Fujifilm X10 'Orbs' Investigated. Does the Firmware Fix Work?

Studio Scene: EXR 'DR' Mode

Something that we have noticed after extended use of the X10 is that the appearance of the white discs/orbs differs depending on the 'DR' dynamic range setting. In full-resolution 12MP mode the reason for this is obvious if you know how the camera works - increasing the DR setting from 100% to 200% and up to 400% increases the base ISO sensitivity from 100 to 200, to 400, and as we've already seen, increasing ISO sensitivity reduces the intensity of the white discs effect.

But what about EXR mode? In 'DR' EXR mode the X10's 'DR' dynamic range setting can be extended up to 1600% using a combination of exposure and tone curve adjustment, and EXR technology. These images were shot in identical conditions to the samples on the previous page, but ISO was set to automatic (400) and because exposure cannot be manually adjusted, exposure compensation was used to match the brightness. 

EXR 'DR' mode, DR 100%, ISO 400 (FW 1.02)	EXR 'DR' mode, DR 100% ISO 400 (FW 1.03)
EXR 'DR' mode, DR 200% ISO 400 (FW 1.02)	EXR 'DR' mode, DR 200% ISO 400 (FW 1.03)
EXR 'DR' mode, DR 400% ISO 400 (FW 1.02)	EXR 'DR' mode, DR 400% ISO 400 (FW 1.03)
EXR 'DR' mode, DR 800% ISO 200 (FW 1.02)	EXR 'DR' mode, DR 800% ISO 200 (FW 1.03)
EXR 'DR' mode, DR 1600% ISO 400 (FW 1.02)	EXR 'DR' mode, DR 1600% ISO 400 (FW 1.03)

As you can see, the white discs decrease in size as you go up the DR scale, but they retain their artificial-looking, hard-edged appearance. Again though, there is no noticeable improvement between firmware versions 1.02 and 1.03. We also tried shooting side-by-side comparisons in EXR 'SN' mode, which is designed to deliver lower noise at 6MP, but with identical results - no improvement with firmware 1.03.

Sample Variation

One curious thing that we've noticed while working through this issue is that not all X10's display the dreaded 'white orbs' to the same extent. We've used two cameras from different production runs, and while both produce images that show the effect, one displays more severe 'orbing' than the other. We've used images from the better of the two cameras in the rest of this article, so you can judge for yourself the severity of the issue, but for the sake of completeness here's an example of the difference between the two models that we've used, at base ISO and full-resolution (12MP). 

Camera A	Camera B
Camera A	Camera B

We would hope that Camera B, in the table above, (from an earlier production run) is the exception, and that more of the X10s on the market behave like Camera A in our example. Even if this is true though, the key point is that the 'white discs' effect is visible in images from both cameras that we have used, and in neither case does updating to firmware 1.03 make a difference. We have asked Fujifilm whether or not changes have been made to the X10's manufacture since production began and we will update this article with a response as soon as possible. 

Summary

We're still working through our testing, but for now it seems clear that firmware version 1.03 does not appear to have a significant impact on the appearance of 'orbs' and certainly isn't the silver-bullet solution that a lot of users had hoped for.

We have been unable to observe a meaningful improvement in either studio or real world testing which strongly reinforces our suspicion that the problem is hardware-related, and cannot be solved by an adjustment to the camera's firmware. The issue is caused by uncontrolled sensor blooming, where signal spills out from photodiodes radially, onto their neighbors, which creates an artificial-looking white disc around blown-out point highlights. Blooming isn't unique to the X10, but this effect - the distinctive 'orbs' - is unlike anything we've seen before from a modern camera.

After drawing a blank in all of our image quality testing we had thought that perhaps the new firmware adjusted the camera's program line in Automatic ISO mode. As we've seen, the appearance of the white discs is less severe at higher ISO sensitivity settings, so perhaps the new firmware biases automatic ISO to use these high settings more frequently? We have seen reports that this is the case in EXR Auto mode and we will continue to work through the issue, but so far we have not been able to consistently demonstrate any clear bias towards higher ISOs with the new firmware. 

So after all this, what have we learned? Here are our findings, summarized.

• 'Orbs' are real, but some cameras may be more prone than others due (presumably) to hardware variation.
• The appearance of the orbs changes as you increase ISO sensitivity, and they become softer.
• By increasing the DR '%' settings you can reduce the size of the orbs, but they remain unattractive unless you increase ISO sensitivity.  
• In itself, firmware version 1.03 appears to have no noticeable intrinsic effect on the appearance or intensity of the orbs.
• However, some reports suggest that auto ISO performance in EXR Auto mode is tweaked in FW 1.03 to select higher ISO settings (where orbing is less noticeable). We cannot consistently demonstrate any high ISO bias in FW 1.03 compared to 1.02 (but we'll keep on looking). 

UPDATE March 5th 2012: Following the publication of this article Fujifilm has promised us a 'definitive solution' to the so-called 'white orbs' problem, which will be announced on the 12th March. We will report full details when we have them.