Generally, I find lens comparisons the most useful. A single pretty picture that looks reasonablly sharp is often not enough.
[a href=\"index.php?act=findpost&pid=173214\"][{POST_SNAPBACK}][/a]
Sure, but what a real useful lens test should do is:
1. Test 10 lenses bought randomly from 5 different shops within 2 or 3 months (to test different batches of lenses),
2. Provide results in terms of
average (that you can hopefully expect to the be the spec) and the
worst performer (that gives an idea of how unlucky you could get if you purchase a bad sample).
Nobody does that today as far as I know.
Testing one lens has a high probability of being enough if the manufacturer does its job in terms of quality control, but if they don't then simple unit lens tests become useless.
We are only to blame ourselves for supporting various publications in print and on the web that define such low level benchmarking standarts for items we are willing to spend x0.000 US$ on.
Indeed, in most areas of human activities, companies design their products relative to the established benchmark along one single criterion, be better than the competition where it can be measured. That's what people running these companies learn in their MBAs.
As long as we, consumers, agree that the benchmark for lens testing is the testing of a single lens by various small companies accross the web, we will keep making it very easy for companies to design lenses with the following criteria in mind: be able to manufacture a few samples with very high quality (those will be provided to testers) and optimize the costs so that most other samples are good enough not to generate complaints from 90% of users (those don't know how good a sample they could have gotten). Better samples are given to the 10% that do complain.
What would really help this industry is the establishement of a third part non profit entity funded directly by working photographers (or amateurs willing to contribute) that would pay a recognized certfication company (TUV comes to mind) to do random test of lenses according to a published methodology (the random test of many samples would be part of that). These test results would be made availalbe for free on their website. Wonder how I hadn't thought about it before.
This would provide tremendous value to all of us in:
1. Identifying these lenses that are really good (and not only those that can sometimes be really good),
2. Force the manufacturers to design for average quality and not for peak quality,
3. Streamline the discussions on forums thanks to the provision of cross brand objective information focussing on the interest of the photographer,
4. Give Ray a life.
Cheers,
Bernard