50D Sharpness/Resolution - my test results

In the process of microadjusting my lenses I developed my own test target, as I was not really happy with the targets I'd found at Northlight Images and the like. I constructed a simple target, using Microsoft Paint, to draw simple box shapes and straight lines in various colours, interspersed with plain black lines. I'm not sure how many pixels wide the finest box edges are but I'm assuming 1 pixel, or 2 at the most.

The target starts out from the centre, with boxes/lines at the very finest pitch, surrounded by more boxes/lines one size up, and after a few of those, more boxes/lines another size up. Finally these carefully prepared boxes/lines were surrounded by further boxes/lines of arbitrary width, with the occasional fine line (smallest size) thrown in in black or white. Here is the test target shown at 100% size.

HTTP response: NOT FOUND | MIME changed to 'image/gif' | Redirected to error image by FLICKR

HTTP response: NOT FOUND | MIME changed to 'image/gif' | Redirected to error image by FLICKR

HTTP response: NOT FOUND | MIME changed to 'image/gif' | Redirected to error image by FLICKR

HTTP response: NOT FOUND | MIME changed to 'image/gif' | Redirected to error image by FLICKR

HTTP response: NOT FOUND | MIME changed to 'image/gif' | Redirected to error image by FLICKR

HTTP response: NOT FOUND | MIME changed to 'image/gif' | Redirected to error image by FLICKR

How long it took to draw that thing with a basic tool such as Paint? It reminds me of some old video games I used to play in the early '90s.

A pixel is formed by three LCD imaging elements. Red-Green-Blue.
You may want to get even closer to capture those individually.

SO did you adjusted the 100-400 w/wo the 1.4x kenko?

adas wrote in post #6709056
A pixel is formed by three LCD imaging elements. Red-Green-Blue.
You may want to get even closer to capture those individually.

I have captured them individually. That was exactly the point of drawing red lines, green lines and blue lines independently, rather than just alternatng black/white lines (although those are present too). That is also the reason I chose to paint the coloured lines against a black background - so that the lit pixels would stand out cleanly. By the way, I disabled my monitor calibration, for the test shots, in the hope that my reds, greens and blues would be presented as pure red/green/blue and not be contaminated by colour corrections from the monitor profile.

Remember I embarked on this exercise in order to calibrate my microfocus adjustment, not to prove that I could resolve individual sub-pixels. I only posted this thread because of the typical whining you get from so many people saying "My 50D (40D) is not sharp.". My point is - how sharp do you need it to be? If I can see sub-pixels on a high density display from 20' away with a soft old lens like the 100-400 at 400mm and f/5.6 I'm happy enough. From these results I'd say the 50D is plenty sharp enough. For people who say the 50D is soft, I wonder what on earth they are expecting.

I expect my primes will be sharper, but I'm not posting any more "proofs of sharpness". I know the "camera" is sharp. I'm only interested in calibrating my lenses as accurately as i can.

By the way, I figured that test target was over-complicated so I've developed a simplified version that should be equally up to the task. If anyone wants a copy I've attached the original and the revised version as BMP files within a ZIP file, since the website won't permit BMPs to be uploaded.

silvex wrote in post #6709090
SO did you adjusted the 100-400 w/wo the 1.4x kenko?

I did check the adjustment separately both with and without the Kenko. The problem is that in order to preserve AF function with the Kenko fitted you need to tape the pins so that the camera can't see it (otherwise AF is shut down by the camera because your f/5.6 lens becomes f/8 ).

The calibration needs are slightly different with/without the Kenko. For the bare lens I need between +3 to +4 dialed in for ultimate sharpness. With the Kenko fitted I need somewhere between +6 to +9. It is hard to be completely accurate because I find it too hard to judge perfectly by eyesight alone, and, to be fair, if I can't even tell the difference between +6 and +9 when viewed at 100% I'm sure it doesn't really matter much which setting I choose within that range.

I was guided in my choice of setting, not only by visual inspection, but also by checking file sizes. I did actually take photographs, captured to raw, and I did notice a trend in increasing file size the sharper the image became. This allowed me to see a statistical pattern emerging, with a sharpness hump (largest file sizes) where the sharpness was greatest. Thus, even allowing for a slight focus blip, the trend guided me to where I should set the adjustment.

So, back to my teleconverter and the fact that it is taped in order to preserve AF function..... Well this means that the camera does not know the whether the teleconverter is fitted or not. So in the real world I have to remember to manually change the adjustment from +4 for the bare lens to (say) +7 or +8 for the teleconverter if I want the ultimate sharpness. Alternatively I could just set it to +6 and leave it there.

Obviously, if I wanted to make the camera aware of the teleconverter, so that it could switch back and forth between the two settings, automatically, then I would lose AF anyway as the camera would see the 560mm f/8 lens. It's a catch 22 situation. So I need to leave the teleconverter taped and make the adjustments by hand. I do wish Canon would make it custom function option to enable/disable AF with an f/8 lens fitted. We all know it is not supposed to work, and nobody should blame Canon if it doesn't, but for those of us who find that it does work, it would sure be nice to see correct EXIF data and not obfuscate things with tape and lies.

You adjusted your micro-adjustment according to what you did on-screen??

sando wrote in post #6709158
You adjusted your micro-adjustment according to what you did on-screen??

I don't understand what you mean. I adjusted my AF based on visual inspection of the captured files at 100%, backed up by the file size trends as confirmation of the sharpness peak.

Without the Kenko I shot 26 images, from -10 to +15 in increments of 1. I knew it was soft at -10 but just wanted to get a broad range and eliminate statistical error.

With the Kenko fitted I started at 0 (I could see that was soft so it was pointless to go back to -10 as a starting point) and went up to +15 in increments of 1.

Holy psychedelic 60's Batman ..