Fuji EXR sensors, the X10, and dcraw
Posting rules: It shouldn't need saying, but... play nice. Please keep your discussions civil. You can disagree, just don't be disagreeable. And, of course, all of the usual stuff like no spamming. Tex adds: I'll be rigorously enforcing this as we go along. We're probably going to be a small community in a little lifeboat, so we can't have members at each others' throats. This is for the sake of the project as a whole. So when you post, pretend you're speaking in person with your very wealthy auntie who has always treated you wonderfully and currently lists you prominently in her will. I won't be tossing anyone out of the forums because we are all in this together, but I'll delete suspect posts right away.
As I expected, dcraw 9.12 is out. It was released on Wednesday. As I expected, it contains support for the Fuji X10.
What I didn't expect was that LightZone didn't like it. The X10 raw conversions in LightZone were Teh Suck!!1! They were unacceptably noisy, even at minimum ISO. Furthermore, the DR200 and DR400 files were dark. I've spent three days learning a lot about EXR Raw files in general, X10 Raw files, and the older Fuji SuperCCD Raw files.
First the easy part: EXR Raw files (X10, X100, or otherwise) that are created with DR200 will appear 1 stop underexposed, and those created with DR400 will appear 2 stops underexposed. This is by design: the extended highlights have to go somewhere. You could always grab the Exposure slider in the Raw tool and push it up to 1.00 or 2.00 as appropriate. Then you'd get the expected exposure with the extended highlights all smashed to white. But that usually would be senseless, because you end up with half as many pixels and no advantage. What you need to be using is ZoneMapper — and remember that two segments in the ZoneMapper is one stop. What I still don't know is: do you want to be using this ZoneMapper in Luminosity mode or in RGB mode? The wrong mode could cause color shifts. If you've got an EXR camera and use DR200 or DR400, let me know what you find.
By the way, the above implies that the EXR Raw data isn't so raw after all. The camera blends the two exposures into a single set of Raw data. It's been reported that, unlike the X100, the X10 also applies vignetting correction to the Raw data. I can say from looking at the sample files that chromatic aberration correction is not being applied to X10 Raw files.
The "noise" in the X10 images turned out to be that LightZone thought that the X10 had a SuperCCD instead of an EXR sensor. The reason it thought that is that in the DR200 and DR400 modes, the X10 produces two sets of Raw data into the RAF file, just like a SuperCCD camera does. After three days of looking into it, I still don't know why. The two images are very similar. I'm satisfied, though, that using the first image — which is the easier path from a programming standpoint — works fine.
The new dcraw update will be out as soon as we can get it compiled. I think you'll understand that some people have higher priorities on the evening of December 24th than compiling dcraw. However, if you're running Windows, click here for the new dcraw.exe. There's also an interim Raw Tone Curve template for the X10. When I get a good reference image (preferably the DPReview studio scene), I'll produce a more accurate one.
Cameras added to dcraw 9.12:
- Canon PowerShot S100
- Fujifilm X10
- Leica V-LUX 3
- Nikon 1 (J1)
- Nikon 1 (V1)
- Panasonic DMC-FZ47
- Panasonic DMC-FZ48
- Panasonic DMC-GX1
- Samsung NX5
- Samsung NX200
- Sony NEX-7
If you need a Raw Tone Curve for one of these cameras, either put in an issue report at github or make a request in the Raw Tone Curve forum here.
- Log in or register to post comments
Comments
I've been doing some more research on the difference between the two raw data sets in the X10's RAF file. The differences are only along the edges, and are particularly concentrated in the "resolution" target in this image.
My analysis suggests that it's just a question of which direction the low-ISO data is merged with the standard-ISO data: left or right. Or is it up or down? With Fuji's 45° sensel matrix, it's all very confusing. I'm not completely convinced, though, because the JPEGs for the alternate image are a bit smaller than the JPEGs for the standard image, indicating less detail, less noise, or something.
Anyway, here are some screen shots of LightZone working with one of the "Dave Box" photos from Imaging Resource. Click any of the smaller images below to get the whole thing in uncompressed JPEG — about 1 megabyte each.
If anyone sees a reason to prefer the "alternate" data set, let me know. Otherwise I'm going to leave dcraw set to use the "standard" data set.
Fit to screen
Standard image
Alternate image
Difference
1:1
Standard image
Alternate image
Difference
Both DPReview and DCResource are reporting that the latest (1.03) firmware for the Fuji X10 does not correct, nor even reduce, the occasional problem of specular highlights being turned into large white "orbs." DCResource notes that the problem is not limited to EXR mode, and affects Raw files as well as JPEG.
Fuji says it does: "Blooming (White disk) phenomena, which can be observed in EXR AUTO mode, is reduced."
Some commenters on the DPReview article confirm that they're not finding any difference, either.
DPReview has just issued its report on X10 firmware 1.03 relative to the orb issue, and found no significant effect.
Some comments claim that the only difference is in Auto-EXR mode, where a higher auto-ISO is used. Higher ISO reduces the "orb" effect. DPReview, however, has so far been unable to verify that there is a difference in auto-ISO selection with 1.03.
In the wake of the "orb" brouhaha, Fujifilm has announced:
It's been noted that Fuji hasn't said anything about whether they will or won't make the new sensor available for retrofit. Their only advice for current owners is, "We encourage any customer with an X10 and X-S1 who has experienced the ‘white disc’ phenomenon to call their local authorized Fujifilm service centre." Exactly what your local authorized Fujifilm service centre will do about it, nobody seems to know.
One DPReview commenter writes, "I just spoke with Steve in the Fujifilm USA Service Center. I was informed that I would have to cover shipping expenses to send it in to have the sensor replaced." Another writes, "I just spoke to the UK Service Centre - really nice helpful people. They register your details and will apparently send you a free-post box when the new sensors are ready. The sensor will be upgraded and the camera returned, all free of charge." The UK version has been confirmed by other UK-based commenters. It's early, but no word yet as to how long the turn-around is expected to be.
It also appears that, for now at least, Fuji will continue selling the X10 and X-S1 with the old sensor.
DPReview's Barney Britton writes, "We've asked for a modified X10 and hope to receive on within the next few weeks." Obviously, question #1 will be, does it really fix the orbs? And question #2 will be, was anything else compromised?
As an aside, I totally don't understand this statement in the announcement: "This ‘blooming’ effect can occur with all digital cameras which use CMOS sensors, to varying degrees." I can't say that I've ever seen blooming on a CMOS sensor. CCD sensors were known for occasional blooming, but not CMOS.
It's nice to see RAF format is making its way into dcraw/LightZone(Zombie)... I've been thinking of getting one of these Fuji EXR sensor cameras for a while now, but have held off because they used to require processing in Fuji's software.