Sigma DP-1 and resolution/print size

The specs I see for the Sigma Foveon chip in the DP-1 reads "2652 x 1768 x 3 layers." So what does this mean in

terms of print resolution? If I print at 300ppi, does this mean my maximum print size will be just under 6" x 9"?

Yes, I know, you can upres, but I'd prefer not to.

Your assumption is right.

Sigma claims 14MP. It's all marketing speak. The camera's real spatial resolution is about 5MP. Say it's 6MP because it's unnecessary to demosiac a Bayer filter.

 

The real advantage of the DP1 is that it's the only P&S form factor digicam with a DSLR sized sensor. I'd say look somewhere else if this isn't important for you.

The DP1 uses the same senson that's in Sigma's SD14 and is aprox equivalent to a 9-10 mp Bayer sensor in observable resolution. It also uprezzes very well. Mant of my shots have been done with a SD14 and uprezzed using PS bicubic to 16" x 24" and look quite nice. Here are some of my SD14 samples so you can judge for yourself:

 

http://www.pbase.com/mikeearussi/sd14_iso_50

Thanks, Mike. Can't tell anything about resolution on the web, of course, but they look to be rich in colour and the right contrast. I'm really tired of the overamped colour and constrast in my old digicam. And am growing tired of scanning 35mm colour neg from my Leica. The DP-1 seems close to what I'd like but not quite there yet. I suppose what I want is an M8 for less than $1500!

<< ... If I print at 300ppi, does this mean... >>

 

Your call as to whether to print at 300 dpi, but some photoshop 'gurus' don't believe that's necessary. In fact, some contend that they cannot see a difference when they drop to 150:

 

http://www.scottkelby.com/blog/2007/archives/767

Viewing distance issues aside, it depends very much on the image content. Basically, not much resolution is needed if you're after Holga'esque prints. This is fine. All things equal though, it is preferable to get the most technically competent gear at a price point.

 

300dpi is magical because it's roughly near the resolving limit of unaided human vision (and not surprisingly that of printing equipment capabilities.) Anything above this is great but is of diminishing returns.

I can't get excited about the F4 max aperture. If you remember all those great old 35mm small cameras the slowest of the really tiny ones that was full frame like the Rollei 35 se was f2.8 the take everywhere ones had F2 or faster lenses. Now we are paying a heck of a lot more for slower lenses.

Theoretically a Bayer sensor is at worst 1/3 as good as a Foveon,

given that four RGGB pixels make the equivalent of one

with an oversampling of green. So a 6 Mp Foveon sensor

(actually a bit less in this case) should be equivalent to 18 Mp.

 

In real-life images green is quite important, so a Bayer sensor

is in practice 1/2 as good as a Foveon, making it the equivalent

of a 12 Mp sensor.

 

In other words, the Foveon really is quite good.

Harvey,

 

Why don't you take your trolling elsewhere?

Bill, I can certainly see you point as far as colour rendition goes, but I don't see the "1/2 as good" logic in terms of resolution, i.e. crispness of detail.

 

And Michael, I can show you 8x10 prints where the difference between 300 ppi and 360 ppi on an Epson 2400 actually shows a difference. Granted, it is like Robert says - it depends on the content - but I've seen specific examples of content where the difference between <300 ppi prints definitely show less refined detail than 300 ppi prints. (Scanned 6x7 film shot on a tripod, mind you)

 

If you apply that 360 ppi resolution to a DP-1 file, you only come up with a print size of 7.4" x 4.9"! It would be gorgeous, but small.

Well yes, the Epson is a 360/720/1440 ppi device.

If you were to print both on a Fuji Frontier,

the 300 dpi would probably look better because that's the native resolution.

<P>

<A

HREF="http://www.trustedreviews.com/digital-cameras/review/2007/02/02/Digital-Photography-Tutorial-Camera-Sensors/p2">

Here is a good intro with pictures</A> of the Bayer pattern sensor.

Bayer has high resolution for luminance, but chroma resolution is lower.

That's why JPEGs are often encoded with 2x1 chroma subsampling:

there isn't more color resolution than that anyway.

I'm not an expert on printing so I can't really address the pixels issue there.

 

I tested a Sigma SD14 against a Canon 5D using a bellows and some Xenotar lenses and found the Sigma to be the equal of the Canon at iso400 and below. Above iso400 the canon was superior. I also did resolution tests using Koren's charts and methodology (http://www.normankoren.com/Tutorials/MTF5.html) and found the system mtf for the 5D to be slightly better than the sd14 using the same lens and Koren's target.

 

Uprezzing the Sigma to Canon native size also showed that the Sigma image was the equal of the 5D at its native resolution. One of those posting on the dpreview forum when I did this did print the two images and could see no difference between them.

 

If you compare the raw file size between a Sigma shot and a 5D shot of the same subject they will be very close to the same size. In other words even though the 2D spatial measurement is 4.7 MP it still contains as much information as does the larger 12MP 2D spatially measured 5D.

