Perhaps it's because the difference in resolution isn't sufficiently great. But I think there's a better explanation: if you take a cluster of noisy pixels and average them out into a single pixel, you still get a noisy pixel. The process of averaging is NOT any kind of noise reduction.
Now, you may argue that the resulting noise is less noticeable but the loss of detail due to smearing is also noticeable, as you can see above.
Repeating the process for cameralabs.com:
As you can see, same outcome. The downscaled 50D image has less detail.
- Canon EOS 40D with BG-E2N battery grip
- EF-S 17-85mm f/4-5.6 IS
- EF 70-300mm f/4-5.6 IS
- EF 100mm f/2.8 macro
- EF-S 10-22mm f/3.5-4.5
- EF 50mm f/1.8 II
- 580EX II external flash
- Cameron/Marumi DRF14 macro ring flash
- Induro C-014 carbon fiber tripod with SA-0 ballhead
- Lowepro Fastpack 250 camera bag
- Canon PowerShot S90 backup camera
For an amateur photographer, and a novice at that, this is pretty good gear. C$5,100 worth.
(For the same price, the EF 70-200mm f/4L is much sharper than the 70-300mm. However, the latter has Image Stabilization, or IS, and a longer focal reach. IS allows for handholdability, without the need for a tripod, and that's very convenient.)
The 40D body alone was effectively C$1,000, slightly more expensive than the consumer-oriented Rebel T2i that you can buy today. But in return, you get:
- durable magnesium alloy construction
- comfortable grip
- ergonomic control wheel
- large bright pentaprism viewfinder (0.95x magnification)
- better AF (9 cross-type points)
- 6.5 fps shooting speed
- top LCD info display
Otherwise, the 40D is a fairly basic body, without such amenities as built-in vignette correction, contrast-detect AF, and VGA screen.
I am primarily interested in wildlife, landscape, and macro photography.