Canon TS-E 17 - Borders are out of focus on Sony A7r III

Shirozina wrote:

Keith Cooper wrote:

Shirozina wrote:

These lenses are quite old now and maybe were never designed to work with such high MP sensors.

Well, it came out at the same time as the 1Ds3 - quite good enough to show problems.

There were issues if you're talking about the original 3 TS-E lenses, but less so the TS-E24 ii and 17mm

The new ones both concentrated on expanding the size of the image circle (as well as the changes in the mount movements)

Canon may not have done a full test and if they did the design spec may not be for a perfectly flat focus field at infinity.

I'm minded to doubt that - the differences in requirements for a 1Ds3 sensor and relatively modern ones are not that distant.

Now, with the R5 and potential higher res versions I can see that a potential RF t/s lens at 14mm will indeed require some nifty optics.

Few lenses are flat at infinity anyway nor are their focus fields.

Indeed, and (even when first launched) there were reports of field curvature causing problems. Given my own testing of the 17mm, I'm inclined to think that this is something that varies between examples. I don't have test evidence for this but it certainly chimes with some of the observations in the excellent Roger C articles here Roger Cicala: the difference between sample variation and a 'bad copy' (Part 2)

I’d say 21mp is not at all in the same league as the current high mp sensors these issues are being revealed on.

I'd say the difference in pixel pitch is not actually that much in terms of sensor resolution. Lack of flatness at this level would be readily apparent.

There is more than just resolution when you consider the optical parameters of the design of Sony's sensors. The TS-E 17 works quite well on the 45MP (DPAF) pixels of the R5.

That suggests that it's likely to be as much the design of the pixels (lenses/cover etc) as the resolution (as mentioned elsewhere in this thread)

Looking at using the lens with a 1Ds3, 5Ds and R5 I'm far more inclined to beliee it's connected with aspects of sensor design, not just megapixels.

In talking recently with a lens company in China about very similar design issues, they had noticed distinct issues with some short FL retrofocus design lenses which were likely sensor design based.

Keith Cooper wrote:

Excellently. I tested it with the plain adapter and the EF->RF adapter with a polariser in it

http://www.northlight-images.co.uk/canon-filter-adapter-with-tilt-shift/

With the plain adapter it was pretty much identical to using it natively on my 5Ds

I've no suggestions as to the Sony related problem I'm afraid.

Hi Keith, thanks for joining the treat.

Yes, I've already read your review, as some of your other very good articles and video.

Cheers,

Jerome

Hi Sittatunga,

That's interesting! I've read a few things on Ken's blog, but I definitely missed this one.

For sure that may explain some things. Thanks for sharing it,

Jerome

Shirozina wrote:

Jerome

These lenses are quite old now and maybe were never designed to work with such high MP sensors. Canon may not have done a full test and if they did the design spec may not be for a perfectly flat focus field at infinity. Few lenses are flat at infinity anyway nor are their focus fields.

On the other hand a lot of architecture photographers I follow shoot on the 5DsR, which is 50 MP and these ts-e lenses are crazy sharp.

Shirozina wrote:

Jerome-140 wrote:

Hi,

I get a friend how owns a Canon 5DM IV. Here is the shot, center focus as the other.

Now I can say that the problem doesn't come from the lens.

But one question stay? How will it perform on an EOS R5 WITH Canon adapter ???

The 5d IV doesn’t have the same resolution as your Sony so how can you be so sure? Also you can’t rule out your adapters unless you have measured them.

31 vs 42 MP is not that much. Compares the shots and look at consistency. The point of my comparison is not how sharp le lens is on a different body in absolute - and it is sharp - but to have center and border with a similar one.

Keith Cooper wrote:

Few lenses are flat at infinity anyway nor are their focus fields.

Indeed, and (even when first launched) there were reports of field curvature causing problems. Given my own testing of the 17mm, I'm inclined to think that this is something that varies between examples. I don't have test evidence for this but it certainly chimes with some of the observations in the excellent Roger C articles here Roger Cicala: the difference between sample variation and a 'bad copy' (Part 2)

Keith, do you also feel with that lens that is best to focus as far as possible, as opposed to look for the hyperfocal point, which will be way closer in the foreground (0.9 to 1.3m depending on the aperture) ?

Jerome-140 wrote:

Keith Cooper wrote:

Few lenses are flat at infinity anyway nor are their focus fields.

Indeed, and (even when first launched) there were reports of field curvature causing problems. Given my own testing of the 17mm, I'm inclined to think that this is something that varies between examples. I don't have test evidence for this but it certainly chimes with some of the observations in the excellent Roger C articles here Roger Cicala: the difference between sample variation and a 'bad copy' (Part 2)

Keith, do you also feel with that lens that is best to focus as far as possible, as opposed to look for the hyperfocal point, which will be way closer in the foreground (0.9 to 1.3m depending on the aperture) ?

My own longstanding heresy in this respect

http://www.northlight-images.co.uk/when-hyperfocal-focusing-is-no-good/

For a 17mm lens like this at the sorts of apertures used, I focus on the main subject and let DOF take care of itself. Usually that's something relatively far away.

A bit of tilt may help, but finding scenes where near and far objects of attention line up neatly on a plane doesn't tend to occur very often with my general choices of subjects/composition. If I do use tilt on the 17, it's far more likely to be to run the plane of focus along a wall/floor/ceiling

My own longstanding heresy in this respect

http://www.northlight-images.co.uk/when-hyperfocal-focusing-is-no-good/

For a 17mm lens like this at the sorts of apertures used, I focus on the main subject and let DOF take care of itself. Usually that's something relatively far away.

A bit of tilt may help, but finding scenes where near and far objects of attention line up neatly on a plane doesn't tend to occur very often with my general choices of subjects/composition. If I do use tilt on the 17, it's far more likely to be to run the plane of focus along a wall/floor/ceiling

Wow great article, thanks

mmarian wrote:

Shirozina wrote:

Used my 17mm + A7rIV on a job yesterday where the 24 was not wide enough and the images at infinity are razor sharp right to the edges at F8.

Maybe not directly related to this topic but the TS-E lenses are quite difficult to focus accurately through VF. I find that the most reliable and accurate way to focus is using LV and 10x magnification on the LCD screen. I understand that this might not be always practicle. I shoot architecture and 99.9% with camera on tripod. These lenses' focusing ring goes past infinity and it is easy to make a mistake and set the focus there. I am not saying that this was responsible for the issue debated in this thread but one needs to be mindful of this when shooting with TS-E lenses.

The ability to set the focus ring beyond the ∞ setting is important with all the TS-E lenses for use with tilt - that doesn't though stop it from being a minor irritation when you accidentally move the focus beyond ∞

With all modern TS-E lenses I find he reduced focus throw of the lenses to be a hindrance to focusing at modest distances.

The new 50/90/135, being 'macro' lenses all have quite a bit of the focus throw devoted to closer settings, making the range of throw from 1m-∞ unnecessarily short.

As an aside, I found the 17mm a delight to use on the R5 where IBIS, the viewfinder level and focus peaking made using it hand-held very easy(heresy I know, but I don't care  )

Maybe not directly related to this topic but the TS-E lenses are quite difficult to focus accurately through VF. I find that the most reliable and accurate way to focus is using LV and 10x magnification on the LCD screen. I understand that this might not be always practicle. I shoot architecture and 99.9% with camera on tripod. These lenses' focusing ring goes past infinity and it is easy to make a mistake and set the focus there. I am not saying that this was responsible for the issue debated in this thread but one needs to be mindful of this when shooting with TS-E lenses.

True, but that's pointless with a mirrorless camera thanks to the EVF

Keith Cooper wrote:

The ability to set the focus ring beyond the ∞ setting is important with all the TS-E lenses for use with tilt - that doesn't though stop it from being a minor irritation when you accidentally move the focus beyond ∞

With all modern TS-E lenses I find he reduced focus throw of the lenses to be a hindrance to focusing at modest distances.

The new 50/90/135, being 'macro' lenses all have quite a bit of the focus throw devoted to closer settings, making the range of throw from 1m-∞ unnecessarily short.

As an aside, I found the 17mm a delight to use on the R5 where IBIS, the viewfinder level and focus peaking made using it hand-held very easy (heresy I know, but I don't care )

Hi Keith,

Right on point as always...

R5 on order, can't wait to check this out by myself!

Cheers,

Jerome

OK, the final step for this post, and final test: My TS-E 17 mounted on the new Canon EOS R5 (with canon adapter) and compared with the Canon RF 15-35 (at 15mm), and the Sony A7RIII with the Sony 16-35 GM (at 16mm).

1. TSE 17

2. RF 15-35 on Canon EOS R5

3. Sony 16-35 GM on Sony A7RIII

All these images are slightly edited from RAW in Lightroom. (Highlight - shadows - contrast) and all are sync with the same settings and white balance. No sharpening added!

My final conclusion:

Adapted lenses on sony - I speak for wide-angle lenses - are not great, and worst on the Metabones V than the Sigma MC-11 (unfortunately because electronics works better on the Metabones).

EF lenses on RF mount with a Canon adapter works great!

EOS R5 is great. Maybe the Sony A7rIII is still very very slightly sharper and highlights recovery a bit better. But, it seems to me that the shadows recovery on the Canon beats the Sony.

What are your thoughts?

Jerome-140 wrote:

EOS R5 is great. Maybe the Sony A7rIII is still very very slightly sharper and highlights recovery a bit better. But, it seems to me that the shadows recovery on the Canon beats the Sony.

What are your thoughts?

Did you compare the Sony files with the standard R5 files or the Dual Pixel RAW files?  The Dual Pixel RAW files give you an additional full stop of highlight recovery.