 

Finally you may want to read Sean Reid's review of the Sigma. It does cost money but from what I hear it is well worth the cost. I haven't purchased or read it because I already have the dp-1 and am in the process of paring down my other equipment since the DP1 does almost everything I need and I expect the succeeding models to fill those gaps.

 

Mike

Many forget that the DP1 and SD14 don't have to have an anti alias filter that drops the effective resolution of a Bayer sensor because essentially it smears out the image over adjoining pixels. Comparative resolution will vary according to the nature of the test (maximally in favour of Sigma using a red and blue test chart), but I think 8-9MP Bayer equivalent is a fair general assessment.

Another data point is that at the PMA (Photo Marketing Association) show, Sigma was hanging 20x30" images from the

DP-1 that looked very good.

<p>

As others have noted bayer cameras have various things going against them for upsampling, which is why you have

learned not to make use of it much - the reality is that bayer images arrive to you already upsampled due to the masking

process and re-assembly of image data from separate spatial color components.

<p>

That is why you are able to get around a 10MP level of detail from the camera. But the whole story is even more

complex than that, due to the Foveon sensor capturing a constant level of detail across all colors in a scene and a bayer

camera being unable to:

<p>

<a href="http://www.ddisoftware.com/sd14-5d/">http://www.ddisoftware.com/sd14-5d/</a>

"... DP1 and SD14 don't have to have an anti alias filter ..."

 

And this is a design fault. Any digitally sampled system has to be fronted with a low pass filter. The

consequence of its absence is having to tolerate false scene components getting folded back into the digitized

image. The link Kendall posted above is instructive; do look at http://www.ddisoftware.com/sd14-5d/5d-ymc.gif

Oops... I linked only to the 5D gif file. Anyways, the rest of it is on the web page.

Robert Lee [Frequent poster] , Aug 06, 2008; 01:39 p.m.

 

>"... DP1 and SD14 don't have to have an anti alias filter ..."

 

"And this is a design fault. Any digitally sampled system has to be fronted with a low pass filter."

 

If that were true the aa filters would not be optional on the digital backs used by professionals.

 

Mike

The assumption is that the user knows better.

 

Implementing a physical AA filter with ideal characteristics is very difficult. Practically, enough suppression of unwanted high frequency components also necessarily throws away meaningful image detail.

 

Even then, the AA filter is generally a good thing. It's fairly easy to compensate for a shallow filter roll-off in post. Indeed, this is the net result of sharpening, whatever the approach, in Photoshop. However, removing aliased image components after the fact is impossible. The classic example is the moire that comes about when imaging woven fabric under some circumstances.

 

Removing the AA works under a some circumstances. One is if the source doesn't have the troublesome image components to begin with. Second is if the optics ahead of the imager already sufficiently limits the resolution, i.e., when shooting stopped way down (or if you're using a coke bottle for a lens.)

 

My recollection is that there were actually small format digital cameras without AA fairly early. I think one was a 14MP Kodak DCS-something (in the era of 4MP mainstream DSLR.) The argument there was precisely that the sensor was dense enough such that the limited lens resolution effectively low pass filtered. I don't know how well this actually worked in practice. My impression was that this was a gee-whiz marketing exercise - and many things old comes around again.

Robert,

 

Well all my sigma cameras do just fine without it. So does my 5D which has only the horizontal filter removed.

I occasionally see some moire in the raw on the 5D but not after processing and there is a noticeable improvement

in the accutance of the images produced.

 

The AA filter is necessary for some shots but not for all shots, for some subjects but not for all. The Leica,

Sigma, and Kodaks all get along fine without it most of the time. Sure there are exceptions but they are few and

far between. If you want a camera that smears details in all shots at the expense of only a few where it would

matter get one with an aa filter. If not be like the pros and get one where you can choose or be like those of

us who shoot without them and suffer the occasional situation where the lack has a negative impact on the image.

In 20,000 + shots it hasn't happened enough to me to worry about it.

 

If I were shooting fashion with lots of fine weave I would want an AA filter and those who do shoot fashion are

primarily those who use the AA filters on MF backs. Roots and rocks folks leave the AA off for the most part.

 

There is no simple answer. If someone shoots more people and events where there are lots of people the AA filter

is probably useful. If one shoots closeups of birds the AA filter might be useful. If one shoots primarily

landscapes, near or distant an AA filter is more of a hindrance. All IMHO of course.

 

Mike

<p>Agree with Mike. I have the DP-1, M8, and 5D, and have considered removing the AA filter from the Canon. I don't consider moire much of a problem at all with the Sigma and Leica.</p>

<p>In answer to the OP, I've printed a DP-1 RAW file to tiff straight out of the box at 180ppi to 12x18" and the print looks great, and I am pretty sensitive to sharpness issues. I think printing only at 6X9" might be somewhat akin to thinking only contact prints are acceptable from film negatives.</p>