My High ISO Evaluation

[Steaphany]

Even though I'll be posting a number of photos to this thread, I'm posting this here, under "Sigma Cameras" since the focus will be on how the SD14 performs through out the available ISO range.

Being inspired by akv's thread

Please, Log in or Register to view URLs content!

where the image was photographed at ISO 1600, I decided to experiment with my SD14 and see exactly what can be expected.

Earlier this morning, as several of my Horses were playing front yard gardener, I went out and shot a series of photos keeping as much possible the same while only varying the ISO through out the 50 to 1600 range. The sky was blue with few clouds, I had my SD14 set to Shutter Speed Priority, Sunlight White Balance, 1/320s Shutter Speed. These photos are pretty much snap shots with the only composition being an attempt to ensure a consistent angle between the Sun and my subject. I specifically chose not to perform post processing in preparing these photos for inclusion just so everyone can see what the SD14 base line performance is. Here are the photos:

ISO 50



ISO 100



ISO 200



ISO 400



ISO 800



ISO 1600, Since my 28mm lens could only go down to f22, I raised the shutter speed to 1/640s



So far so good, no significant noise, and the only obvious difference is a apparent reduction in color saturation.

 

[Steaphany]

These photos actually surprised me, so I pulled out my X-Rite ColorChecker card, shot another set and proceeded to code a spread sheet to perform some actual comparative calculations. I chose a shutter speed of 1/500s which allowed it to remain constant across the full ISO range.

Using the documented sRGB Red, Green, and Blue values as a base line and computing how the "as photographed" ColorCheck RGB color values differed from the ideal, I have come to some interesting conclusions.

Through out the ISO range, the SD14 delivers a constantly higher color saturation, while the mean luminance was within an acceptable range to the X-Rite ColorChecker expected values. These average errors were computed by comparing the SD14 performance to the X-Rite standard and averaging the results of all 24 color samples. Please note that my spread sheet calculated a wide range of results across the 24 ColorChecker references:

Please, Log in or Register to view codes content!

Interestingly, the individual color channels also show a consistent bias towards boosting the Red while also suppressing the Blue:

Please, Log in or Register to view codes content!

This effect has been discussed at length in the thread

Please, Log in or Register to view URLs content!

When comparing the error of individual Primary color squares, squares which are strong in the Red, the Red and Yellow squares, have a consistently have a boosted Red with a reduced Blue while strong Blue, Blue and Cyan, results in the opposite effect regardless of the ISO.

The image quality was acceptable regardless of ISO and surprisingly, there was no significant color noise even in the dark regions of the ColorChecker card.

Here are the photos:

ISO 50



ISO 100



ISO 200



ISO 400



ISO 800



ISO 1600



Seems like we may be able to find situations where the high ISOs may be useful.

 

[akv]

Good work Steaphany, I'm glad that my picture inspired! For good I hope! lol

I mostly use the high ISOs for black and white images in low light, like the salt shaker image.
In the horse pictures, the 100-200 ones were most pleasing to my eye. The 50 ISO had the shadows too dark, and in the 800 image I could really notice a magenta tinge to it. The 1600 was surprisingly pretty clear though.
In the color chart pics, the 1600 ISO looks a little "washed out", the colors look more pastel with muted hues on my laptop. The yellow of the case and the similar color on the chart (first column on the right, second row) seemed to match up best on 400ISO on my laptop. I doubt they're the exact same color though.
Thanks for the comparison!

Oh I read on another forum that the 100 ISO setting is actually more like 64. Not sure what their source was or how they came to that conclusion.

 

[tc95]

Steaphany,

I would have to concur...it does not do bad...it just does not produce the colors like the lower levels...

Great study...

Tony C.

 

[Guest .]

Hi All,

just do have a closer look

Please, Log in or Register to view URLs content!

Foveons HIGH ISO results are not that bad as one might think. Unfortunately SIGMA's "In-Cam-Picture-Processing" does not do a lot to make the best from their sensor's output. We can do that ourselves! I did some research in the issue!

See you with nice HIGH ISOs

Klaus

 

[Steaphany]

Thanks Klaus, I'll post my High ISO Photoshop Fix there