Just catching up on these exchanges and information. Thanks to all for posting the latest links and comments here as they will be a big help for people like me considering a purchase. For now, it seems the case to upgrade from my two 80D bodies is not compelling. I don't particularly want the extra pixels of the sensor as I suspect the optical limitations of the 100-400 Mark II that I use will not produce much improvement in detail or ability to crop. Perhaps I'm wrong? I don't know. It may be a different story if I was using a big white prime. Either way, I'm happy to sit and wait this one out a bit longer.

Of relevance to the discussion about AF hunting, I was recently in the field (Gluepot) with Tim Van Leeuwen and Con Duvestyn who are both using 5D MkIV bodies. In strong backlit situations at some hides at Gluepot, focus hunting and inability to lock on the subject, especially small subjects was a significant issue. Interestingly, the 5D IV was also struggling in these conditions. Tim was using AI Servo. I was using single shot AF. Both struggled. For some action shots, I eventually decided to pre-focus and switch AF off. This seemed to work but would not work in most situations. Certainly not for BIF.

Glenn

Hi Glenn, hope you enjoyed the trip!

On the lens limitation issue, I am not sure about this. I think you have in your mind the DxO 'perceptive megapixel' concept which suggests that the lens is only allowing a percentage of the overall 'theoretical' resolution to be realised, in effect turning the 24 megapixels into (if memory serves correctly) about half that in real life. I find this concept quite confusing. While I understand what they are trying to achieve (given the problems and complexities of understanding MTF numbers) it is also evident that the perceptive megapixel measurement changes depending on which body the lens is attached to. It attempts to measure the resolution of the overall package rather than the lens itself. This suggests that having a greater sensor resolution (ie megapixel count) would result in a different and probably higher perceptive megapixel reading. That said, as I have noted, the 90D is only has about 15 percent more pixels horizontally and vertically, so the overall resolution gain is not spectacular. I don't think that having the extra resolution is a bad thing, particularly given that tests seem to suggest that high ISO rating (at least up to about 1600) is no worse than the 80D. Whether this is worth the outlay for the 90D, though, is the question given that other upgrades seem mainly in the live view experience.

I think there are two distinct (but related) issues with through-the-viewfinder focusing. The first is the focus hunting issue. As you know this is inherent in the DsLR phase detection focus method, as the focus sensor struggles in low light and in low contrast situations. Your comments suggest that it is no better in higher end models than in the 80D. This is very old technology now and perhaps we are at a limit where the cost benefit ratio is simply not worth the redesign of the sensors to increase the sensitivity, given the decision to put resources into mirrorless. The light sensitivity of phase detection focus also means that the number of focus points available is reduced as the aperture gets smaller, and this focus method often does not work well at apertures smaller than f/8. At 400mm, the 100-400 is at f/5.6 which is about as low as you want to go and still get reasonably good reliable focus in low light. I have found the 80D to be fine in straightforward situations with well defined edges and contrast - fast and accurate - so its not a huge problem, but from time to time as you note, it gets completely fooled. Clearly (and perhaps reassuringly) its not unique in this. Likewise, I think that the contrast detection system in mirrorless also struggles in 'uncontrasty' situations, as this uses the point of highest local contrast as the focus point. But there does seem to be steady improvement in both focus speed and sensitivity of mirrorless as companies develop their on-sensor systems, to the point that I expect that on-sensor will be preferable in terms of both speed and accuracy (we are perhaps already at that point).

The second issue is focus tracking - ie the ability of the focus system to follow a moving object. The 80D (and the 90D) seem to have problems both with the ability of the focus to 'hang on' as the subject moves, and the ability to intelligently track the chosen object - when viewfinder focus is used. I expect the 5D series (and the 7D2) are better in this regard, but have never used them. In this area, the live view experience is very good (if ultimately not quite as good as Sony's most recent iterations) particularly in terms of 'colour tracking' for things like face and eye detection. I've tried this in live view on the 80D and it seems to work pretty well, and by all accounts the 90D is better. In contrast, I find the viewfinder experience to be pretty hopeless with the focus points easily confused by background. Its better for birds against the sky where the bird is the only 'hard' object in the frame.

You must have done a lot of BIF on your pelagic - how did you find it?

I may well end up getting the 90D because I want to have a second body as a back stop when travelling (which would then be my current 80D). However I will wait and see how prices go in the lead up to Black Friday, Christmas deals etc. The M6 Mk2 looks surprisingly competent (and apparently works well with EF lenses with Canon's adaptor) but I don't like optical viewfinders.

Simon

Just a quick response: I didn't go on the pelagic. I'd left by the time it was run as it had to be rescheduled due to weather. Either way, I've done a fair bit of BIF and do find the 80D to be weak on tracking. I have nothing to compare it with though so perhaps my expectations are unreasonable. There are also quite a few parameters to tweak for tracking and perhaps I have the mix wrong (although I have tried different combinations in the past and have reviewed Ian Wilson's article on this on occasion). Either way, I do get some keepers from BIF and some 'good' shots although very few 'exciting' shots.

As for the extra pixels of the 90D and whether this will be of practical benefit, I can't comment further. If you do get hold of one, I'd be very interested to see if this does prove a significant advantage.

For fun I downloaded a few raw files from Imaging Resource to look at and compare the 80D and the 90D to see what difference the extra 8 odd megapixels makes. I used files from one of their standard still life series showing a range of different objects, such as bottles, brushes and fabric swatches. I downloaded files for ISO 200 and ISO 800.

The 80D files are in the .CR2 format and are about 31-34 Mpix in size. The 90D files are .CR3 format and about 41-43 Mpix. As far as I am aware Imaging Resource use the same 70mm macro lens for both cameras.

The only program I have that reads both files is Canon Digital Photo Professional, so I opened and compared files in this. I used the Neutral picture setting, with all sharpening and noise reduction turned off.

There is a small but visible difference in resolution in these sample files at 100% viewing, in the 90D's favour. Obviously the 90D file looks slightly bigger on the screen at 100% which accounts for some of the perceptual differences. Overall the 90D is better at resolving what I will loosely call 'line pair' detail ie the ability to resolve subtle differences in things like brush bristles, and slightly clearer, more legible text. Also, in these sample files, one of the things that distinguishes really good sensors from others is the ability to resolve the individual thread patterns in fabrics. Here, the 90D is better, with the weave pattern visible in fabrics to a higher degree than the 80D (in some cases showing the weave whereas the 80D does not). There are also subtle differences in the look of the files. In particular the 80D seems to have slightly more local/micro contrast with less subtle differences between light and darker local areas. The way that reds are rendered is also a little different to my eye - again the 80D seems to show a bit more contrast in red patterns. Noise is not a significant issue in either at these ISO settings.

I also looked at some of the DP Review test files on their web site again (note- not downloaded) and notices similar differences; slightly more detail in fine black and white lines, for example.

There is certainly not an 'oh wow' difference like you get when you compare massively detailed files of these still life images from the latest full frame or medium format cameras from Sony and Fuji; as you would expect, the details are modest and at common output sizes (for screen display, BLP submission, A4 prints) I expect most of us would not be able to pick one from the other. If you are a satisfied 80D user, I am not sure that you would find a compelling case to rush to get the 90D purely for the better image quality. On the other hand, if I were in the market for a new APSC camera, I think the modest improvements in the 90D would be worth getting albeit it's somewhat more expensive than the 80D. I don't know how long both models will be available together before the 80 D is phased out. I've not compared with the 7D2 but the relevant raw files are available for download for anyone interested.

Simon