Marsu42 said:
LetTheRightLensIn said:
It has better image quality than the 70-200 f/4, 70-200 2.8 non-IS, and 70-200 2.8 IS.
It seems to be better at 70 and 200mm than the 70-200 f/4 IS but not as good in the middle (and much better than the 70-200 f/4 IS with a TC).
It really would be helpful if people were citing their sources, because after all I read (see link below) and the replies I got from asking around I think this statment is simply not correct - actually, to me it seems to be exactly the other way around. See for yourself:
http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=404&Camera=453&Sample=0&FLI=0&API=0&LensComp=738&CameraComp=453&SampleComp=1&FLIComp=0&APIComp=0
From a test that was very carefully carried out (for 70-300 L vs Tamron 70-300 VC vs Canon 70-200 f/4 IS).
Indoors, constant lighting, 25' distance to target (so no near MFD garbage messing the test up for normal usage), best of 6-12 manual focusing attempts using 10x zoom liveview for each aperture tested at each focal length tested, edge and mid-frame tests re-focused (doesn't reveal field curvature but on the other hand you don't get weird results because the setup was not parallel to the 0.001cm).
I doubt TDP used so many focusing trials, I don't believe they refocus away from the center so the mid and edge performance they test is a mixture of lens ability, field curvature and how off they were from parallel perfection (and it all it takes is a TINY tweak for results to be altered) and I believe they use a target much, much closer than 25' I think someone said they often use a target just 3-4' away for the lenses that can focus that close (not sure though). I haven't found a single other site that agreed with their claim that it has much worse CA at 300mm than at 70mm. For whatever reason I've found TDP results to not match what I see as often as not, although their newer tests match a bit more often (I believe they use to use AF and best of only 3 tries or something early on). And look at their Tamron 70-300 VC results where the 70-200 f/4 IS + TC blow that away at 300mm to a huge degree. That's not what to many others have seen.
The photozone.de results were much closer to what I saw between the 70-300L, 70-200 f/4 IS, tamron 70-300 VC, even if not exactly matching.
Of course some of it might be copy to copy variation in the lenses. It is possible that it may play a larger role than we think. Look at lensrentals who tested lots of 100 macro and 100L macros and they found the better 100 non-L outperformed the lesser 100L copies! Although the best 100L did better and, on avg, the 100L definitely did better.
It is probably best to use a wide array of sources what with testing and copy to copy variation.
Comparing the 70-300L to the 70-200 2.8 IS and non-IS were based upon how the 70-300L did against the f/4 IS and about how the f/4 IS did against the 2.8 non-IS in a past, and not as carefully carried out test, and how the 2.8 non-IS did compared to the 2.8 IS in yet another test, also not quite as painstakingly carried out as the recent test.
I see lots of reports and test agreeing with 2.8 non-IS doing better than the 2.8 IS and lots also saying that the f/4 IS does better than the 2.8 non-IS. There are also other tests including photozone.de and on some guy's blog, forget his name (EDIT: I see someone else posted the link, Dan Carr), where the 70-300L did better at 70mm than the f/4 IS. Even looking at Canon's simulated MTF charts you can see that this is predicted behavior.
But with copy to copy variation, who knows what your two copies would do for sure.
. 
