Tuesday, January 10, 2012

Fuji X-Pro1 Sample images are available … some thoughts …

So the new X-Pro1 is out. A nice looking body with a new mount, the X mount. A cool new sensor that groups pixels 6 at a time instead of the Bayer 4 pixel grouping. And the groupings are random, which attempts to simulate film and thus head off moiré.

They say that allows this camera to have no AA filter, which should bring with it some fairly serious sharpness. Not quite Foveon sharpness though, since there is still the need for demosaicing the color for each pixel.

Anyway, the sample images are pretty, and Fuji broke a long tradition by including a high ISO image. Good on them.

A quick peek at the sample images shows that we are not really able to judge much yet. The opening shot with the 18mm f/2 prime is at 200 ISO (looks like this sensor does not do 100 ISO) at 1.9s and the sky is kind of grainy with what looks like banding. Shot at f/10 in A-Prio mode for some strange reason. Aperture as high as f/10 should not be needed on an expensive prime. Weird …


The image is pretty and the color is nice. The silhouetted elements at the bottom are quite blurred. Perhaps a cheap tripod was used … that would not surprise me.

What is curious is the fairly heavy banding of the gradient from the bright parts at the bottom left into the dark parts of the top right. My TN panel clearly shows banding. The IPS monitor is better, but zooming in on the data shows that there is a surprising amount of posterization of colors.



Nothing smooth about that. The other issue that worries me is the CA. For a prime that will cost almost as much as the X10, the lights are not pretty at all.


These could be preproduction images, but I suspect that the lenses are a done deal. Of course, at typical sizes, people won’t notice much. But the X10 has a real issue with edges of things like branches, with halos and CA mitigating sharpness. I hope that this camera won’t suffer similar issues.

Caveat: Fuji took the unusual step of using Velvia film mode, which makes the color stand out but also uses a stronger tone curve. This may have affected the outcome somewhat.

The 1600 ISO image is quite attractive, but the background is even more obviously grainy. Not sure why that is, but I’ve tested the X100 and it seems nice and clean up to 3200 and quite usable at 6400 …

This one was shot at a more reasonable f/2.4, which is wide open on the X mount 60mm macro lens. The subject part looks nice, as it was intentionally shot for limited depth of field. This one is also Provia and the shutter came out to a resonable 1/58s … which of course eliminates the need for a tripod for many people. (Critical sharpness is always better with a tripod, but 1/focal length tends to make for reasonable sharpness hand held.)


Looks great at web sizes. But again the background seems noisy for 1600 ISO on a modern sensor. And again I see some posterization that is disturbing.


When you get even closer (354% here) you can see that this is not fun noise to deal with. I like noise that is fine grained. This does not give me that impression. This looks more like nasty JPEG compression issues.


Note: This was shot at –2EV, which tells me that the JPEGGs continue to be metered very hot in the X series. I really don’t like that very much …

The file size of a 16Mp camera without compression would nominally be 48MB. The file size of this JPEG FINE image is 4.557MB, which is 10.5:1 compression. That’s the highest compression I have ever seen for a FINE image. The busiest image of the samples is 7.250MB, and is thus 6.6:1 compression. That’s still fairly high compression for JPEG FINE. Perhaps this is where these artifacts are coming from …

Let’s peek at one more. Since the lack of AA filter is supposed to be a huge plus where sharpness is concerned, let’s get close to the subject on an image that has the potential to show this off.


The image on the sample site is shown rotated 90 degrees counter clock wise, but frankly I much prefer the image in this orientation.

What is fascinating here is that the background is perfectly smooth.


Now that’s more like it. Shows how great this thing might be.

This was shot at 200 ISO and f/13 (macro shot needed more depth of field) with the PROVIA film mode. DR100 was used, and that is good too, since it allows lower ISO. Shutter was 1/8 so I presume tripod was used.


Straight from the cam, the subject looks pretty awesome. The clarity and cleanliness combine for a beautiful 3-dimensional rendering. The tones are perfectly rendered with no posterization in sight. This is how images should look on large sensors.

Fuji cannot be faulted for their honesty. You can see the issues that are present in the JPEGs at high ISO and in night shots with gradients. RAW is where this thing will be most interesting. Of course, with that kinky new matrix, I’m sure that we are 6 months away from ACR support, unless Fuji got smart this time and opened discussion with Adobe with preproduction hardware available …

Bottom Line

Some concerns about high JPEG compression. Looks like skies are posterizing in JPEG FINE, and that’s not acceptable. But an easy firmware adjustment, so no big deal.

The CA on the lights is small enough that it won’t matter much, but for night shooter of architecture who want to print big (and of course this sensor should support that at this price point) there is no way that JPEGs should show any CA. Again, something easily addressed in firmware and of course in RAW with a proper lens profile (again, Fuji should be addressing this with Adobe NOW.)

So those minor glitches should be easily removed …

The real star here is the absolutely liquid presentation of that last image. Gorgeous tones and smooth gradients make for an amazing dimensionality that you see in  the very best lenses and sensors. This thing could be a huge player if this is how the images will look most of the time.