SD1 vs SDQH Detail Comparison
So after talking a bit with @tirpitz in the other thread about film scanning. I decided to do a quick test of the detail possible out of each camera. I was curious if the different arrangement and design of the Foveon sensors affected how much detail is available in macro photography or film scanning. Both were shot at f/8 at 100 (base) ISO. I used -0.3 exp comp on the SD1 and reduced the SDQH -0.4 in post because I forgot to set it before hand. I want to also note that the compos ion and lighting were not identical between setups. But here are my results, the SD1 is on the left and the SDQH on the right:
SDQH25c by Cory Maben, on Flickr
SD125c by Cory Maben, on Flickr
comp2 by Cory Maben, on Flickr
comp1 by Cory Maben, on Flickr
comp4 by Cory Maben, on Flickr
comp3 by Cory Maben, on Flickr
To my eye, they are (surprisingly) nearly identical. I could not tell any major differences in terms of detail. The Quattro has more contrast and at various points on the image one is sharper than the other, which I put down to simply different focus points and the slightly different compositions.
It is interesting that the level of detail is so similar to me given that the resolutions are so different. The SD1 has 3, 15mp layers, while the SDQH has 1 20mp blue layer, then two 5mp green and red layers. So I suppose if I had an object that was red. That would really highlight the differences between the two cameras. Maybe I will try to go find something at a thrift store to test and attempt to bring out the differences. But as it stands with this object under these conditions they look basically identical to me.
Thank you for the comparison, they indeed look quite similar, maybe with the slightest edge to the SDQH (which for the sake of precision has a 25.5MP blue layer, so on paper has almost twice the “real” resolution of the SD1).
I’ve never compared the two at macro level honestly, I recall that the SDQH’s increased resolution was way more apparent in very fine details at infinity. At macro level seems that the Merrill can hold its own even at 15MP if “real” resolution.
Quote from Lorenzo Rossi on May 22, 2024, 2:47 pmThank you for the comparison, they indeed look quite similar, maybe with the slightest edge to the SDQH (which for the sake of precision has a 25.5MP blue layer, so on paper has almost twice the “real” resolution of the SD1).
I’ve never compared the two at macro level honestly, I recall that the SDQH’s increased resolution was way more apparent in very fine details at infinity. At macro level seems that the Merrill can hold its own even at 15MP if “real” resolution.
Maybe that will be the next test. I always think about doing something like that but then bringing both my cameras and a tripod and setting up a shot feels like too much work and I never do it lol. Or maybe you could set up it, I'd be interested to see that. I still want to do the macro shots of something red. I didn't get out today due to not feeling well, but maybe sometime next week I'll set it up.
It's very difficult for me to figure out what the pixel count of the SDQH is. Depending on where you look you get different numbers. I was looking at this diagram on DP Review
But you're right, looking at the actual output files they are ~25.5, so I'm not sure why this image has them at 20mp. It's very hard to tell how many pixels are at each layer because I think sometimes people report only the color pixels and sometimes they include the phase detect pixels. But then DPReview also list that the effective pixels is 45mp, which 25+5+5 = 35, so I have no idea where that 45 number is coming from. Sigma takes it a step further and claims it's a 51mp equivalent camera. But if you look in Sigma's official spec sheet it lists:
the top layer (blue) as being: 6200 x 4152 (25.74mp), which aligns with the output from my camera.
the middle layer (green): 3100 x 2076 (6.44mp)
the bottom layer (red): 3100 x 2076 (6.44mp)
With a listed total (approximate) megapixels as 44.7mp. which must include the phase detect pixels because 25.74 + 6.44 + 6.44 = 38.62.
https://www.sigma-global.com/en/support/download/SIGMA_sd_Q_H_Catalog_en.pdf
So it's a complete mess and every number starts to sound totally arbitrary. I wish they were more clear about what the numbers are.
Unfortunately I recently sold my SDQH, so the only comparison I can try to do would be between my DP0Q and the SD-1 with the 14-24 F2.8 Art, but in the meantime I can try to find some of my old comparisons that I should have somewhere on my hard drive 🙂
Quote from Lorenzo Rossi on May 23, 2024, 6:33 amUnfortunately I recently sold my SDQH, so the only comparison I can try to do would be between my DP0Q and the SD-1 with the 14-24 F2.8 Art, but in the meantime I can try to find some of my old comparisons that I should have somewhere on my hard drive
That would be awesome if you had anything left in your files. Just out of curiosity, what was your reasoning behind keeping the SD1 over the SDQH?
