Drinking Sigma's Kool-Aid.. DP-1M wow..

[finbarqs]

I've been obsessing over the Foveon sensor since the SD-1 was announced. I thought no way was I ready to start investing in a whole new system. I'll just stick with my canon setup, 5D Mark2, 24-70 f/2.8L, 17-40 f/4L, 70-200 f/2.8L IS II, 580ex2, and the new 600RT with a couple of pocket wizards.

However, doing several weddings and portraits, I've come to realize that every time I take my 5D Mark 2 out, it's just such a hassle, and I feel like I'm working rather than having fun. So I invested in a Fuji X100. I started taking that with me to concerts, and the low light ability blew me away... it looked as good as my MK2, with better Auto Focus! (Sorry my MK2 is ~ 50% at best -- hoping to make the move to a mk3 soon) Image quality on the Fuji X100 was also amazing, the Astia shots were amazing for portraits, and my velvia shots were stellar. This was truly a photographer's compact camera.

Then sigma announced the DP-2M and the DP-1M. Compact version of the SD1, but carries the exact same sensor (15MP APS-C). I thought about it, and decided to sell my X100 in favor of the DP-1M/DP-2M. Didn't know which one, and I realized that since I'm not really gonna be taking a lot of portraits with that camera, I'm gunning for the DP-1M. I realize that this was going to be a huge sacrifice: no flash, horrible write speeds - even on the fastest card, awful low light ability, laughable video, and probably the worst battery life to ever graze a digital camera (compact at that!) and no Photoshop/Lightroom support -- meaning Sigma Photo Pro for raw files only. Plus I would have to spend money to obtain one!

But still I had to have it. Sold my Fuji X100, and got the sigma. took it zion with my 5D Mark 2, slapped on a 24-70 f/2.8L, and decided to take some shots, to compare. Though this comparison is a bit flawed (i'll make more adequate comparisons) I was completely shocked. Obviously photogs know that Full-Frame doesn't give a camera it's "image quality" but the ability to view more in frame (compared to a crop sensor). But anyways, this was going to see what kind of kool-aid every sigma user's been drinking...

This is the original image from my 5D Mark 2 with the 24-70 f/2.8L
I used photoshop lightroom to fix exposure, and to color correct (balance). Did not do anything with CA.

This is the original image from my Sigma DP-1M. I imported the image into SPP and it came out like this, pre-balanced. I thought, awesome. I'm just going to export it.

Though they are 2 completely different images, albeit different focal lengths, the same aperture, the same iso, I figure this was flawed from the beginning. Sure the color of the sigma looked more natural, but that wasn't what blew me away.

This is the 100% crop of the 5D Mark 2 of the sharpest, most in focused point of that frame.

This is the 100% crop of the Sigma DP-1M of the sharpest, most in focused point of that frame.

Wow I was blown away. Though they are 2 completely different pictures, and the test was flawed from the beginning, the Canon can only get marginally better. It will not be Sigma DP-1M good.

Now I'm drinking Sigma's Kool-Aid also!!!

 

[runawayprisoner]

Immediately, I can see the "full frame" sensor delivering smoother bokeh and higher dynamic range. Those circles showing up in the DP-1M shot look quite... unappealing to me.

 

[finbarqs]

I fail to see the higher dynamic range. I do not have highlights nor shadows that the mark 2 shows superiority over the sigma. If anything the 5D 2 has more contrast than the sigma image (looking at the full image) showing that the sigma has the upper hand on DR.

I do see smoother bokeh, which means it's another fault of mine: shooting at different focal lengths. I did shoot the 5D @ 70mm vs. sigma's 19mm.

I do see the out of focus elements being circular (judging that they look like flares coming through, I'm guessing it's the outline of the aperture) But besides the bokeh, and the circles, the sigma ran circles around the 5D in sheer detail resolving, and the amount of noise, and the visible CA.

But i'll take something a little more fair, probably gonna pull out my 17-40L put both cameras to f/5.6 and do a direct test at the same focal length (19mm vs. 19mm)

 

[CuriousMike]

It looks like the Canon was shot at ISO 12800 and the Sigma was shot at ISO 100.

 

[runawayprisoner]

I fail to see the higher dynamic range. I do not have highlights nor shadows that the mark 2 shows superiority over the sigma. If anything the 5D 2 has more contrast than the sigma image (looking at the full image) showing that the sigma has the upper hand on DR.

From what little I understand, contrast is actually not related to dynamic range... contrary to what some people would like to believe. When I wrote "dynamic range", I really meant how the 5D2 captured an image that "looked" like it covered a wider range of colors, and allowed a more vivid (saturated) look to come out without compromising details. I didn't mean how different the darkest and whitest points were in those photos.

Of course lighting and other factors come into play as well, but judging from that point alone, I don't think the DP1M is holding any edge over 5D2.

