Canon 5D Mark IV -- Time to Discuss!

CyberDyneSystems wrote in post #18141842
I have heard a number of stories similar to this one, and am still baffled by it as my own experiences have been totally contrary. On both my 500mm f/4L IS, and the replacement 500mm f/4L IS II, I saw no difference in AF performance with the MkIII 2X vs the MkII (until I got a 5D4 that is) with 1D4, 7D2, or 5D3 the behavior was much the same as i had always come to expect from the MkII 2X T-Con, (or even the Mk-I)

Perhaps not all 2xII TCs are created equal, or maybe it depends a lot on the amount of contrast in the scene, as the 2xII tended to have lower contrast, and could reduce it to difficult levels in some situations.

Tapeman wrote in post #18141852
About the only thing I miss with my MKII converters is the ability to stack them. When I had my 300 f/2.8L IS I would occasionally use them stacked to get to 840mm.
Now with my 500 I use the 1.4X when hand holding frequently, but the 2X almost always tripod mounted.
On the tripod my 5DSR has the advantage over my 5DIV due to its amazing cropping ability.
Canon's new converters definitely are better.

I kept the MkII 2X for that reason alone. Some day I may need 2000mm f/16

Tapeman wrote in post #18141852
About the only thing I miss with my MKII converters is the ability to stack them. When I had my 300 f/2.8L IS I would occasionally use them stacked to get to 840mm.
Now with my 500 I use the 1.4X when hand holding frequently, but the 2X almost always tripod mounted.
On the tripod my 5DSR has the advantage over my 5DIV due to its amazing cropping ability.
Canon's new converters definitely are better.

the 1.4mk2 is actually a smidge sharper in the center but the III is much much better in corners/edges and also controls CA better improving the overall clarity.

Scrumhalf wrote in post #18141802
John, please. Don't talk down to me. Okay? I am just using the colloquial term. I know all about pixels on subject. I'm fact, I made that very point in my post.

I am not one to get riled up over forum posts but, really.

I am talking down to the concept that crop = reach. In fact, I talk down to "reach" by itself, because there is no such thing, except getting the entrance pupil closer to the subject, and changing perspective.

sploo wrote in post #18141840
Somewhat relevant to recent discussion, I've knocked up a quick Excel spreadsheet for calculating equivalent focal length/crop/resolution factors. Yes, I know, a smartphone app or web page would be better, but I don't have time at the moment. You can get it here: http://spikyfish.com …valence%20calcu​lator.xlsx and it looks a bit like this:

Hosted photo: posted by sploo in
./showthread.php?p=181​41840&i=i244878686
forum: Canon Digital Cameras

Green cells can be edited, grey are generated. Some of the drop down data is on sheet 2, and can be changed for your tastes.

Set the crop factor for Camera 1 (1=Full Frame, 1.5=Nikon DX crop, 1.6=Canon EF-S crop, 2=Micro 4/3). Enter the camera's resolution and lens focal length, and if you're using teleconverters you can choose from a variety of one or more 1.4x and 2x factors. The total TC "magnification" will be represented in the Teleconverter effect column. Finally, the effective FF focal length is shown in the last column.

Now, assuming you have a subject filling the frame with Camera 1, you can find out what you'd need to do to also fill the frame with Camera 2.

In this example, Camera 1 is a 7D2 (1.6x crop factor, 20.2MP, no teleconverter, and using a 400mm lens). Camera 2 is a 5D4 (1x crop factor, 30.4MP, using the same 400mm lens, but with a 1.4x TC).

You can see that Camera 2 requires a further pixel crop of 1.14x to match the framing of the 7D2, and this results in a 23.28MP image.

The actual amount of sensor area used in this comparison is 1.96x that of Camera 1 (the 5D4 sensor is 1.6^2=2.56x bigger than the 7D2, but we've cropped a bit so we didn't get the full "benefit" of the extra area). This doesn't take into account the actual amount of light, which would be affected by the use of the TC on the 5D4.

A sensor area ratio <1 means you're using less sensor area than Camera 1. For example, with a 400mm len on both cameras and a 2x TC on the 7D2 (effective 1280mm in full frame terms), using no TC on the 5D4 (effective 400mm) would require severe 3.20x crop to match the framing - giving a 2.97MP image and use just 0.25 (25%) of the 7D2's sensor area.