So here is my test results for the macro colors. Same setup as normal, f/8, 100 ISO, etc.. This time though I used Daylight white balance to make sure everything was consistent between shots and between cameras. The SD1 wanted to overexpose so it took a bit of work to get everything to come out correctly. The SD1 seems to struggle with guessing long exposures correctly. For the comparisons the SD1 is on the left and the SDQH is on the right and as always just ask if you want the raw files:
SD1-1 by Cory Maben, on Flickr
SDQH-1 by Cory Maben, on Flickr
COMPO by Cory Maben, on Flickr
COMPY by Cory Maben, on Flickr
COMPR by Cory Maben, on Flickr
COMPG by Cory Maben, on Flickr
To my eye the SD1 was the clear winner with green, which is not a surprise considering it's larger MP green layer and it was also better with orange.
Yellow looked basically identical to me and surprisingly so did red, which I did not expect given the difference in greens.
I do see what you mean now with the SDQH having more vivid colors SOOC. But my conclusion is that it is not a night and day difference in terms of detail unless you have a lot of green/orange in your image. Which may suggest that the SD1 is better for landscapes. But like you mentioned before it's not clear if their macro color performance is going to be parallel to their non-macro subject performance.
Quote from Cory Maben on May 23, 2024, 8:00 amQuote from Lorenzo Rossi on May 23, 2024, 6:33 amUnfortunately I recently sold my SDQH, so the only comparison I can try to do would be between my DP0Q and the SD-1 with the 14-24 F2.8 Art, but in the meantime I can try to find some of my old comparisons that I should have somewhere on my hard drive
That would be awesome if you had anything left in your files. Just out of curiosity, what was your reasoning behind keeping the SD1 over the SDQH?
Actually it was somehow related to my recent purchase of the Mamiya ZD and the investment in lenses for that one that I decided to do.
First of all I needed to monetize something and the SDQH was a best sell than the SD-1 and second to me the ZD is able to produce basically the same results in terms of fine details (or even a tad better in terms of sheer acutance) as the SDQH but has its dynamic range much more towards the highlights (which I prefer) plus the "medium format look", so I decided to sell the SDQH and keep only the excellent DP0Q for the "Quattro look" and the SD-1 for the "Merrill look" and the ZD as the main stalwart 🙂 (I'm totally in love with that camera at the moment)
Here are some landscape-like comparisons I did some time ago between the SDQH and the SD-1, all files processed and sharpened at my best (also trying to make look the more similar possible). Some of them are not exactly at the same focal length unfortunately.
The actually look very close to my eyes, maybe with just a tad more "perceived" sharpness of the Merrill one, but the Quattro H's increased resolution allows to discern a lot of fine details in the distance that cannot be resolved by the Merrill due to less available resolution.
Based on my tests, I think that a Merrill sensor with the same real resolution of a Quattro sensor will likely trounce it in terms of perceived acutance though.
Thanks for sharing those. First of all they are extremely beautiful images! Where were these taken?
My immediate thoughts are that the SDQH has a stronger blue cast to the atmospheric haze than the SD1. But in pixel peeping around it feels like a toss up. The SD1 seems to do better at retaining detail in the trees and foliage, while the SDQH seemed to do better in retaining the geometric details in buildings and rocks and such. But you would never notice without pixel peeping, at viewing size they both look basically identical. One thing I did notice, especially in the photos of the manor(?), was that the SDQH has a digital-looking sharpness to it that makes me prefer the image out of the SD1, which I think looks more natural in that regard.
But to your point about resolution; I think that we have both been in agreement that the images are extremely similar, which is disappointing when I remember that the SDQH has a 70% increase in spacial resolution just to break even. Which speaks to what you were saying about given the same resolution the SD1 sensor design of 3 identically sized layers would be extremely impressive at 25 mp or beyond. But I feel certain that the full frame foveon camera will also have an asymmetric layer size structure like the SDQH, especially given all the reported problems they've had with getting a functional design.
I find what you mentioned about the ZD producing images on par with the SDQH really interesting. Several people over time on here have had that camera and I've always liked the results people got from it.
Thank you very much, they were taken in Bolzano, in South Tyrol in northern Italy, I can confirm it’s a really nice place 😊
Yes, SDQH need more “digital” adjustments like sharpening and “crispness” in SPP to look similar to the Merrill files, so guess that’s what you are noticing. If you don’t do that, usually you incur in the “mushy” look that a lot of people complain about in Quattro files. I tend to prefer the “crunchy” look I’m used from the Merrills, so save for portraits I tend to process all Quattro files like that.
Yes, the ZD was for me almost a shocking surprise, considering it has “only” 22MP, but they are REALLY good pixels, at low ISO with plenty of light of course. I’ll try to post something my on the future.
