dSLR autofocus ability...

You ever printed your 630's output at the equivalent size of the D-60's 100% on-screen size before comparing?

Not that that is relevent anyway, since the 20D is 2 generations more advanced than the D-60.

thats more about resolution though, and cameras today resolve to incredibly high levels...

Yes, they do resolve very well. And that's part of the problem.

A lot of people look at the 100% screen crop and determine that the focus is off by a few mm on an image that, if printed at some normal size, wouldn't be visible. That is to say, they are expecting AF performance based on a circle-of-confusion much smaller than what is utilized in the "standard" 8X10 or 6X9 or 11X14 print (viewed at a distance equal to the diagonal of the image) that most camera manufacturers use for determining focus accuracy.

I call it the curse of the 100% crop.

indeed. although I've had a couple of A2 or bigger prints from my P200 and they are stunning...and thats 7mpix, granted the lens isn't L quality but its still good zeiss stuff. had some very very good shots and at 100% yes you can notice the break-up...if I got a dSLR (after the disaster of the first i'd have to choose wisely) I'd be printing at quite big sizes and using the high res as such, good glass obviously helps. as a result an AF system thats accurate is quite high on my list. the 20D I tried wasn't doing it for me, perhaps the 'next' one will...? rumours are starting to fly in now so its just a matter of time...

I'm amazed anyone ever gets a decent shot from any camera based on this discussion. It's like a how much fluff can we pick out of each others belly buttton before we realise the only people who know the answers will never tell their customers. What I do know is I get better shots on full manual using manual focus and half decent glass than I ever got with Canon gizmo-ology

true true, this is a way around it no doubt but then some times i manual focused and it still produced an oof shot...sounds like either I need my eyes checked or the focus plane was out a touch.

Yeah, my 350D managed that consistently this morning when I tested it... shortly before it went back to Canon....

There is some real misinformation going in this thread about pricing products. A film camera in the old days was a mechanical machine. The new digital cameras are electronics. Having spent many years in the electronics and semi-conductor industries, and also working in marketing, I would say that some of you are really off base in this thread.

First off, companies are in business to make money for the stockholders. That is why they exist. At some point, someone decided they could make more money by making and selling cameras, than they could get by putting their investment money into a bank or some other business. The customers are a means to an end... very important, but the stockholders are why businesses exist... so that they get a return. All of the recent attempts by the media to make big business look bad have skewed reality.

Back to electronics. Electronics components, specifically semiconductors, have a large up front investment cost and a manufacturing cost that goes down over time. The reason for this is that early in the life cycle the effective yield of the product is pretty low (they may get only 10% good out of a batch) and steadily goes up until the product is mature (they may get 90% good out of a batch) and eventually becomes obsolete. If they priced their product day one to fully compensate for the higher cost, no one would buy it. So, they price it with a lower profit... profit steadily climbs and the company slowly recoups their full investment and makes a net profit. Then based on competitive pressures, the price may go down over time until a new product is introduced.

It cost way more to make a big sensor on a 1DsMk2 than it does to make a tiny one on a P&S. One of the big reasons for that is that there is going to be alot lower yield (higher cost) on those big sensors early on. This is why the huge LCD panels cost so much to produce. The bigger it is, the greater the odds that there is a single bad part of if that scraps the whole thing.

In a lot of technology markets, companies can actually produce faster/better stuff but will hold it back until the market catches up and the competition is pushing them. This way they are bringing it into the market with a lower cost.

With memory chips and processors, they will make and batch and then test them to see which speed tests they pass. The ones that pass the higher speed tests are given a higher price tag. The ones that can't pass the higher speed tests are given a lower price tag. It costs the same thing to make each of them, but they may get a 3x price on the fastest. Lets say that their yield goes up til they are getting almost all the chips to pass the faster test. Well guess what, they may have to take some of the fast chips and mark them with the slower product part number and lower price. The reason is that there is a market for each, and they need to serve both markets. They are NOT going to mark them all as fast and lower the price on all of them. That would be a silly way to run a company, but obviously really good for the buyers. This fact is why overclocking of computers is so popular. People will try to find the faster chips and run them faster than their specs say they will go. Often times they are successful.

