Go Back   Steve's Digicams Forums > Digital SLR and Interchangeable Lens Cameras > Sigma dSLR

Reply
 
Thread Tools Search this Thread
Old Nov 7, 2002, 12:50 AM   #1
Administrator
 
steve's Avatar
 
Join Date: Jun 2002
Posts: 2,535
Default Sigma SD9 First Look Posted



I just posted our Sigma SD9 First Look review.

It's not done yet but there is a starter set of sample pics and more will be coming shortly.

-Steve
steve is offline   Reply With Quote
Sponsored Links
Old Nov 7, 2002, 9:45 PM   #2
Senior Member
 
Join Date: Jun 2002
Posts: 259
Default

Hi,

At the time I ordered my D60, I was really taken by the hype about the Sigma/Foveon combo, particularly by their sample pix. So, I looked at your real-world pix using the SD-9 today. I may be off, but these were my observations:

1) Either a blown pixel or dust particle visible near the top left of the sky pix.

2) Slight blue cast.

The rest of these are relatively minor, and the overall detail is excellent.

3) Less sharp and less contrasty than the unprocessed D60 pix.

4) A hint of boosted edge contrast on the lower left image, where the treetops on the horizon meet the sky.

5) Some noise in the blue sky?

Does anybody else see these the way I do?
WalterK is offline   Reply With Quote
Old Nov 8, 2002, 3:39 AM   #3
Guest
 
Posts: n/a
Default

Is this the off the shelf product or pre production
  Reply With Quote
Old Nov 8, 2002, 1:02 PM   #4
Administrator
 
steve's Avatar
 
Join Date: Jun 2002
Posts: 2,535
Default

The SD9 we have came directly from Sigma Japan, it is in a retail box. It has production firmware.

And yes, it does have some dust on the sensor, the most notable is the large "dot" in the upper left which is very visible in blue sky pics. There are also other smaller specks that become more visible with smaller apertures. I didn't want to remove the dust protector and do any imager cleaning for fear of doing damage to it. These cameras are extremely rare at the moment and I'd rather put up with a little dust than no camera...

There is noise in the blue sky areas and some chromatic aberation problems but overall the daylight color is good and the sharpness and far field detail is awesome for a 3.4 mpixel sensor.

-Steve
steve is offline   Reply With Quote
Old Nov 11, 2002, 9:22 PM   #5
Senior Member
 
wsandman1's Avatar
 
Join Date: Jun 2002
Posts: 318
Default

Steve,

To simplify things, does the theory behind the FOVEON technology mean that the 3.4 mega-pixel image can be enlarged without losing as much detail as a super CCD image capture device? In other words the image could be enlarged to the equivalent of an image produced with a 6 or 9 mp camera and still have the same detail?

I am seriously thinking of getting this camera because I can get my hands on a SA-9 and several Sigma lenses, very cheap. But I'm a bit leary because of the small file 3.4 mp image size and the large multiplier ratio of (1.7).

Thanks,

Bill
wsandman1 is offline   Reply With Quote
Old Nov 12, 2002, 8:50 PM   #6
Junior Member
 
Join Date: Nov 2002
Posts: 1
Default

Guys,

Granted, I haven't seen too many sample photos from prosumer digicams, but the high level of detail in these photos is striking to me. With any other digicam I've used, I'd have to resample to a lower res and run an unsharp filter to get photos that look that good.

Steve, how was the battery life on this cam, considering that the CMOS sensor is supposed to use less power than a CCD?

I do have one question about the image quality: what caused the "red dots" I see on the petals of the flowers? It almost looks like a photo in which the color depth was reduced.

Do you know of any other manufacturers that plan on using this image sensor?
Mike T. is offline   Reply With Quote
Old Nov 13, 2002, 1:25 AM   #7
Administrator
 
steve's Avatar
 
Join Date: Jun 2002
Posts: 2,535
Default

The image quality in daylight ISO 100 shots is pretty amazing given the 3.4 megapixel "real" resolution of the SD9 images. It is especially noticeable in the far field detail like trees in the background, the sharpness of buildings and etc.

I don't know if I agree with Foveon's claim of 3.4 megapixel times 3 equalling 10 megapixels but against any other camera with 3.x megapixels it is clearly superior -- including cameras like the Nikon D1 or Canon D30. And it is definitely superior to any 3.x megapixel consumer digicam I've ever used.

There is visible noise in low light shots or exposures beyond 1/2 second, this seems to be the achille's heel for this sensor. That and Sigma definitely has a final assembly QC problem with dust on the sensor as ALL the samples I have seen from my camera and others exhibit a lot of specks and spots in pictures shot at F11 or smaller apertures. You can of course remove the dust protector and clean the sensor but you shouldn't have to do this with a brand new camera. I hope Sigma starts using better clean room facilities in the future.

To answer all the other image quality questions - this is why we post the sample pics people. Download them and print them yourself. Ultimately image quality is in the eye of the viewer.

