p.1 #1 · IS generations in a side-by-side comparison
Background:
I have the 70-200/2.8 IS for about 1.5 years and enjoy it a lot. No complaints at all. That is, until my last trip. It turned out to be too heavy so I left it at a friend's hands and took his 70-200/4 (non-IS) for the trip. It turned out to be a wise decision as most of the time I only used the 17-40/4.
That got me thinking. Is it worth to have a lens which I can't take with me to all places? To help myself answer this question I bought the 70-200/4 IS. I will be testing both and using both for a while until I figure out which fits my shooting style best. As I have my eye on the 135/2, I guess the 2.8 will be the one to go. However, that will all be in the future. In the meanwhile I will post some comparisons between the two. This is the first one. Subject: IS. How can Canon's first lens with third generation IS compete with Canon's first lens with forth generation IS?
Testing:
I was shooting handheld with both lenses set to 200mm and camera set to Tv. I was shooting JPEG and apart from cropping, no PP was done. These are 100% crops from the center.
As you can see, speeds go from slow (1/30) to ridicules (2 sec). Reason: I was curious.
My conclusions:
1. Both generation are very impressive.
2. Yes, forth generation IS is indeed better than third generation IS.
3. No, it is not that important. Reason: The photographer is also a variable. The 1/2 sec pictures are very similar whereas in the 1/4 sec pictures the f/4 shows it's clear superiority.
4. These speeds are really getting ridicules….
Happy shooting,
Yakim.
P.S. If - for any reason - you can't see the pictures use my login. Username: yakim. Password: alon.
p.1 #2 · IS generations in a side-by-side comparison
Yakim,
thanks for this very interesting test! Did you try several shots at the same shutter-speed for both lenses? I'm asking because I'm sure that there will be some shot-to-shot-variation, which may skew the results.
Overall, I agree with your impression that the f/4 seems to do a bit better than the f/2.8. It's pretty amazing what image stabilization can do! Just for fun, you should also add comparison shots with IS turned off!
p.1 #3 · IS generations in a side-by-side comparison
Rainer wrote:
Did you try several shots at the same shutter-speed for both lenses? I'm asking because I'm sure that there will be some shot-to-shot-variation, which may skew the results.
No.
I wanted to imitate a real life situation.
See point #3.
p.1 #4 · IS generations in a side-by-side comparison
Tests of this sort are interesting, but unfortunately not very reliable. While it is possible that the IS on the f4 is superior, handheld tests are probably not accurate enough to show it without a large sample of shots and statistical analysis of the results.
This can be seen by the fact that the f4 1/2sec image is so much worse than the f4 1sec image. Should we take that to mean that the IS works better at 1sec than it does at 1/2 sec. The more likely explanation is that your handshake is inconsistent and you were shaking more at 1/2 than at 1sec. With that degress of inconsistency, how much trust can we place in the other results.
I understand your statement about wanting to create a "real life" test, but you introduce real-life inconsistency into a process that is better carried out without such inconsistency.
p.1 #5 · IS generations in a side-by-side comparison
Hand-holding is a little bit like rolling a dice. If you roll it once, and you get a "4", do you conclude that that particular dice always rolls "4"s? Certainly not! In the end, there is a certain probability for getting sharp shots when hand-holding at a certain shutter-speed. And even this will depend on whether you're at rest, or out of breath because you chased that deer through the forest!
Actually, finding your own probabilities for sharp shots is not a bad idea. Let's say you know that at 1/30 s, you get 90% or more sharp, at 1/8 s it's only 50%, and at 1/2 s it's less than 10%. This will help you to decide when to bump up the ISO in order to get a shutter speed that will give you a sharp shot.
p.1 #6 · IS generations in a side-by-side comparison
Yakim, I wonder if you shot ten shots with each lens at a threshold speed where the 2.8 performed badly but the 4.0 performed well (looks like 1/8sec is a good choice), what would the results look like? Might you get 50% keepers with the 2.8, but 80% with the 4.0? Or 0% with the 2.8 and 100% with the 4.0?
I agree that consistency is the big variable with shooting IS handheld. I've gotten 1/4sec shots out of my 28-135, but I don't bank on it!
p.1 #8 · IS generations in a side-by-side comparison
Very interesting test Yakim. While obviously hand-holding tests are not going to be nearly as accurate as tripod tests, I do find it useful as I shoot mostly hand-hold shots. Not only does it compare 3rd-gen IS vs. 4th-gen IS but it also shows the difference that weight makes. I'm sure some of the blur from the 2.8IS is caused by its weight (1470g of the 2.8IS vs. 760g of the 4IS... almost twice the weight!)
I hope you'll have a chance to compare hand-held low-light action shots... obviously the F2.8 has the advantage there, but I wonder if it will blow away the F4 since the F4 has the advantage in lower hand-shake (IS and weight).
p.1 #10 · IS generations in a side-by-side comparison
Stunnaz wrote:
I hope you'll have a chance to compare hand-held low-light action shots... obviously the F2.8 has the advantage there, but I wonder if it will blow away the F4 since the F4 has the advantage in lower hand-shake (IS and weight).
The f/2.8 is vastly superior here. The F/4's AF speed in low light is unacceptable, at least in my book.
p.1 #11 · IS generations in a side-by-side comparison
nathanlake wrote:
I understand your statement about wanting to create a "real life" test, but you introduce real-life inconsistency into a process that is better carried out without such inconsistency.
You are right but look at point 4. I - personally - am not going to shoot at speeds lower than 1/30, with any lens. Reason: The probability to get a good shot is greatly reduced. The results of this test prove this.
p.1 #12 · IS generations in a side-by-side comparison
I know from experience that the 70-200/2.8L IS gives me 3 stops over my own hand-held non-stabilised performance. That's exactly what Canon claims for this lens, they say it can provide "up to three stops".
The 70-200/4L IS is Canon's first 4-stopper. Al that I've read seems to confirm that those "up to 4 stops" are there for the taking, it's no hollow claim.
Your testing seems to confirm this in part: It shows the 70-200/4L IS is better, but it doesn't really quantify how much better it performs.
