A7R+TS-E 24: Internal reflections in Metabones III adapter and what could help

[marc aurel]

Hi everybody.
I'm a long time reader of this forum but didn't contribute yet. I want to share an observation I made with the metabones III adapter:
I made a comparison with TS-E24 II on A7R and 5DIII, using shift. In the centre of the unshifted lens, the A7R has more resolution. Most obvious in the meander of the tympanon and the grid in the window. In the extreme corners of the shifted image resolution of both cameras is very close.
But additionally there is a degradgation of contrast in the shifted area in the A7R image compared to the 5DIII - the A7r image looks worse which I didn't expect - the black isn't really black. I suspected internal reflections in the adapter - something that has been reported about the TS-Es in another forum too. So I bought some self-adhesive velour to cover the interior of the metabones - Canon uses something similar for the interior of the TS-Es. I repeated the test. It was difficult to find a day with similar weather (no direkt sun), so the test is not perfect (light was a bit more contrasty on the second test).
I used tripod. Focused on extremes corner with live view. Raws converted in Lightroom 5.3. Adjusted white balance a bit to make images look similar. Sharpening for all: Amount: 50 - Radius: 0,7 - Detail: 30 - Mask: 4. Test 1 is before, Test 2 after changing the interior of the adapter. 100% crops (I tried up- und downscaling one or both to match sizes; results were similar).

In my eyes one can clearly see that in Test 1 the A7R image of the shifted area is less contrasty. In Test 2 - with the metabones with velour - it isn't. A7R still doesn't show a resolution advantage in the extreme corners - maybe the lens doesn't give enough resolution there. But it's not worse than the canon any more. And in the centre and most other parts of the image the A7R is much better.
I suspect that TS-E causes more reflections than other lenses beacuse the alrger image circle hits more surfaces of the adapter interior.

marc aurel

[rainer_v]

thanks for this test. i will do the same with my adapter ...

[marc aurel]

Hi rainer,

it was a bit tricky to do - the geometry of the adapter is not very simple. Below is how it looks.
I made a cutting pattern which consists of several perts (see below). It's not absolutely precise, but it worked well. The jpg with the cutting pattern is scaled 1:1, so you could use it to try. Your own risk  . The self-adhesive velour i used could be removed without problems.
At first I tried to cover every surface. Then I noticed that at a shift of more than 10mm the velours interfered with the optical beam path (the opening of the adapter is just about 36x24mm). When shifted up, the lowest part of the image bottom went darker. When shifting down, it went darker on top.
So I took off the velour on one surface (marked green in the cutting pattern), but only the one that interfered with the shift up (cause i don't use more than 10mm of shift down). The problem was gone and it didn't visibly change the image quality. Either this was not the problematic surface, or it is sufficient to treat most of the surfaces to prevent a ping-pong-effect of reflections. Image 2 shows how that looks. You can also see that - despite of the matte black finishing - the metal does reflect.
Maybe it's not to everyones taste do such things - but now i get the best image quality available for this sensor size when shift is needed. 

marc aurel

[derAngler]

Thank you for posting the pictures. Did you think about using matt black varnish, or why did you choose the velour?

[marc aurel]

Thank you for posting the pictures. Did you think about using matt black varnish, or why did you choose the velour?

I think it wouldn't work as well. All even surfaces do reflect. Look inside a canon lens - they do the same.
marc aurel

[marc aurel]

I tried to visualize what happen's with that combination. Section unshifted and with full shift (see below).
With its image circle of 67mm the TS-E hits a lot of surfaces inside the metabones that can reflect. Maybe the rear element of the lens reflects back to the sensor.

[adammork]

Hi Marc

Thank you so much for the effort - and for sharing this!!

Best from Copenhagen

[Paul2660]

Interesting post, and thanks for the info.

I would curious if shooting an LCC for the shifts at least on the Sony might help fix this?  Capture One would be an easy way to test it. 

Metabones makes a point of how their adapters are coated matte black inside to prevent this type of problem.  But your results seem to show that the velour makes a difference for sure.  Now I know why Canon has always used it. 

Paul Caldwell

[uaiomex]

+1

Greetings from Tulum

Eduardo

Hi Marc

Thank you so much for the effort - and for sharing this!!

Best from Copenhagen

[marc aurel]

Interesting post, and thanks for the info.

I would curious if shooting an LCC for the shifts at least on the Sony might help fix this?  Capture One would be an easy way to test it. 

Metabones makes a point of how their adapters are coated matte black inside to prevent this type of problem.  But your results seem to show that the velour makes a difference for sure.  Now I know why Canon has always used it. 

Paul Caldwell

thanks everybody for the positive reactions. I'm curious to see if others experience an improvement too.
Would be interesting to see if there is a difference in the corners of non-shift-lenses too (I rarely use them so I didn't try).