I suppose it would be useful if it could take into account the actual amount of light for noise purposes - so maybe those in the know could chime in on how to update it?

You have to take into account the resolving power of the teleconverter, like approximately 80% resolution with the 1.4xiii and 60% resolution with the 2xiii. Those are rough figures, and adding those into the mix will definitely muddy the waters. 30mp like the 5DIV at capacity can probably only resolve 25mp due to the AA filter, ugh yeah, it's muddy.

A more simplistic train of thought, FF + 1.4xiii is midway between a 20mp FF sensor + bare vs 20mp + Crop + bare sensor if framed the same. 20mp FF + 2xiii = 20mp Crop + bare if framed the same. However 20mp FF + 1.4xiii == 30mp FF + 2xiii if framed the same? Theres a question if should we multiply 50% resolution before or after the total resolution loss? (60% x resolving power) x 150% or (60% x resolving power) + (50% x resolving power)... getting messy

Charlie wrote in post #18141882
You have to take into account the resolving power of the teleconverter, like approximately 80% resolution with the 1.4xiii and 60% resolution with the 2xiii. Those are rough figures, and adding those into the mix will definitely muddy the waters. 30mp like the 5DIV at capacity can probably only resolve 25mp due to the AA filter, ugh yeah, it's muddy.

It should be muddy. Ideally, a system should not perform so that it scores the same number of perceptual MPs as actual MPs; any system that does has too few pixels for the lens and can not place edges accurately in an image, and will snap them to the pixel grid. The AA filter is not the ideal way of doing things, but with low pixel densities, it is the only way to prevent luminance aliasing (it does not prevent red and blue channel aliasing).

The AA filter is bypassed quite a bit by using a TC. Yes, from a pixel-level perspective, the TC lowers acuity even more beyond the acuity-lowering of the AA filter, but from the perspective of the subject matter's scale, the radius of blur of the AA filter gets smaller, as does the scale of the color filter array.

John Sheehy wrote in post #18141955
It should be muddy. Ideally, a system should not perform so that it scores the same number of perceptual MPs as actual MPs; any system that does has too few pixels for the lens and can not place edges accurately in an image, and will snap them to the pixel grid. The AA filter is not the ideal way of doing things, but with low pixel densities, it is the only way to prevent luminance aliasing (it does not prevent red and blue channel aliasing).

The AA filter is bypassed quite a bit by using a TC. Yes, from a pixel-level perspective, the TC lowers acuity even more beyond the acuity-lowering of the AA filter, but from the perspective of the subject matter's scale, the radius of blur of the AA filter gets smaller, as does the scale of the color filter array.

I hope Im not the only one on here who has absolutely no idea what all that means

don1163 wrote in post #18142011
I hope Im not the only one on here who has absolutely no idea what all that means

Way above my pay grade

Charlie wrote in post #18141882
You have to take into account the resolving power of the teleconverter, like approximately 80% resolution with the 1.4xiii and 60% resolution with the 2xiii. Those are rough figures, and adding those into the mix will definitely muddy the waters. 30mp like the 5DIV at capacity can probably only resolve 25mp due to the AA filter, ugh yeah, it's muddy.

A more simplistic train of thought, FF + 1.4xiii is midway between a 20mp FF sensor + bare vs 20mp + Crop + bare sensor if framed the same. 20mp FF + 2xiii = 20mp Crop + bare if framed the same. However 20mp FF + 1.4xiii == 30mp FF + 2xiii if framed the same? Theres a question if should we multiply 50% resolution before or after the total resolution loss? (60% x resolving power) x 150% or (60% x resolving power) + (50% x resolving power)... getting messy

Clear as mud.

Glad I asked

Maybe if someone has both, if they can be bothered, try both and and report back with the vibe.

I would be happy with the Vibe.

Talley wrote in post #18141846
<--- this one is for you sir.

I don't usually swing that way, but buy me a drink and I'll think about it

Charlie wrote in post #18141882
You have to take into account the resolving power of the teleconverter, like approximately 80% resolution with the 1.4xiii and 60% resolution with the 2xiii. Those are rough figures, and adding those into the mix will definitely muddy the waters. 30mp like the 5DIV at capacity can probably only resolve 25mp due to the AA filter, ugh yeah, it's muddy.