We'll try to answer all of these questions in our conclusion so just be patient and let us finish our review and take more pictures. The weather has not been great lately and I only have another week with the camera.

-Steve
steve is offline   Reply With Quote
Old Nov 17, 2002, 9:48 AM   #8
Moderator
 
Join Date: Jun 2002
Posts: 1,139
Default On X3 and resolution.... not equal to six megapixel bayer's

Resolution is a very tricky term - Foveon's early claims of 8-9 or more bayer sensor equivalency from the 3.4 megapixel sensor are highly exaggerated. Does the X3's optical resolution reach even the six megapixel bayer sensors level? No - not if we consider resolution in the classic sense.

The X3's (chip designation) big theoretical advantage is in its color fidelity. There are a few bugs there as evidenced in color shifts, but it does offer decided advantages to the bayer sensors in this respect. As to optical resolution, resolution charts indicate interesting aliasing patterns. Does is have more resolution in the classic sense than the D30 or S1? Yes - but it doesn't approach the 6 megapixel bayer sensors in this respect: For distant detail or fine detail where there are curves or circles (anything other than 180, 90 or 45 degrees) the X3's optical resolution equivalency is like any other 3.4 megapixel sensor. The "rope-like" aliasing is evident and even using the best interpolation algorithms (step bicubic, Genuine Fractals, Lanczos, etc.) will not render a smooth image without the aliasing on these areas being quite evident on serious enlargements.

What does this mean in a practical sense? Simply that at time of capture, if insufficient pixels are allocated for areas of fine detail to properly define boundaries, interpolation software will quickly reveal the flaw. For amenable subjects like head and shoulders portraits, or macro shots, the X3 (disregarding excessive noise) images can be enlarged like the D30 or S1 or D1H, etc., to poster sizes with excellent results. But if fine detail of curved or circular edge nature is present, the enlargement potential is essentially the same as with any 3.4 megapixel bayer sensor.

Below are a couple crops made from identical captures with a D60 and S9. The first is a uninterpolated (disregarding bayer interpolation) zoom from PhotoShop showing identical spots on a curved metal bar surrounding a large metal tank. The stair-step aliasing is clearly evident in the Foveon capture, and much less so in the Canon six megapixel capture. There is some strange color shift in the Foveon X3's capture which is evident in the full sized image as well. That's another issue entirely - and for another thread.

The second image is a crop of the same area with the Foveon's capture interpolated in PhotoShop to the same pixel count as the Canon. The "twisted rope" appearance is still quite evident in the S9's image (both are enlarged 300 % from the normalized sizes for demonstration).

What does this mean then to the photographer?? That for serious enlargement of detailed subjects, the S9 will not equal the six megapixel bayer interpolated sensor's smoothness, but acts more like a 4.5 megapixel bayer sensor in absolute optical resolution. Resolution chart samples from the S9 show this aliasing which might be interpreted as "false detail." It becomes apparent in some strange patterns seen when enlarging the S9's images, but only a problem when pushing the limits of enlargement potential.

Lin



S9 top image - D60 bottom image below.....


Lin Evans is offline   Reply With Quote
Old Nov 17, 2002, 10:40 AM   #9
Moderator
 
Join Date: Jun 2002
Posts: 1,139
Default Another look at the Foveon aliasing problem from Steve's -

Here's another example of some of the strange aliasing in the X3's capture. Here are crops from the 1Ds and SD9 of areas of wood grain from the surf-rider stacks.

Look at the stair-stepping on the Foveon's capture. Not a problem until enlargements are made, then it becomes an issue. Same pattern is seen in resolution chart captures.....

Lin

Lin Evans is offline   Reply With Quote
Old Nov 28, 2002, 6:49 PM   #10
Moderator
 
Join Date: Jun 2002
Posts: 1,139
Default More on the SD9

Aside from the aliasing and other bugs common to a first iteration processor, the SD9's "sharpness" at ISO 100 is absolutely astounding. The single pixel detail is better than anything I've ever witnessed in a bayer sensor, including the 1Ds.

There are some "niggles" to contend with. Green colors are often muddy, browinsh green and there is more noise in the red channel that I would like to see. The sky blue is most generally an unnatural shade, and it's prone to highlight clipping and blown whites if even slightly over-exposed.

But damn! That shapness is hard to ignore. In fact it's so good that I bought one and a 28-70 ED lens to test it. It should arrive early next week. I don't want to invest in too many Sigma lenses until I determine if it will be useful for me in a professional venue, but after printing some samples I'm impressed enough with the edge detail and sharpness that I want to explore it further.

Hopefully Steve will get some decent weather and get some more samples for us to play with - and I'm anxious to try it on butterflies, if I can get one of my existing flash units to work with it.

Best regards,

Lin
Lin Evans is offline   Reply With Quote
 
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT -5. The time now is 4:44 AM.