Sony vs Sigma blue and gray resolution measurements

Started 5 months ago | Discussions
docmaas
Veteran MemberPosts: 4,789
Like?
Sony vs Sigma blue and gray resolution measurements
5 months ago

For methodology please read these posts first for background on what kind of testing I am doing: http://www.dpreview.com/forums/thread/3669097#forum-post-53665179 and http://www.dpreview.com/forums/post/53653391

Sony Sigma civil war the Blue and the Gray

Well, here's blue from the Koren chart provided by Erik.

conditions: aperture f5.6 sigma 18-35 f1.8 at 18mm.

Camera further away from the target than yesterday.

SPP used to develop Sigma with sharp -7. Sony IDC used to develop sony with all neutral settings. Iridium Developer used to develop both with all neutral settings.

Note while you might be able to assess relative mtf it still has nothing to do with any other mtf measurement.

First the blue:

Sony A7r response to the blue chart. f5.6 18mm sigma 18-35 f1,8 iso 100

Sigma SD1 response to the blue chart. f5.6 18mm sigma 18-35 f1,8 iso 100

The sony response to the blue is surprisingly fat. It makes me wonder if the IDC is boosting the blue. Let's take a look at the same chart via iridient developer:

Sony A7r response to the blue chart. f5.6 18mm sigma 18-35 f1,8 iso 100

Indeed, Iridient shows a much more neutral response. However note that Iridient seems to find more detail in the sony image than sony does. Interesting.

I wonder how iridient compares to spp for the blue:

Sigma SD1 blue response as developed by Iridium -- nearly identical to spp

Note that while spp sharp is set to -7 no sharpening at all on Iridium. Looks like -7 is very close to 0 sharpening in actuality.

Now the gray:

Sony a7r aps-c mode response to the gray chart. f5.6 18mm sigma 18-35 f1,8 iso 100

Sigma SD1 response to the gray chart. f5.6 18mm sigma 18-35 f1,8 iso 100

Now the crops of the charts:

Sony

Notes:

In both cases the Sigma outperforms the sony.

The sony response to both the blue and the gray is surprisingly much more plump from sony Image Data Converter than that from Iridient indicating that IDC may be boosting response during processing.

Sigma spp results match those from iridient almost perfectly. In particular the sharpness setting of -7 seems to match the no sharp setting on Iridient.

Mike

Sigma SD1 Sony Alpha 7R
If you believe there are incorrect tags, please send us this post using our feedback form.
Aku Ankka
Contributing MemberPosts: 591
Like?
Per pixel? Per image?
In reply to docmaas, 5 months ago

In a hurry now, so not really reading about your methodology. Are you measuring "per pixel"? By this I mean SD1 vs. 15Mp crop of A7r? If so, I don't see what is the point as the results don't tell anything about how much either camera resolved details.

Anyhow, if I have the time, I hope to repeat your test on my camera and to post the result. Terribly busy now though and my printer is broken so can't print the test target at the moment.

Cheers, Aku

-- hide signature --

Quack!

Reply   Reply with quote   Complain
rjjr
Veteran MemberPosts: 6,438
Like?
How do the images look at ISO 400, 800, 1600...?
In reply to docmaas, 5 months ago
No text.
Reply   Reply with quote   Complain
xpatUSA
Senior MemberPosts: 3,591Gear list
Like?
Re: Sony vs Sigma blue and gray resolution measurements
In reply to docmaas, 5 months ago

docmaas wrote:

First the blue:

Sony A7r response to the blue chart. f5.6 18mm sigma 18-35 f1,8 iso 100

Sigma SD1 response to the blue chart. f5.6 18mm sigma 18-35 f1,8 iso 100

The sony response to the blue is surprisingly fat. It makes me wonder if the IDC is boosting the blue.

Thanks for taking the time to publish the results of what is obviously a lot of work. In the Sony above the rise of amplitude with frequency does mean that IDC is messing with the image data. That would put me off if I were a Sony candidate. The Sigma response is much 'better' from my POV.

Now the crops of the charts:

Sony

Compared to slant-edge testing, the results gained from this method do seem to be easier to assess and more visually representative.

