What would you do body or glass upgrade

[Franglais]

This is the boat I'm in. I have a canon 10d with a scratched sensor filter. $400 plus to fix. The lenses I have are of mediocre . No L lens only one 28-70 2.8 lens. I have really enjoyed the 10d.
My question would you replace the Canon 10d with a 50d and kept the lens I have. Or would you get another 10d and upgrade on the lens.

I know this is a vague general question but really wondering if a 10d compared to a 50d body will make that much difference with the same mediocre glass.

Ray Still

What do you mean by "the lenses I have are mediocre"? Can you give us a list?

I'm in a similar situation with my 6MPix Nikon D70 and my newer 10MPix D200 and 12MPix D300. I have a load of lenses that I bought back in the film days plus newer ones designed for digital. As each new generation of digital bodies comes along I do a test to see if they are still up to it:

- my designed-for-digital 18-70 that came with the D70 is my benchmark. It exceeded the performance of the D70 and even with the D300 it still hasn't reached its limit. The 18-200 is slightly behind but its the same story
- many of my designed-for-film lenses were OK with the D70 but even with the 10MPix D200 they got fuzzy, way behind the 18-70, and I stopped using them. (Examples = a plastic 28-100, 5-element 28mm f2.8, 24-85 f3.5-4.5).
- my 3 pro f2.8 constant zooms (17-35, 28-70, 80-200) are slightly ahead of the 18-70 and 18-200 but there's not much in it. The 18-200 replaces all of them in one small, light lens and that's what I use most of the time
- my prime lenses are only slightly ahead of the 18-70 at apertures the 18-70 can do. At wider apertures (f1.8) the 18-70 can't follow but performance of the primes drops off catastrophically. I only use them if I have to.

My advice: wait for the tests of the 50D with lenses. If you decide to get a 50D then get a lens that is designed for the camera. You might find that the new 18-200 from Canon is a possibility.

[Franglais]

I have been collecting lens tests for years (Chasseurs d'Images - excellent magazine). This gives me an opinion on most lenses on the market. But the important thing is my lenses on my cameras. Here's how I do my tests:

Always take the lens to be tested plus a benchmark lens - one that I use and whose performance I know. If possible test a third lens as well - fairly easy in my case as I have the consumer lenses, pro lenses and primes.

Use the same subject each time, with lighting that doesn't vary. I use the view out of my kitchen window on a bright morning. The subject is the house across the way (about 50 metres). This includes lots of fine detail like brick walls, leaves, shadows under the trees, white balconies with a dark wire grill..

I set the camera up on a heavy tripod (important).

For the lens under test I choose at least three focal lengths (widest, middle and longest) and three apertures (widest plus two useful ones like f5.6 and f8). I shoot the test lens at each setting then the Benchmark lens at exactly the same settings then the other lens.

I load all the JPG's onto the computer and I do side-by-side comparisons.

First a blow-up to 200%, looking at center and each corner. I'm looking at how well the detail is rendered (espacially leaves and the brick wall). The balcony with the wire grill is usually at the corner of the image. On a mediocre lens the grill is smeared and imprecise.

Then I look at the overall image. On an inferior lens the colours just aren't the same, the whole thing looks dull and lifeless. A superior lens gives some extra purity and vibrancy. If in doubt I do a few 8x10 prints.

I also check out the corners for signs of light falloff - pretty rare with an APS-C sensor. If in doubt I do a second test of just a patch of clear blue sky.

That's all. Each image is either the same, worse or better than the Benchmark result. I know what to expect.