Pages: [1]   Go Down

Author Topic: Fuji XPro1 samples  (Read 3021 times)

32BT

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3097
    • Pictures
Fuji XPro1 samples
« on: April 04, 2012, 09:19:34 am »

Let me first say that the very first digital camera I bought was a Fuji. One of those pocketable superCCD types. The one that brought so much controversy because of the 6Mpx designation, even though they only actually captured 3MPx. I liked the little critter, even though it was slow and cumbersome. But it had an excellent little lens for the package which combined well with the bracketing mode.

And yes, I also think that Fuji should be commended for its innovation attempts. And no doubt that the new X Pro1 will make a useful camera for a lot of people.

however…

(you could see that one coming from a mile away, no?)

however…

As I already mentioned in an earlier post here (shameless selfpromotion), the new sensor layout is not going to provide the benefits that are currently touted. And the recent samples provided by DPreview immediately showed the problems I have predicted.

That is to say, it is not immediately visible, but that is because they try to mitigate the problem by excessive color-noise-reduction, and this is exactly why I am posting this: 

Apparently the touted lack of color-aliasing is NOT a result of the sensor layout, but is a result of the excessive color-smearing. It is for example specifically visible in the tree images, which show the remains of the color-aliasing problems in the tree-branches (as vague yellow, green, and red blotches), but also show excessive smearing on the edge of the grassland and waterfront.

The amount of color-smearing is such that a normal bayer setup would easily produce an equal lack of color-aliasing.

I therefor would like to challenge(?) the reviewers and owners to look into this problem specifically. I might be totally wrong on this issue, but if it really is true then I feel this is important for at least these 2 reasons:

1) For landscape photography, if the problems mentioned really exist, they will appear in a significant number of images.

2) Obviously, if you tout a specific advantage due to sensor layout, it would be useful if a true comparison will actually reveal the advantage, and the advantage in one area should clearly not be detrimental in several other areas. (the color smearing).
 
Logged
Regards,
~ O ~
If you can stomach it: pictures

Alan Smallbone

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 779
    • APS Photography
Re: Fuji XPro1 samples
« Reply #1 on: April 05, 2012, 04:40:02 pm »

Oscar,

I am not sure I understand it all but I was curious of the images show by DPreview, which ones in particular showed the artifacts that you are talking about? I am not trying to argue just trying to see what you are talking about. I am interested in the camera and was thinking of renting one.

Thanks,
Alan
Logged
Alan Smallbone
Orange County, CA

32BT

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3097
    • Pictures
Re: Fuji XPro1 samples
« Reply #2 on: April 06, 2012, 06:50:28 am »

Image 9992 would be a good example. The one with the bald trees next to the lake. It is designated as a velvia shot which may or may not exacerbate the problem, but that's not the point obviously. (note that the examples below are additional jpg compressed screen grabs, download the original file for actual examination).

Example 1 indicates the problem areas with markers. Note that this is definitely NOT in pixel peeping territory. And I'm pretty sure this should appear in a normal sized print as well.

Example 2 shows how it also affects the waterfront. But in general you can easily see it all over the image. Note for example the unnatural appearance of the clouds. Both the clouds themselves, as well as the edges with blue sky. Or note how the tree trunks change color depending on the background due to color blurring.

It is not meant as Fuji bashing. Far from it. Thinking about this for the past days, I thought that perhaps this sensor setup would fare much better in architecture, as opposed to nature. The sensor isn't that much different from a normal bayer sensor, you can think of the sensor as having a rotated red and blue filter array. In order to rotate the R&B array in a fixed pixel grid, you have to widen the gap between the pixels, which allows more green pixels but reduces the number of R&B pixels. Now there is a lot of leeway in the amount of color information that you can ignore (read "blur"), but Fuji doesn't quite seem to have struck the right balance.

At this time I am unsure whether it is even possible to overcome these problems. I guess we will have to wait for third party converters to take a stab at it. (And yes, I would personally love to give it a try, just to get a better "feel" for the possible results).




Logged
Regards,
~ O ~
If you can stomach it: pictures

Alan Smallbone

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 779
    • APS Photography
Re: Fuji XPro1 samples
« Reply #3 on: April 06, 2012, 09:50:56 am »

Well that is interesting but I am going to see what I can see when I get the camera, I should have it on Tuesday, I am renting it for a short time, along with 2 lenses. I will write up my review and sample images. Be nice if there was another raw converter soon.

Alan
Logged
Alan Smallbone
Orange County, CA

32BT

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3097
    • Pictures
Re: Fuji XPro1 samples
« Reply #4 on: April 06, 2012, 10:28:02 am »

Well that is interesting but I am going to see what I can see when I get the camera, I should have it on Tuesday, I am renting it for a short time, along with 2 lenses. I will write up my review and sample images. Be nice if there was another raw converter soon.

Nice. Make sure to do some B&W as well.

Here's another test for you. My architecture theory just went down the drain visiting Steve Huff Photo (.com). At the bottom of his review there is an image with orange/red tiled roofs. Then suddenly the green/yellowish diagonal is staring me right in the face. I presume the 100% inset was a correct representation, and this is not some resampling problem. It's visible in other roofs as well.

(clip for "journalistic purposes" reproduced without permission).





Logged
Regards,
~ O ~
If you can stomach it: pictures
Pages: [1]   Go Up