For camera performance analysis, I try to exclude as much converter processing as possible and have lately been using output from RawDigger and such. It's surprising how soft and noisy 'raw' images can be! Less surprising is how much 'pre-processing' goes into crap like Adobe Camera Raw which is guaranteed to destroy any X3F fed into it 

-- hide signature --

Cheers,
Ted

 xpatUSA's gear list:xpatUSA's gear list
Sigma SD9 Sigma SD10 Sigma SD14 +2 more
Reply   Reply with quote   Complain
mike earussi
Veteran MemberPosts: 4,857
Like?
Re: Sony vs Sigma blue and gray resolution measurements
In reply to xpatUSA, 5 months ago

I assume by your use of the Sigma 18-35 on both cameras you're using only The APS-C portion of the A7r.

But if you have the time I'd be interested in seeing the color response of the A7r FF compared to the Sigma.

Thanks again for your effort.

Reply   Reply with quote   Complain
Erik Magnuson
Forum ProPosts: 12,081Gear list
Like?
Gradient assumptions
In reply to xpatUSA, 5 months ago

xpatUSA wrote:

Thanks for taking the time to publish the results of what is obviously a lot of work. In the Sony above the rise of amplitude with frequency does mean that IDC is messing with the image data.

While it's hard to be sure from these crops, the Sony looks almost black in the center of the thinner blue lines.  This may have to do with the demosiacking algorithm IDC uses making assumptions how the image is changing over the "gaps" in the blue pixels for certain gradients.  Then again, it could also be due to JPEG chroma subsampling.

That would put me off if I were a Sony candidate.

I doubt you would use ever Sony IDC at default settings.

-- hide signature --

Erik

 Erik Magnuson's gear list:Erik Magnuson's gear list
Canon EOS 5D Mark II Canon EOS 450D Sigma SD10 Sony Alpha NEX-5 Nikon D3200 +28 more
Reply   Reply with quote   Complain
docmaas
Veteran MemberPosts: 4,789
Like?
Re: How do the images look at ISO 400, 800, 1600...?
In reply to rjjr, 5 months ago

Good question.  I'll test it today.

Mike

Reply   Reply with quote   Complain
docmaas
Veteran MemberPosts: 4,789
Like?
Re: Per pixel? Per image?
In reply to Aku Ankka, 5 months ago

FF or aps-c on A7r is irrelevant because the pixels don't change size. The only thing that changes are the 2d dimensions. The center section and the target will be exactly the same size regardless of the setting on the camera.

It's an indicator of how well similar size pixels (4.9 micron sigma and 5 micron sony) resolve.  It's particularly interesting because the a7r has no aa filter so we can see what the absence of the aa filter does for the bayer image.  Of course a Nikon D800 (not D800e) in the mix would make things even more intersesting but incompatible mounts make that very difficult.

The whole point is see how much difference there is in resolution.  You do need to read the links I posted to get informed as to the theory behind the experiments.

Mike

Aku Ankka wrote:

In a hurry now, so not really reading about your methodology. Are you measuring "per pixel"? By this I mean SD1 vs. 15Mp crop of A7r? If so, I don't see what is the point as the results don't tell anything about how much either camera resolved details.

Anyhow, if I have the time, I hope to repeat your test on my camera and to post the result. Terribly busy now though and my printer is broken so can't print the test target at the moment.

Cheers, Aku

-- hide signature --

Quack!

Reply   Reply with quote   Complain
Roland Karlsson
Forum ProPosts: 20,942Gear list
Like?
Re: Per pixel? Per image?
In reply to docmaas, 5 months ago

docmaas wrote:

FF or aps-c on A7r is irrelevant because the pixels don't change size. The only thing that changes are the 2d dimensions. The center section and the target will be exactly the same size regardless of the setting on the camera.

It's an indicator of how well similar size pixels (4.9 micron sigma and 5 micron sony) resolve. It's particularly interesting because the a7r has no aa filter so we can see what the absence of the aa filter does for the bayer image. Of course a Nikon D800 (not D800e) in the mix would make things even more intersesting but incompatible mounts make that very difficult.

The whole point is see how much difference there is in resolution. You do need to read the links I posted to get informed as to the theory behind the experiments.

And ... we already know that, given the same amount of MP, the Foveon will resolve more. And we already know that an AA less camera, with a sharp enough lens, will have color artefacts when doing aliasing.

Nice to get it quantified though.

 Roland Karlsson's gear list:Roland Karlsson's gear list
Sigma DP3 Merrill Sigma DP2 Quattro Sony Cyber-shot DSC-RX100 III Pentax K-3 Pentax smc DA 15mm F4 ED AL Limited +8 more
Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads