Sony A7 - Thick filter stack issues

From vahonen.info
Jump to: navigation, search

Sony A7 - Thick filter stack issues

Lens mount flange to sensor distance makes it possible to physically fit almost any lens to Sony A7 series cameras. Sony A7 cameras are first full frame cameras, which have working live view (Leica M implementations have been horrible), electronical viewfinder and modern sensor. Therefore Sony A7 cameras are well suited to adapt lenses from other cameras as manual focusing is very easy.

People have adapted high quality SLR[1] lenses (e.g. Zeiss C/Y and Leica R) for a long time to cameras. But with Sony A7 adaptation possibilities are much wider. Many lenses created for rangefinder cameras, such as Leica M and Contax G, have very good performance, small size and optical characteristics preferred by many.

However Sony has used very thick filter stack, which decreases image quality of all lenses, unless specially designed for the thick filter stack. Mainly this is shown by field curvature [2][3] and astigmatism [2] [4].

There are two ways to mitigate these issues:

  • Modifying camera so that the filter stack is thinner
  • Using filters in front of lens to "reverse" thick filter stack

I have used both options and combination of the two options.

I'm personally interested about the possibilities to use Zeiss ZM, Leica M and Contax G series rangefinder lenses. Additionally some SLR[1] lenses, which worked quite OK with Canon 5DmkII and 1D-series were not working well with standard Sony A7 - for example Carl Zeiss C/Y mount Distagon T* 2.8/28 does have bad astigmatism, and it's really hard to get good photos unless closing down to f/8-11.

Mitigation option 1: Camera modification

There are few companies, which perform camera modifications. However in practice there is only one option, and due to customs and shipping fees option is only feasible for people living in North America. This company is Kolari Vision. I have modified two Sony A7 (mk1) using this company. Due to shipping and custom fees, and all the extra work with customs, I'm not sure would I like to do it again. The actual work, communication etc. performed by Kolari Vision was top notch - no complaint there, just the cost paid to DHL and customs + negotiations with customs was little too much. Order process from Finland (copy paste from discussion forum): 1. Discussed with Kolarivision (=Ilja) as there are v2 and v3 currenty available. Decided to go with v2 as the color performance of v3 is still slight mystery. As I ordered outside USA and converted 2xA7 the Kolarivision webshop was not having option for this = Ilja sent me PayPal invoice via email and I paid it.
2. Finland local time 5:09PM left package for DHL to deliver on 15th Feb. Paid 161EUR to DHL. Remember to ask them to email proforma invoice, export certificate etc. what you need when cameras get back and you need to proof that you have sent them to USA and you are not trying to avoid custom fees and VAT by ordering camera from USA.
3. DHL delivered and person x at Kolarivision received package 13:37 EST on 16th Feb. Delivery took 27hours and 28minutes (timezones taken into account).
4. Kolarivision did modify the cameras and packed the cameras and certificates to box.
5. Kolarivision shipped package using xxx-shipping company 9:53 EST 24th Feb.
6. xxx-shipping company sent a boat over Atlantic Ocean or somebody did ski over north pole or something else ridiculous as package arrived to Finland 2nd March.
7. xxx-shipping company took 7 days to clear customs (normal for DHL is <1 hour) - today finally at 10:30 they cleared customs.
8. xxx-shipping company delivered package today (9th March) 17:20 Finnish time.
9. xxx-shipping company sent invoice of 241,73EUR, of which ~75EUR is service, and rest is customs and VAT.

If DHL would be used for both ways shipping turnaround time would be easily under 2 weeks.

Modified camera works the same way as standard A7. Only negative effect is that sensor "dust shaker" was disabled. In practice I have not seen more dust on sensor than before - also sensor cleaning can still be done the same way as with standard; using Eclipse fluid.

Some people complain white balance, but generally they have workflow using Auto-WB, and they are not familiar with calibration of cameras. In practice if you have RAW-based workflow and know how to use software white balance should never cause any issue for you.

Mitigation option 2: Filters

On 2016 October discussion started about using lens front filters for mitigation. This discussion can be found from Fred Miranda Sony Forum thread "Front-End Filter Improves Corner Smearing".

Forum recommendations for correcting filter

Stock camera - Zeiss ZM Lenses
Lens Filter needed Comment
Carl Zeiss ZM Distagon T* 4/18 Software calculated:
3000mm Plano-Convex p51 #6

Experimental:
Proxar 0.5m p7 #14 p18 #11
How to remove shims: p12 #2Youtube

"I did not need a rentention ring on the 18mm Distagon, as the step down filters internal diameter had a lip below the 55mm thread which had a smaller inside diameter then 55mm. I instead laid the lens against the front of the 18mm Distagon, placed a strip of nitrile rubber completely around it to hold it in the centre, and used the step down ring to lock it in place against the Proxar lens." p7 #14

Carl Zeiss ZM Biogon T* 2.8/25 Simulation: 2500mm optimal, but others work as well p65 #20, simulation 2000/2500/3000 p67 #6



SLB-50-1500PM p7 #4 p7 #14, Proxar 1m p18 #11 p33#19, also Hoya +1 ok p46 #18

46-55mm -> thin 55mm empty frame -> SLB-50-2500PM -> Elpro retaining ring p74 #3



"46-55 than 55-52 step down than add a 52 retaining ring" for Proxar p38 #3 Photo p39 #2 More p31 #9 Pictures of Guys's setups 46mm Hoya and step-up based p59 #16 More, also other method of shortened Hoya +1 46mm filter p60 #2- and without retaining ring p42 #4