Paul - I guess LCC wouldn't help on this problem because I see no color shift or vignetting in the images with the unmodified adapter. It's rather scattered light that increases overall brightness and reduces contrast this way. But I'm not familiar with LCC myself so maybe I'm wrong. Let me know what you find out.

with best regards -
marc

[tho_mas]

Marc -
Extremely valuable post - thanks for sharing!

[Paulowen]

Another big "thank you" for taking the time/trouble/effort for this 

[tn1krr]

Brilliant post/work and thank you so much for taking time to document it so well for the rest of us. I'm usually gathering a lot of useless stuff, but for this once it works for my benefit. I bought a pile of adhesive light trap material to kill reflections on my home theater and still have a lot left. Just in case anyone is interested, this material is usually used to coat insides of star telescopes so it kills reflections extremely well.
 
http://www.protostar.biz/flock.htm

[MrSmith]

What are the vertical lines in the archway shadow on both canon and sony image? I'm looking on an ipad but I don't think it's to do with the retina screen.

[marc aurel]

What are the vertical lines in the archway shadow on both canon and sony image? I'm looking on an ipad but I don't think it's to do with the retina screen.

You're right - it's not an artefact. This is a wire mesh for protection against birds nesting up there in the towers. For me it's a proof of the capability of that lens.

I did a rough calculation of it's dimensions:
The church front is about 20m wide, the whole image about 30m (maybe a bit more). So on the A7R we have 7360 pixels to resolve 30000mm. 1 Pixel represents about 4mm. The grid width of the mesh is 5-6 pixels in the A7R image which corresponds to 20-25mm. Looking at it on spot (altough from far below) I estimate the diameter of the wire used for the mesh with not more than 1mm. This would only be 1/4 of a pixel. I added a screenshot of the A7R-image from under the clock at 1000% enlargement in photoshop.
Nice to see that this lens is able to show at least a trace of this structure - in that part of the image circle!
The fact that you only see the vertical wires and not the horizontal ones hints towards a difference between saggital and meridional resolution.

For comparison and more information about how the wire mesh looks: I added a photo taken with contax-zeiss pc-distagon 35mm shift-lens from closer distance. By the way: in my opinion the best 35mm shift lens for this sensor size - and it is great on the A7R. Didn't have the time to test it on the same subject and A7R yet).

marc

p.s.: in the lower part of the church there's a different mesh with much thicker wire - so you can't compare resolution in centre and corners by this - which would be quiet a nice and large test target for real life photography (the focus distance when people test wideangles with shift using the usual test charts is a bit short)

[chiek]

Very interesting Posts. Thanks for sharing.

It's very helpful for my New design camera.

Happy new year !!!

http://www.chiek.co.kr/cam

[David Campbell]

I took Marc's sketch (he did all the hard work) and drafted it up in a CAD program and made a PDF template.
It is not 100% perfect and you will need to trim in some places but if it helps more people get around the reflection problem, then it is of some use.

Cheers
Dave

[tho_mas]

For comparison and more information about how the wire mesh looks: I added a photo taken with contax-zeiss pc-distagon 35mm shift-lens from closer distance. By the way: in my opinion the best 35mm shift lens for this sensor size - and it is great on the A7R.

Due to your post and sample I've bought a Contax PC-Distagon. In fact a great lens on the A7R! Really sharp at the edges of the image circle (respectively at max shift of 10mm on the long side of the sensor) at f8. Many thanks for the advice!

[marc aurel]

Due to your post and sample I've bought a Contax PC-Distagon. In fact a great lens on the A7R! Really sharp at the edges of the image circle (respectively at max shift of 10mm on the long side of the sensor) at f8. Many thanks for the advice!

Good to hear. I really love this lens.
For the correction of distortion there is a profile provided by the free Alpa Plugin for Photoshop. You can simple enter how much shift, rise or fall you used and get a correccted result. Very comfortable.
Marc

[tho_mas]

For the correction of distortion there is a profile provided by the free Alpa Plugin for Photoshop. You can simple enter how much shift, rise or fall you used and get a correccted result. Very comfortable.

I've seen you mentioning Alpa LC support for the PC-Distagon in the MFD forum. Very valuable info - many thanks!
First I was a bit confused how to set the shifts when the lens is rotated (say 10mm shift at 45° rotation). On my tech cam the back is shifting so when I use 10mm rise/10mm lateral I can simply enter the same shift values in the Alpa LC plugin. With the rotating lens you have to re-calculate the shifts (10mm rise at 45° roatition = 7mm rise/7mm lateral). Anyway... the Alpa LC profile works just fine.
Thanks again!