How is this SD1m portrait photo shot ?

adrianashoots wrote:

I found on this page some very nice photos of Sigma SD1

https://m.eprice.com.tw/tech/talk/1187/17202/1/

https://timgm.eprice.com.tw/tw/dc/img/2011-05/23/17202/oscar210_5_Sigma-SD1_2415914c6d0ac5f92b2ee209f74dcde8ori.jpeg

How it is this Merrill photo shot ?

The skin looks smooth, and colors are pure. Not leathery, dry in any way, but healthy.

A very interesting portrait. Personally I find the flash highlights a bit rough and the face a bit overexposed.

But I also think Merrill raw files are very malleable.

The photo is a bit front focussed but that may be intentional.

Generally speaking, not referring to this particular image, I think excessive sharpness can be avoided or coped with during post processing. I think Foveon sensors can deliver exceptional image quality also for portraits.

I think this is excellent

adrianashoots wrote:

I think this is excellent

I agree.

Scottelly wrote:

xpatUSA wrote:

adrianashoots wrote:

I found on this page some very nice photos of Sigma SD1

https://timgm.eprice.com.tw/tw/dc/img/2011-05/23/17202/oscar210_5_Sigma-SD1_2415914c6d0ac5f92b2ee209f74dcde8ori.jpeg

How it is this Merrill photo shot ?

The skin looks smooth, and colors are pure. Not leathery, dry in any way, but healthy.

Most of the skin is out of focus which is why it looks smooth with not much detail:

GIMP edge detection filter - black equals no edge equals out of focus,

Ted, I'm sorry to disagree with you, but it sure looks to me like most of the skin would be in focus. Just look at the hair that is showing as in focus in your image, and on the left side that hair is obviously behind her cheek, while the hair on the right side is obviously in front of her face, so in front of her cheek. It looks to me like the focus is perfect on her eye, making most of her cheek, nose, chin, lips, and some of her hair in good focus, but for some reason your image is showing mostly just some of her hair, lips, and eye in focus, when we know that most of her nose has to be in focus, because all but the tip of her nose is behind the hair on the right, and in front of the hair on the left. Most, if not all of her cheek falls into the same plane of focus, yet your image makes it look like neither her nose nor her cheek are in focus.

For the above, I applied Gaussian edge detection, then Levels, then Threshold conversion to pure B&W. From Scott's comments the threshold  settings may be too high.

Here are some Edge Detections - without threshold conversion:

In many modern cameras, focus "Peaking" uses a similar method of edge contrast, as does CDAF.

The top left is too much ... the entire cheek and all the hair appears to be in focus!

Of the other three, I lean toward 'Gradient' at top right, followed by the other two.

Knowledgeable comments about edge detection always welcome.

Here are some photos I took with Foveon cameras

Quattro

https://www.instagram.com/p/BmOHZdoD3RK/ (slightly weird skin tones but I love the way the dp3 Quattro renders her hrands)

Merrill (the first two are unedited)

https://www.instagram.com/p/Bk_0VOKDSq6/ (you can see the color bleeding issue where the camera pulled the gold from the skin into the relatively unsaturated shadows of the teeth)

https://www.instagram.com/p/Bk_yOqtDron/

https://www.instagram.com/p/BPgz--gAW5O/

https://www.instagram.com/p/BO1HybAA49p/

https://www.instagram.com/p/wMhQ6rRfWg/

High resolution versions of a couple of the above photos here:

https://medium.com/ice-cream-geometry/an-ode-to-the-sigma-merrill-506dd0864169

If I want to smooth Foveon skin a bit, I export at 2x resolution and apply some skin smoothing plugin in Photoshop - the extra resolution helps the algorithm and skin looks great. But mostly I'm happy with Foveon skin - skin tones and sharpness - even if somewhat over sharpened by Sigma's algorithms.

After that, I think people should see this too (your blog):

http://sodium.nyc/blog/

…

High resolution versions of a couple of the above photos here:

https://medium.com/ice-cream-geometry/an-ode-to-the-sigma-merrill-506dd0864169

If I want to smooth Foveon skin a bit, I export at 2x resolution and apply some skin smoothing plugin in Photoshop - the extra resolution helps the algorithm and skin looks great. But mostly I'm happy with Foveon skin - skin tones and sharpness - even if somewhat over sharpened by Sigma's algorithms.

The portrait taken with the Merrill sensor in the linked article (“An ode to …”) has just an amazing, wonderful picture quality. It is pictures like that, that make me think there is something very special to this kind of sensor.

Obviously the person behind the camera is at least as important. Also the lenses are important etc. But still, IMHO those sensors can produce images that give a very realistic impression and have are particular intensity.