The DP1M does have a cleaner look, though, if nothing else.

 

[gus6464]

Is there 2 versions of the DP1? I ask because you say yours has the same sensor as the SD15 but the Sigma website now states that the DP1 has the same sensor as the SD1 which is a completely different beast altogether.

http://www.sigmaphoto.com/shop/dp1-merrill-compact-digital-camera

Same Foveon X3 46megapixel Direct Image sensor as the SD1 DSLR camera

 

[finbarqs]

I have the DP1-M which is the DP1-Merrill, which has the same sensor as the SD-1. Dynamic Range of photography literally means the brightest to the darkest points. Dynamic range of color means the brightest green, to the darkest green.

obvious, those pics aren't going to win any awards. But it was just a thought of how much detail, and how much color accuracy the sigma can pull in straight from the RAW file. The 5D2 pic has been saturated, contrast corrected, and balanced corrected.

Anyone knows that secret of digital photography lies in the post processing. The photography is actually the easy part. But sharpness, detail resolving, and color accuracy, and probably actual dynamic range, sigma has it hands down. Even the chromatic abberations the sigma controls it very well! I've had my 5D2 for about 4 years (since it came out) and no way will I use my sigma for weddings. It's too slow. But with patience, it will pay off.

P.S. Foveon X3 46mp is actually 15MP. Because the foveon sensor has 3 layers (red, green and blue), you'll have all the color information on 1 pixel site. Contrary to the bayer pattern, it works in R,G,G,B, so basically 4 pixels will make a full color range. But through the use of technology advancements in bayer technology, the interpolation of the missing color information is not bad at all.

 

[elitejp]

Sigma really has a great idea behind them i really hope that they can continue to put some r&d into their tech. As for a non sensor tech savvy person it appears that this sensor design should be the best out of sony/nikon or Canon. I have no idea about hasslebad but not many people have the money for one of those either.

 

[Don Vito Corleone]

I imagine you've seen this, but here is a pretty entertaining little video showing the DP-1M and DP-2M Merrills in action in the field. https://www.youtube.com/watch?v=f3VjyHQiqdE

These are impressive little cameras in terms of image quality but I can see why they're not more popular - they seem wildly impractical for anything other than outdoor, high-light environments with static subjects, shot from a tripod. If that is the only kind of shooting you do, I would agree the Sigmas are a valid alternative to a DSLR as long as you can deal with the other shortcomings (particularly the horrible high-ISO performance, the battery life, and the need to edit RAW files only in Sigma's software).

 

[SecurityTheatre]

It looks like the Canon was shot at ISO 12800 and the Sigma was shot at ISO 100.

Interesting comparison!

 

[finbarqs]

yeah isn't that crazy? so much noise involved it's ridiculous... They both were shot @ iso 160. Of course, those are zoomed in 100%

 

[finbarqs]

I imagine you've seen this, but here is a pretty entertaining little video showing the DP-1M and DP-2M Merrills in action in the field. https://www.youtube.com/watch?v=f3VjyHQiqdE

These are impressive little cameras in terms of image quality but I can see why they're not more popular - they seem wildly impractical for anything other than outdoor, high-light environments with static subjects, shot from a tripod. If that is the only kind of shooting you do, I would agree the Sigmas are a valid alternative to a DSLR as long as you can deal with the other shortcomings (particularly the horrible high-ISO performance, the battery life, and the need to edit RAW files only in Sigma's software).

I haven't seen that yet, but it's nice how they compared it to a D800. But yeah, the raw files out of the camera were incredible. The color, the detail, everything was like perfect right off.

 

[Throckmorton]

To be fair, the Sigma is lower res because it's really 5 megapixel, with 3x as much real data per pixel. So of course a 100% crop looks better.

 

[Aharami]

OP, can you please do an apples to apples comparison - ie, shoot the same subject with the same light? I'm genuinely interested in what you find

 

[finbarqs]

will do!

And sigma really is 15MP. I never considered it to be 45MP, but then again, Bayer patterns are considering themselves to be 20-30MP. So who knows, it's a marketing game out there.

Of course, prints are where it counts in doing our type of photography, otherwise, we'll stick with our instagram. And sigma does not disappoint, in printing large prints!

 

[Throckmorton]

will do!

And sigma really is 15MP. I never considered it to be 45MP, but then again, Bayer patterns are considering themselves to be 20-30MP. So who knows, it's a marketing game out there.

Of course, prints are where it counts in doing our type of photography, otherwise, we'll stick with our instagram. And sigma does not disappoint, in printing large prints!

Oh wow I didn't realize the resolution was that high.


Why the hell don't Canon and Nikon use their sensors? We need to ditch the Bayer grid.

 

[AkumaX]

will do!