So Canon is a big company and wants to make money. They are not going to take all the technology that exists in the 1DsMk2 and slap it in a P&S body and sell it for $99. Who knows if they could do this or not... it wouldn't be a good business decision. There are consumers that want a $99 camera and don't expect it to be a pro camera. There are pros that pay top dollar to have the better performing camera.... market demands, and value to the customer. It is worth it to the pro to pay more. It is not worth it to the consumer to pay more in many cases. For those that it is, they buy the pro product.

By the way, Canon ranked number 2 this year behind IBM for the most patents files in the US Patent Office. IBM had 2941 and Canon had 1828. Canon moved up from number 3 last year, overtaking Matsu****a, as did HP that moved from 4th to 3rd. Of the top 10, 7 are Japanese companies. IBM, HP, and Intel are the only US companies in the top 10 filing patents in the US.

I think that small viewfinder of 1.6 crop cameras is one of culprits of misfocused images. On 35mm or FF cameras it's easier to estimate if picture is OOF and to make correction – especially with FTM lenses.
Maybe slight OT, but what are experiences with other dSLR systems: Nikon, Olympus, Pentax... I guess they are even less accurate and definitely slower than Canon (except maybe Nikon) as they use motors inside bodies?

surely USM is a motor ?

I agree ...I suspect if many took a day using only MF they would find that their keeper ratio increased significantly. They might also find their approach to a subject changed ... perhaps for the better in some cases.

Yes, that was always my intention as well Scott. I intended to either update to the 10D, mainly for the AF improvements, or the 20D, finances permitting. A full frame DSLR is out of the question due to costs - I simply cannot justify the artificial prices that Canon is demanding for their full frame DSLRs.

Even 2nd hand, these cameras are ultra expensive, as an example, a 2nd hand 1d mark II is going to set me back at least au $6500 (and the 1ds range you can add another $1500 on top of that). My good ole eos1n was a fair bit cheaper, and is a pro camera. Even allowing for inflation at 2% per year, for the past ten years, the cost of the eos1n still doesn't approach the cost of either the 1d or 1ds (and that's 2nd hand) - I paid au $4500 new, with inflation set to 2% per year (which is higher than what the inflation has been in in the Australian economy for most of that time anyways), after ten years the cost of $4500 goes up to $5485.

My take is that Canon is cashing in on this digital demand/craze with artificial prices, and it knows people will pay it because they're desperate. They're locked into the Canon lens system, and it's far too expensive to swap to another marque's lens range. I was telling a mate about the Canon dilemma, he's not a photographer, and his very first comment was, switch brands. I can honestly tell you now, if I didn't have my investment in several Canon L series lenses, I'd switch to Nikon in an instant.

I had a play with the 20D and the AF seemed fine in the camera store, but so would the D60 in that type of scenario. The AF was fast, subjectively speaking I don't think as fast as the eos1n. I didn't try the AI Servo, and of course the lower lighting focussing comes into play. If I was to consider the 20D, I'd really want to test it more thoroughly before buying, otherwise I'd just grab a 2nd hand 10D (cheaper and pretty near the ability imho) and spend money on glass!

I hope the next version (30D) has good improvements to the AF, spot metering, weather protection etc. I guess I'm gonna have to settle for the fact that nothing but the pro models are going to have a really fast and decent AF or full frame sensor. Pity.

Dave

...on a side note I read about the difference between the 20D and 1dmkIIN low-light performance here:
http://www.digitaljour​nalist.org/issue0512/w​estfall.html
a bit down the page...
interesting, is it a canon rep trying to justify the 1d's slower performance...? seems so, but then it just makes the 20D better for that situation, although the 1d beats it elsewhere.

The 20D and 5D are both able to focus at -0.5 EV, while the 1-series requires 0 EV. At least according to the specs.

When I had the 5D and 1D II together, I took some very low-light shots. The 5D was able to achieve focus more often than was the 1-series under those circumstances. There's plenty of situations where the 1-series is more capable, though.

Mike, what you wrote should be read by all those who push rumors of big improvements in upcoming specs. The changes are always gradual.