carl's Raw files and bounce back.

Regards

 

[JOHNBLUE3]

I just want to post the URL to Rytterfalk's site:

http://www.rytterfalk.com/

 

[LUIS A GUEVARA]

I did not mean to start out by getting involved in a big debate on the SD1. But I would suggest that you down load the Sd1 Raw files directly and view them using SPP5. Again, Carl has a set of five Raws that you can download. Look at those photos and show me the resolution shortfalls.
Now regarding the resolujtion requirements:
There are 15MP on the fron of the APS-c sensor with 1.5 aspact ratio. This means that, the number of lines along the x axis is given by,

(Nx)^2 = 15,000,000/1.5, ao that Nx = 3162 lines

When this is divided by 24mm, that's 132 lines per mm or 66 line pairs/mm

It's as simple as that. Now I am sure you know that just showing jpgs on the Web is no substitue for the real Raws, particularly when viewed on an excellent and well calibrated monitor. Please examine Carl's Raw files and bounce back.

Regards

No . It is not "as simple as that". Digital sensors are Two Dimensional devices. They spread the total number of Photosites over the two spatial directions . The SD1 15 Megapixels per Layer , (of the Trilayered stack of photosites) ,are allocated as follows:

4800 Pixels in the Horizontal Spatial Direction of 24 mm , and 3200 Pixels in The Vertical spatial direction of 16 mm . This sometimes in Bayer cameras give rise to different Resolution in those directions .

According to Sigma published figures , the SD1 spatial resolutions will be :

• Horizontal= 4800/24/2=100
• Vertical =3200/16/2=100

This equality in the Horizontal and Vertical directions accounts in part for the excellent Definition of Foveon Images .

In cameras that have different resolving power in both directions , it is customary to use the Horizontal Resolution as being the Resolving Power of the Sensor , but when actually measured by shooting Target Images , for Optical determination of Resolving Power , actual measurements are given. Those have not been determined yet for the SD1 , or at least they have not been made publicly available, so for the time beeing we have to live with "Calculated" rather than "Measured" resolutions. Looking at different photographers results with the SD1 can serve to Rationalize somebody's Impulse to go and buy an SD1 , but will not change the facts published by the Manufacturer itself.

Sigma SD1 is a 15 Megapixel camera with a Resolving power of 100 Lines-pairs/Millimeter and an Image Quality of great definition. But it is still an APS-C sensor , although one of the best made so far. That might change when the new models from Sony , Canon and Nikon are released.

There are many different ways of manufacturing a True Color Sensor Photo-site . Sigma is only one of them , but Sony and Canon have been working on this for far longer than Sigma and they have been making their 3 CCD cameras for a long time. There are many Patent applications for multilayered sensors architectures filed already, by Sony ,Fuji and Canon .

The photographic field is about to change for the better. In the mean time , if you can afford it , an SD1 converted to Leica-R or Zeiss ZF lenses can provide the ultimate Image Quality.




Luis
SIGMA CUM LAUDE
Please click here to visit our new Forum
--
http://photo.net/photos/Luis-A-Guevara
http://www.pbase.com/luis_a_guevara/galleries
http://gallery.leica-users.org/v/LUIS+A+GUEVARA/
http://www.summiluxart.com/
http://www.sigmacumlaude.com/

 

[JOHNBLUE3]

Luis,

I apologize for my serious arithmetic error. My calculation is two dimensional; actually the same as yours. I mistakenly divided by the 1.5 crop factor. What I attempted to do was to calculate the number pixel along the horizontal axis, thus:

Total pixels = Nx*Ny, but Nx=1.5Ny, so it should have been,

Nx^2 = 1.5*15.3MP => Nx= 4800 appr.

but all of this is an aside from my main point. That is that if you look at lens specs from mfgrs. they typically spec sharpness wide open at 30 line pair/mm as a function of radial displacement etc. And since all lenses lose sharpness at large distances off axis, the smaller APS-C sensor is less demanding in this regard. It is the difficulty of manufacturing a lens that's sharp over the larger sensors, that drives up the price substantially.
As to measuring lens resolution, it's easier said than done in the field. I'm sure you are familiar with the Imatest Program, for example, which has been devised to do such measurements. The problem is that you always get a kind of convolution of things when you try to analyze the response to various step function targets. So, as an example, if both the lens and sensor had equal resolution on axis, say, The convolved measured resolution would be down by a factor of 1.4 or so. The only way to accurately measure a lens' resolution is to do some sort of scanning slit measurement. Let me end by asking the question; Why do you think Sigma states that it has over 40 lenses suitable for working into the SD1?