And sigma really is 15MP. I never considered it to be 45MP, but then again, Bayer patterns are considering themselves to be 20-30MP. So who knows, it's a marketing game out there.

Of course, prints are where it counts in doing our type of photography, otherwise, we'll stick with our instagram. And sigma does not disappoint, in printing large prints!

Can you add in an X-Pro filter, a line bokeh effect, frazzy frame border, and resize it to 640x640?
Kthnx :wub:

--

P.S. how will you do apples to apples? 5D2 + 85mm @ f/2.8 vs DP-1M + 50mm @ f/2.0?

 

[Don Vito Corleone]

Oh wow I didn't realize the resolution was that high.


Why the hell don't Canon and Nikon use their sensors? We need to ditch the Bayer grid.

I can only assume (but would be happy to be corrected by someone with more knowledge of the Foveon sensor than I have) that the very slow write times and poor low-light performance of the Merrill cameras is something intrinsic to the Foveon sensor, and the big guys consider it a non-starter for that reason (not to mention the giant file sizes). On the other hand, the DP-3M, which was just announced, supposedly has a 7 FPS burst rate, so it may be that the slow write time on the 1 and 2 is due to lack of RAM and/or a clunky image processor.

 

[finbarqs]

line bokeh effect- you mean tilt-shift? They do on the SD-1, but it's the actual lens

I will do an apples to apples test (or my best) I will stick the Canon 5D2 + 17-40 @ 19mm @ f/4 compared to the Sigma @ 19mm @ f/4. And I'll just crop the canon to get the same FOV. DOF should be EXACTLY the same. I'll just do raw conversions to JPEG, and post it here.

I do believe in the future, they can get better with noise, and battery life (with the technology) Throw on FF, and I'm sold. If they made a FF SD-1 with canon mount (sorry I have canon glass right now) I'd buy it in a heart beat.

 

[finbarqs]

Test shot to try and equalize both as much as possible. Shot in terrible lighting, bad subjects, but we're here after the detail!

sigma @ F/4

canon 5D Mark 2 17-40L @ F/4


100% crops from both

canon

sigma

Both were shot on the tripod.

 

[finbarqs]

here are the f/8 pictures.

canon

sigma

here is the 100% crop

canon

sigma

Immediately, the color is different on both cameras, you can definitely see that canon is warmer than sigma's cooler approach.

But this test was shot on a tripod, at the same focal lengths, and cropped to make them similar in FOV.

 

[Charles Kozierok]

Heh, Sigma.

I remember having discussions about them when the Foveon sensor first came out, and arguments about whether it was better or not, and whether their marketing claims were accurate or not, etc.

At that time, Sigma was notorious for having serious color accuracy issues, likely due to a combination of the Foveon sensor and their own processing engine. The requirement to use their own software was also a major complaint that still seems to be around.

I'm not going to rain on your parade, but do keep in mind that most of the "uber sharpness" associated with Sigma images is due not to their sensor, but their deliberate design decision to omit the antialiasing filter that is commonly used on most other cameras (though not all). This does increase sharpness, but also the potential for aliasing artifacts, which can range from non-existent to severe depending on the subject.

 

[GoSharks]

But this test was shot on a tripod, at the same focal lengths, and cropped to make them similar in FOV.

I don't understand why you are comparing 100% views when your FOVs are different. Frame the comparison shots such that the FOVs are identical and then look at the 100% for detail and sharpness. No cropping.

Or am I just reading you wrong?

 

[finbarqs]

Heh, Sigma.

I remember having discussions about them when the Foveon sensor first came out, and arguments about whether it was better or not, and whether their marketing claims were accurate or not, etc.

At that time, Sigma was notorious for having serious color accuracy issues, likely due to a combination of the Foveon sensor and their own processing engine. The requirement to use their own software was also a major complaint that still seems to be around.

I'm not going to rain on your parade, but do keep in mind that most of the "uber sharpness" associated with Sigma images is due not to their sensor, but their deliberate design decision to omit the antialiasing filter that is commonly used on most other cameras (though not all). This does increase sharpness, but also the potential for aliasing artifacts, which can range from non-existent to severe depending on the subject.

not talking about the uber sharpness, but mainly for the detail that it can pull. It doesn't need AA filter because of the sensor technology. It doesn't show moire (which is the aliasing artifacts you're talking about)

The reason why I cropped in for the FOV so I can get the EXACT DOF at the same focal length. Besides, that's what a cropped sensor is, cropped! But when you zoom in 100% is still the same when you zoom in 100%

This time, I think it's a much more fair comparison because we both were @ 19mm, at the same aperture, same ISO, etc. I actually kind of prefer Canon's "warm" side on this one.

Remember, these were straight RAW files converted into JPEG. Canon used lightroom 4, and sigma used Sigma Photo Pro 5