Scottelly wrote:

After that, I think people should see this too (your blog):

http://sodium.nyc/blog/

and many thanks for your portrait skin tone JPG comparison:

http://sodium.nyc/blog/2020/05/camera-jpg-portrait

I couldn't agree more that the combination of the EXR Pro processor and X Trans I sensor as used in the Fujifilm X-Pro 1 and X-E1 cameras (= row C in your test) yields very very nice skin tones straight out of camera.

My own experience with Foveon (esp. Merrill sensor) skin tone that it is hard work to get it right - it really needs precise exposure, precise white balance, careful post-production etc. I don't use the Merrill now for family portraits any more as I always get the comment "why do I look so old?"

In contrast, with the Fuji X-E1 one can be a happy snapshooter (esp. around family) and the skin tone comes out very very nice in JPG without much if any effort. And nobody ever complained to me about the Fuji images making them look old..

so my thumbs up to the 10 year old X-E1, which can be had for little nowadays (I would look for the kit with the very nice 18-55 zoom).

paulsch wrote:

…

High resolution versions of a couple of the above photos here:

https://medium.com/ice-cream-geometry/an-ode-to-the-sigma-merrill-506dd0864169

If I want to smooth Foveon skin a bit, I export at 2x resolution and apply some skin smoothing plugin in Photoshop - the extra resolution helps the algorithm and skin looks great. But mostly I'm happy with Foveon skin - skin tones and sharpness - even if somewhat over sharpened by Sigma's algorithms.

The portrait taken with the Merrill sensor in the linked article (“An ode to …”) has just an amazing, wonderful picture quality. It is pictures like that, that make me think there is something very special to this kind of sensor.

Obviously the person behind the camera is at least as important. Also the lenses are important etc. But still, IMHO those sensors can produce images that give a very realistic impression and have are particular intensity.

Yeah, they're special. Let's hope Sigma hits a home run with the FFF too.

allineedislight wrote:

Scottelly wrote:

After that, I think people should see this too (your blog):

http://sodium.nyc/blog/

and many thanks for your portrait skin tone JPG comparison:

http://sodium.nyc/blog/2020/05/camera-jpg-portrait

Thanks everyone for the kind words. I've been lucky enough to try multiple camera systems and test them for my own knowledge - I'm happy other people find value in what I share.

I couldn't agree more that the combination of the EXR Pro processor and X Trans I sensor as used in the Fujifilm X-Pro 1 and X-E1 cameras (= row C in your test) yields very very nice skin tones straight out of camera.

I don't know what it is about the X-Pro1 but it's got the best skin tones of the bunch. While I may reach for a Nikon for natural light, if I had to have one camera for the rest of my life it would be the X-Pro1.

My own experience with Foveon (esp. Merrill sensor) skin tone that it is hard work to get it right - it really needs precise exposure, precise white balance, careful post-production etc. I don't use the Merrill now for family portraits any more as I always get the comment "why do I look so old?"

I'm well used to manual exposure settings and studio lighting - I would only ever use a Foveon sensor in a high key setup where there were almost no shadows and everything was controlled.

But when I do... the result is worth it.

In contrast, with the Fuji X-E1 one can be a happy snapshooter (esp. around family) and the skin tone comes out very very nice in JPG without much if any effort. And nobody ever complained to me about the Fuji images making them look old..

so my thumbs up to the 10 year old X-E1, which can be had for little nowadays (I would look for the kit with the very nice 18-55 zoom).

👍

Scottelly wrote:

After that, I think people should see this too (your blog):

http://sodium.nyc/blog/

Some nice pictures, but he has some crazy ideas about the evolution of colour vision.

Most mammals have blue<-->yellow colour vision. The B/Y opponent signal goes from the retina to the brain. This signal did not evolve last, as the author guesses.

Higher primates such as humans are also evolving red<-->green opponent colour vision. This is "work in progress" and individuals vary greatly.

The problems with reproducing skin and hair colours are mostly failures of metamerism, caused by the use of a crude system of three highly saturated primaries in photography. It is in theory possible to mix any of the subtle colours found in skin from RGB or CMYK, but the mixtures are very unstable and the slightest unbalance is very obvious.

If you want to see good reproduction of skin colours, look at original paintings (not reproductions) by Rembrandt, Velasquez, Van Dyke, etc. They used iron oxide pigments such as Raw Sienna, Venetian Red, or Raw Umber, and they built up the colour in layers so that the white pigment gives a Tyndall effect as in real skin. They would not use the same pigments for brightly coloured clothes.

Haemoglobin is likewise an iron based pigment and has a similar spectrum.

Personally, I think portrait photos are best in B&W. But you can reduce the bad effects of colour photography by reducing the vibrance -- perhaps in the skin areas only if the person is wearing brightly coloured clothes.

Skin colours from Foveon sensors may not be as bad as from some other cameras. The Color Checker does have two squares that use iron oxide pigments to simulate skin colours, so making a profile with the Color Checker can help.

Don