Wednesday, April 13, 2011

F550EXR versus F200EXR – Low light shootout #1 – ISO Ladder in good light at ** High Resolution** — Review Part 31

This is where the rubber meets the road. People have been lamenting Fuji’s strategy of reducing sensor size with gnashing of teeth and wailing over the loss of the great 6mp sensor of the past (despite their being housed in bodies that feel like shooting a brick.)

So when the F550EXR was announced with yet another 1/2” sensor instead of the 1/1.6” sensor from the F200 and S200, there was a predictable outcry from the F200 aficionados (that words sounds so much nicer than zealot.)

But Fuji changed the game this time … they brought out a BSI sensor with EXR overtones. And a spectacular improvement to their JPEG engine to boot (this was long overdue.)

Some felt that the F70EXR held the F550EXR to a tie in my head to head of the 4 long zooms (all of which I own at this time.) But, of course, you had to really stretch to say that the F70 was a winner … I think it lost out pretty obviously as ISO got really high. And that was without drawing on RAW.

This time, the challenge is great as the F200 is known to be the best gig in town when it comes to F series enthusiast cams for low light shooting. Can the F550 actually take on a larger sensor and win? Well, stay tuned.

This ISO ladder is a little different from past ladders … in fact I shot 6 images for each camera and will use three crops from each image to demonstrate how the cameras resolve various types of detail and handle low contrast details plus edges. Each camera is shot at the exposure its meter wants in M mode. Each camera uses auto white balance, but that’s not a problem since the light coming in behind the blinds is sunlight and perfectly balanced.

The F550 is downsized to 4000x3000 to match the F200. Thus, the F200 is not disadvantaged at all. A tiny amount of sharpening is added to the downsized images to account for interpolation edge softening.

The goal of the test is to determine if there is more detail in one or the other crop … not to make subjective assessments about tone or color. So there is an attempt to equalize tone across the crops. Other than that, the jpegs are left alone, so the inherent details are as they come from the cameras. The RAWs are heavily processed in Silkypix, ACR6 and CS5 in order to get the best possible image.

So on with it … and note: you should click through to see the crops as they are 1500px tall and 900px wide. They are 1:1 crops and thus represent what you would see if you printed an image at about 40” wide and looked at it from about 20” distance. So, a bit unrealistic, but then these tests are generally only relevant to people who make moderately sized prints anyway.

Update 25 April 2011: Wolfgang, in the comments, suggests that the RAW images are in fact quite poor. Such things as jaggies on the crayon diagonals and saturation differences masking the little “2” shaped thread in the middle of the wool. So I thought I would do two things to answer … first, I will add two macros from the D700 – one of the coin and one of the wool. These are full sized and have been sharpened hard to bring out every bit of detail. They show what is there so we know what we are comparing. Second, I will add, after the set of ISO 100 crops, a reprocessed version of the ISO 100 RAF file. It still has jaggies, but what I noticed is that there is a fine balance between getting the sharpness and getting the jaggies. I chose to pull more detail and live with the jaggies. The key point is that, for these jaggies to be visible to the naked eye in a print, the print would have to be almost 40” wide. And you’d still have to get uncomfortably close to see them. So this is a bit of a red herring. The saturation differences are real though, Silkypix default to very high saturation. It is a difficult converter to control well.

D700 Coin and Wool

Click through to see full sized images. Focus is not perfect as these are very close macros shot with Sigma 105mm 2.8 and Canon 500D. The top to middle of the coin is the focus point. The little feather on the right of the red is the focus point in the wool. They look over-sharpened here because the browser is shrinking them, they were not properly reduced in size for display using interpolation.

ISO 100

RAW reprocessed

I see a tie in jpeg except for extremely fine details in the flat part of the coin and the texture of the surface of the tip of the dark crayon in the middle. This would be almost invisible with a print because of ink bleed, but it certainly is visible here. But very subtle.

The RAW brings the F550EXR into a full-on tie in all details. It also brings better saturation.

ISO 200

This is a tie across all three with the RAW again showing better saturation (I won’t mention it again as it exists at all ISOs) and this time it shows a tendency to handling very fine details better than the F200, which shows ever so slight a smoothing here.

ISO 400

Again the jpegs are near-clones of each other. The F550 shows a slight tendency to very fine grained noise in surfaces. This is not something that is all that visible and certainly not in moderate sized prints, but it is visible in a crop here. The RAW image is now clearly showing better fine detail, the surface of the coin being a clear example.

ISO 800

Here, the very fine grained noise in the jpeg does reduce its impact slightly. The amount of detail loss is minimal, but measurable. So the F200 could be consider better here, assuming you were making a huge print (and even then, these are very tiny details.)

On the other hand, the RAW image is smoother than the others and retains the detail, although there are a couple of spots on the paper of the crayons where it looks like a tiny dot has disappeared.

For most people, this will be pretty much a tie.

ISO 1600

Now it gets interesting ….

Here, the F200 has lost control of the noise. Yellow blotches are hinted at on the coin and heavy grain and edge destruction are ruining the crayons and hurting the wool. The one area that looks decent is the details in the coin … edges there are working out ok. The RAW here is a pretty clear winner.

And finally ….

ISO 3200

The contest is over here … the F200 has rampant color noise and grain and the edges are gone. The F550 jpeg has finer grained noise (well understood to remain more pleasing to the eye longer) and there is still little chroma noise. The RAW is much better than either except for an over saturation of the red.

So … some things are obvious:

  • In good light both cams are excellent to 800 ISO at high resolution.
  • With careful processing, 3200 ISO is reasonable in RAW with the 550.
  • Jpeg shooters who make moderate prints will be very hard pressed to see any difference between the F200 and F550, assuming that the colors are matched etc.

The F200 still makes lovely images, but it is no longer the defacto Sherriff in this town.

And just to make my point about moderate prints and web images crystal clear, here are the three 3200 ISO images at 800 pixels (click through to see them at 800px wide.)

F200EXR ISO 3200

F550EXR ISO 3200

F550EXR ISO 3200 RAW

No comments: