This post on the FTF outlines a few problems that beginners have … and I’ll examine the issue in some detail.
Maybe I am just a lousy photographer. But the HS10 doesn't seem to do well. After taking a few pictures, I put it on my pc and bring it up at 100% and everything is smudged no detail. What am I doing wrong. I never had any camera do this to me before. The S6000 was beautiful from day one until now that it started having problems. Never a smudge problem, always clear details. So now what do I do? I put it in P mode, now we are having heavy rain, can't go out until it stops. Any tips would be appreciated.
thanks in advance.
Fuji’s S6000 has a somewhat large sensor by current standards … 1/1.7” with a pixel density of 14MP/cm sq. This was identical to the F11, a camera I have shot for some time now.
So here is basically what that poster was seeing at 100% with the S6000fd (from my F11 of course) – a building shot from a long way away in Boston, 100% crop shown in ACR5:
What happens when you start using really small sensors like the 1/2.33” sensor in the Panasonic ZS3 and load it up with 10MP (actually 12.7MP on the sensor, which is 45MP/cm sq – or triple the density of the F11) is that the edges start to get smudged, even in very bright sunlight. Noise picks up as well, even at base ISO.
But then again, you have more pixels so this should still be fine, no? Well, yes and no. Once you have shadows in the image, and / or the light begins to fade and ISO rises, this smudging takes off and become rampant with all subjects. They can literally turn to mush. Especially with weak jpeg engines, as many compacts have.
And 10MP over 6MP only leads to a fairly small difference in linear dimensions, as they only go up by the square root of the area. This is the fallacy of adding resolution. Once you are at 6mp, you can make superb 8x10 images (I have made superb 12x18 images from the F11) … so all the extra pixels are for marketing purposes.
Which means that you end up with people with helmet hair, shady spots in treed areas with *no* detail whatsoever, and buildings with terrible looking edges and smudges everywhere. Unfortunately, that kind of damage is visible even at small web sizes, never mind in prints.
Case in point is the comparison I performed between my compact cameras and the dSLR in the woods last week. The squirrel who sat quietly, unmoving while I changed camweras 3 times for a total of four examples:
The Panasonic looks horrid in some parts of the images:
This mush shows up at web sizes. Which is why I *never* shoot the Panasonic ZS3 for general purpose photography. My standards are simply far higher than it can deliver, despite having excellent side angle at 25mm and a sharp long zoom at 300mm.
These stats are actually shared by the HS10 … excellent wide angle, excellent long zoom (to 300 or 400mm … after that, users report the inevitable softening of pixels). Also a 1/2.3” sensor with 10.3mp on the sensor and density of 36MP/cm sq. This lower density looks promising for the HS10. Doesn’t it?
Well, yes and no. On the one hand, we know that the Canon G10 had a similar pixel density with its 1/1.7” sensor and 15MP on the sensor … and it made stunning images:
The G10 was easy to shoot in RAW. Images were captured and stored quickly, and the camera never felt sluggish. Basically, it was a high end compact camera that was able to shoot superior images in RAW at low ISO. Not so great at 800 ISO or above, requiring heavy processing. But it *did* maintain edge integrity for quite some time, ressisting the urge to go to mush too quickly.
The HS10 should have exactly the same potential, since it has similar pixel density. But the HS10 has horrid shot to shot times in RAW, so shooting RAW is not an option for 99% of people … no one has the patience to wait 5 seconds after *every* shot. And if you use burst mode, you get a few shots quickly and then wait *10* seconds.
So the HS10 is strictly a JPEG shooter. Oh oh … we know from experience that Fuji has an *extremely* aggressive jpeg engine, and often mistakes small low contrast details for noise, thus rendering mush.
And *that* is what this poor fellow is seeing. I’ve seen it in many samples and have been rather unkind to this camera. Because, as I have stated, I cannot stand mushy pixels. They show up all the time, even at small image sizes.
Here is a nice, bright image that was shot at low ISO by the HS10. It was used as a sample in the review on Photography Blog.
A crop to show you what I mean about mushy pixels:
That’s not half bad, actually. But we’re still missing the edges of a lot of the stones that make up these walls. They have superb low contrast details that are simply gone. And it does not really matter what you do, you won’t be getting those back.
Here is an F11 shot of London, specifically of Picadilly Square, at night and at 400 ISO. All disadvantages … and yet:
Edge details. Fuji had a superb JPEG engine in the early F series. They left a lot of details and had only the absolute minimum of noise reduction. Their software team was forced to ramp up the NR in order to handle the high pixel densities they are shipping today.
An example at 200 ISO … Trafalgar Square, the steps of the National Gallery. Swarming with people, as it does on a summer day. The detail available from the F11 will blow your mind:
Even in shadow, it retains excellent edges and some hair details. The HS10 can come close to that, but always seems to add a little mush to the mix. That’s the NR kicking in. As it must.
One last remarkable example of detail from a distance with the F11 …
The shirt is blown there, which happens, but the details in the brick are quite something.
Here’s another HS10 example from the review:
A snippet from the middle, just above center, shows the issue again:
Slightly blurred details, thick edges, and no edges between bricks. Not much can be done, as this is already base ISO. But … this is also a bit of an unfair comparison. After all, the F11 is only 6mp, so when making 8x10 or less, that’s all you need. To make this fair, let’s bring it into CS4 and reduce it to 6mp (essentially 3000px wide.)
Well, that’s actually a bit better. Not quite as crisp as the F11, but not too bad.
The bottom line is that newer models like the HS10 have fairly high pixel densities. That leads to edge issues and noise reduction smoothing of low contrast details. It is unavoidable. But pixel peeping does not help. Looking at pixels at 100% can tell you the difference between how two camewras render details as light drops … but it does not tell you how much difference it makes at web size or in small prints.
So don;t get too excited about soft edges … but *do* get excited about mushy pixels. Real mush, where the whole area gets softened and blended into a uniform mess, is destructive at all display sizes.
I would not shoot the ZS3 day to day because of this issue. And I would not shoot the HS10 day to day because of the same issue.
Let me leave you with some great examples from Fuji’s other new camera, the F80EXR.
At 12mp, high pixel density of 39MP/cm sq:
Decent edge detail, for sure. But *MUSH* in the shadows.
Now, when the same camera shoots in binned mode, which cuts density in half to 19.5 MP/cm sq (just above the 14MP/cm sq) of the F11, things change.
A rather dramatic difference. On brighter monitors, all the detail in the shadow is visible and it looks rather stunning. Almost G10 like.
But wait … there’s more. What if you drop the pixel density one more time, to, say, 33MP/cm sq and 16.5 MP/cm sq when binned? I am of course referring to the F70EXR. That’s basically F11 territory in the binned size.
The larger one at 10MP:
The mush has returned, but is less obtrusive than that of the 12MP image from the F80EXR, since the density is lower and NR cuts in as details get smaller … that makes higher densities Fuji’s enemy!
Binned at 5MP:
Even more detail present. And there is the rub … Fuji could be making stunningly good cameras with all the modern accouterments – long zoom, high dynamic range, HD video, and so on – but they always choose to try to squeak out a wee bit more pixel density … and they always push it beyond where the image quality can hold.
The F70EXR has excellent detail under all lighting, the F80EXR does not. The HS10 has pixel density just below that of the F80EXR, and we know *exactly* what that means. WHen the light is not absolutely *perfect*, the details are rapidly lost.
That’s what that poster is seeing, and there is nothing for it. You either live with shooting only in good light and not looking at the pixels at 100%, or you shoot something better.