High ISO sample images from cameras old and new

[Four Thirds User editorial team]

It's an issue that has bugged Four Thirds users since day one - high ISO camera performance. Is it really that bad? And how do the latest cameras fare? You can make your own mind up!

More...

[catkins]

Always interesting to get comparative shots like these.
Out of interest to compare how the E-30 that I have compares at a higher ISO with the E-5 which seems to get good reviews, I've downloaded two .orfs -

Image 6: E-5 ISO: 3200
http://fourthirds-user.com/sample_im...w/P5100634.ORF
and Image 10: e-30 ISO: 3200
http://fourthirds-user.com/sample_im...w/P5104430.ORF

Comparing the two side by side in ACDSee Pro 4, I'm surprised by how good the E-30 is when the whole image is seen. But the surprise to me is when zooming in to show more detail - although the E-5 seems to have less obvious noise this seems entirely due to far greater noise reduction with presumably in-camera editing that has lead to a noticeable softness in the detail of the overall E-5 image.
However, in places such as the text on the herb jar this blur is 'shadowed' and I wonder whether this may be due to movement of the camera or the IS system?
Are you able to judge what the blur is due to - over zealous noise reduction or movement?

Regards

[fluffy]

That's very interesting. I compared the 1600ISO RAW and jpg of both the E-PL2 and the E-5 (I have one and wouldn't find the other). And what I took away from the comparison is that in most cases (no DR problems) I might as well stick with best quality jpgs on the E-PL2 and not bother with the RAW.

I also did the GH-2 and E-PL2 comparisons also at 1600 which is past what I ever use and the GH-2 wins every way over the E-PL2 but I love the E-PL2 colors.

This was a very, very helpful exercise.

[dingenus]

Well, use an other, better, converter like Lightroom or Viewer2. The E5 has not less obvious noise trough more noise reduction but trough lower sharpening while using a weaker AA filter and an other NF algorithm. Looking at that ORF it seems not made very carefully. AF not on the car but on the front of that TV HD box. The text on the herb jar: this blur is 'shadowed' you say. I can't see it in V2. In V2 you can also use the slider "false color suppression". If you set it on max, the yellow noise will nearly completely fade away.
Olympus Viewer 2 is a free download. It shows all the camera settings with the ability to change them.

[Ian]

Originally Posted by dingenus

Well, use an other, better, converter like Lightroom or Viewer2. The E5 has not less obvious noise trough more noise reduction but trough lower sharpening while using a weaker AA filter and an other NF algorithm. Looking at that ORF it seems not made very carefully. AF not on the car but on the front of that TV HD box. The text on the herb jar: this blur is 'shadowed' you say. I can't see it in V2. In V2 you can also use the slider "false color suppression". If you set it on max, the yellow noise will nearly completely fade away.
Olympus Viewer 2 is a free download. It shows all the camera settings with the ability to change them.

All the focusing was manual and the intention was to get the lettering on the herb bottle in focus. The most difficult to do this with (no live view) was the E-1

Ian

[Ian]

Originally Posted by fluffy

That's very interesting. I compared the 1600ISO RAW and jpg of both the E-PL2 and the E-5 (I have one and wouldn't find the other). And what I took away from the comparison is that in most cases (no DR problems) I might as well stick with best quality jpgs on the E-PL2 and not bother with the RAW.

I also did the GH-2 and E-PL2 comparisons also at 1600 which is past what I ever use and the GH-2 wins every way over the E-PL2 but I love the E-PL2 colors.

This was a very, very helpful exercise.

I haven't yet fully examined all the results, but I did like the Olympus JPEGs more than the GH2's, but that wasn't too surprising.

Ian

[Ian]

Originally Posted by catkins

Always interesting to get comparative shots like these.
Out of interest to compare how the E-30 that I have compares at a higher ISO with the E-5 which seems to get good reviews, I've downloaded two .orfs -

Image 6: E-5 ISO: 3200
http://fourthirds-user.com/sample_im...w/P5100634.ORF
and Image 10: e-30 ISO: 3200
http://fourthirds-user.com/sample_im...w/P5104430.ORF

Comparing the two side by side in ACDSee Pro 4, I'm surprised by how good the E-30 is when the whole image is seen. But the surprise to me is when zooming in to show more detail - although the E-5 seems to have less obvious noise this seems entirely due to far greater noise reduction with presumably in-camera editing that has lead to a noticeable softness in the detail of the overall E-5 image.
However, in places such as the text on the herb jar this blur is 'shadowed' and I wonder whether this may be due to movement of the camera or the IS system?
Are you able to judge what the blur is due to - over zealous noise reduction or movement?

Regards

If you are viewing any RAW file, all you are seeing is the default interpretation of the RAW data by that RAW viewer application. View the same RAW file in 5 different and unrelated RAW viewers and you will have five different looking views

So what you are seeing is the ACDSee view of things. If you look at the JPEGs, then that's all the work of the camera and is more valid from a comparison point of view.

But the best test is what you can do with the RAW file in your chosen program.

Ian

[dingenus]

Ian, I have no doubts about your positive intensions. They are very good. The problem with this things is that the modern NF does different things (I think) with detail in focus and detail out of focus. Maybe in the rawpart too. In the image is not really fine detail as clothing fabrics or hair exept the fluff on the hood of the car. ( I only looked at 3200iso) But thats not in focus. The focussing on the letters is different between EPL2 and E5, just a few mm but with this close focussing the sharpness impression is just different. Therefore I think that the relation between Noise Filter and detail needed other finer detailed images wit parts in focus, and parts slipping to "out of dof".

[Ian]

I'm sorry if there was a slight inaccuracy with the focus, but let's face it, this was never intended to be a lab test. Criticality is not something to expect with high ISO photography. At f/7.1 and at a reasonable viewing distance the focus discrepancy described will not be significant. But I do hope these comparisons will still be useful for evaluating noise grain, banding, etc.

Ian

[dingenus]

I understand what you mean. Sorry if I put the bar too high. Have you already overhauled the sharpness test E5-GH2 ?