A more simplistic train of thought, FF + 1.4xiii is midway between a 20mp FF sensor + bare vs 20mp + Crop + bare sensor if framed the same. 20mp FF + 2xiii = 20mp Crop + bare if framed the same. However 20mp FF + 1.4xiii == 30mp FF + 2xiii if framed the same? Theres a question if should we multiply 50% resolution before or after the total resolution loss? (60% x resolving power) x 150% or (60% x resolving power) + (50% x resolving power)... getting messy

Yes, this is a very good point, and one I certainly didn't take into account (in the 10 minutes I took to write the spreadsheet ).

I suppose with a significant bit of trawling through DxOMark's lens data it might be possible to build up a database of the actual resolving power of a particular lens (or that lens with a TC).

I suppose the critical issue would be: given the resulting size of the image (in pixels) vs the actual sensor area from which that image was taken (i.e. the pixel density) does that lens still out resolve the pixel resolution. If so, you're good. If not, the image will have a lower perceptual resolution than the actual pixel resolution.

For decent lenses, I thought it was usually a case that a TC will give better results than pixel cropping, but how you compare a FF sensor + TC + lens vs a crop sensor + bare lens is beyond the number of brain cells I have working at this time of night.

don1163 wrote in post #18142011
I hope Im not the only one on here who has absolutely no idea what all that means

My two main requests to John would be:

1. Please keep posting - you clearly have knowledge that's interesting to read
2. Please understand that most of us are laymen, and you probably need to think about how you'd explain it to high school kids for it to be understandable!

In other news, my 5D4 has come back... and... the problem with the left sided AF points is exactly the same

Maybe I'm being unrealistic, but running through a DotTune process again gets me ranges that are centred around fairly small values for the centre and right sided AF points (like +2 or +5), but the left sided AF points would need correction values around +13 (maybe higher, as the focus dot is still strong at +20, so the end of the range is likely >20).

As such, if I set a small positive correction value and take a number of shots with the left, centre and right sided AF points (with a fairly wide aperture) the shots using the left sided points are clearly not sharp.

sploo wrote in post #18142098
In other news, my 5D4 has come back... and... the problem with the left sided AF points is exactly the same

Maybe I'm being unrealistic, but running through ....

I've had that happen several times with Canon support but with lenses front/back focusing. I had to send it back several times.

You checked that on a tripod? With mirror lockup? I don''t use any lens adjust stuff, and don't have a problem. When I miss ,I figure I wasn't steady enough. I haven't had the chance to do any real check of the 5D4 AF system yet, but I've not had a problem with any of my previous bodies. I'm just pointing out the obvious stuff here, I'm not trying to sound like a know it all. How bad of a problem are you experiencing?

Time for an exchange from the original place of purchase.

gjl711 wrote in post #18142102
I've had that happen several times with Canon support but with lenses front/back focusing. I had to send it back several times.

I'm a bit frustrated as that repair centre has always done good work for me before, and I was very specific with the issue I was seeing. They claim to have calibrated it, but I've sent an email to indicate there's still a problem.

eric1 wrote in post #18142104
You checked that on a tripod? With mirror lockup? I don''t use any lens adjust stuff, and don't have a problem. When I miss ,I figure I wasn't steady enough. I haven't had the chance to do any real check of the 5D4 AF system yet, but I've not had a problem with any of my previous bodies. I'm just pointing out the obvious stuff here, I'm not trying to sound like a know it all. How bad of a problem are you experiencing?

Yep. Tripod. I've done a bit too much of this sort of testing to be entirely sane. When you have a board slipped down the back of one of the living room sofas with an ISO 12233 test chart you know you have a problem

I like the DotTune method because it eliminates the variability of having to take a bunch of shots using different MFA values (as you usually need several shots at each setting to see a trend). However, four shots (racking the focus way off each time) is more than enough to see that the left sided points have a problem.

At 3.5m camera-to-subject distance, 70mm and f/2.8, it's visibly "off", so that's going to be a problem for portraiture.