Carl Zeiss ZM Biogon T* 2.8/28 Proxar 1m p7 #4 or SLB-50-1500PM p15 #1 p19 #3 tests p18 #11 or SLB-50-1000PM p15 #8. 1500/2000/2500 simulated p68 #5 Seems 2000/2500 is best for non modified camera.
Carl Zeiss ZM Distagon T* 1.4/35 SLB-50-5000PM p12 #17 p13 #15 p25 #10 p25 #18 pros&cons p27 #19 Test p48 #12 - reversing 5000mm might give better result p56 #8 Shim removal p21 #19, "49-55mm step-up > PCX 5m > 55mm empty filter > 55mm retaining ring" p42 #14 - better version "Start with slim 49, 49-55, blank 55 I used a tiffen UV, optimasigma .5 than ELPro 3 retaining ring. " (slim 49 only needed if using hood) p51 #4, summary of plano convex mounting by Fred p.54 #14
Carl Zeiss ZM C-Planar T* 1.5/50 SLB-50-5000PM? p23 #15, 5000mm doesn't work p55 #11
Stock camera - Contax G Lenses
Lens Filter needed Comment
Contax G Biogon T* 2.8/21 Reversed SLB-50-1500PM p1 #2 "55mm Blank Filter + SLB-50-1500PM(reverse orientation,rubber tape adhered on the edge for centering) + ELPRO4's Retention Ring" p44 #7
Contax G Biogon T* 2.8/28 Reversed SLB-50-1500PM [1] p34 #19 p37 #20 shortened version: "46-55 + 55-52 + SLB-50-1500PM + 52-55" p44 #20
Contax G Planar T* 2/35 SLB-50-6000PM but 5000 also OK p4 #16
Contax G Planar T* 2/45 SLB-50-5000PM p4 #16, "tormoder" tests with 5000mm p54 #12
Stock camera - non-Zeiss Lenses
Lens Filter needed Comment
Leica Summilux-M 35mm f/1.4 ASPH SLB-50-5000PM p41 #4 p47 #14 Which versio (usually same lens has billion version if it's Leica)
Voigtländer Ultron 1.7/35 VM SLB-50-5000PM p49 #15 How to build: p49 #17
Leica Summilux-M 50mm f/1.4 ASPH 10000mm p57 #13
Kolari modified camera (0.8mm sensor stack)
Lens Filter needed Comment
Contax G Biogon T* 2.8/21 From PM discussion with Haruhiko: assuming 1.5mm filter stack 2000mm optimal, but also 2500mm work.

Calculated for 0.8mm sensor stack: SLB-50-3000PM p1 #4, SLB-50-4000PM p3 #14 - Samuli comment after testing: 4000mm is not enough correction with Kolari v2.
"55mm Blank Filter + SLB-50-1500PM(reverse orientation,rubber tape adhered on the edge for centering) + ELPRO4's Retention Ring" p44 #7
Contax G Biogon T* 2.8/28 Calculated for 0.8mm filter stack: SLB-50-4000PM p3 #14 "<shortened>46-55 + 55-52 + SLB-50-1500PM + 52-55" p44 #20

My plan

As I'm shooting with four cameras, of which two are having Kolari conversion. There are lots of choices and testing to be done; which lens works with each camera and filter.

TODO:

  • DONE: Get Hasselblad Proxar 0.5m, 1.0m and 2.0m set was obvious, and also cost was just 50EUR.
    • DONE: Remove lenses from Hasselblad filter frames.
      • In order to remove lens from the Hasselblad filter frame I needed rubber lens tool from eBay (search: camera lens repair tool rubber set ring filter removal) to remove the glass. Usually Hasselblad B57 filter glass retention rings have dents, from which lens spanner wrench can rotate the ring, but my Hasselblad B57 filter didn't have any dents.
      • Rubber tool and Proxar filter in pieces
        RubberLensToolToRemoveLensFromHasselbladFilterFrame.jpg
  • Plano Convex lenses - Most obvious set of Plano Convex lenses seems to be 1500mm, 4000mm and 5000mm - In addition ordered 3000mm to test with lenses like Carl Zeiss Biogon T* 2/35 ZM
    • DONE: Order
    • Wait delivery
    • Assemble "stacks" and test
  • Get needed step-up and step-down adapters
    • DONE: Ordered punch of adapters
    • Wait delivery
  • DONE: Get ELPRO filter

Important measurements

Item Measurement
Proxar's lens element Diameter 51.0mm
46-55 step up Inner hole diameter: 43.35mm
Outer diameter: 57.49mm
55mm part thread inner diameter: 54.26mm
46mm thread height: 2.20mm
55m part height: 4.33mm
Leica Elpro retention ring Inner hole diameter: 49.1mm
Height: YY.YYmm
Contax 55m UV filter frame Inner hole diameter: XX.XXmm
Inner thread diameter: XX.XXmm
Outer diameter: 59.05mm
Inner thread height: YY.YYmm

Carl Zeiss Biogon T* 2.8/25 ZM

25mm Biogon has so steep ray angle to sensor that it does not work even with thin filter stack modified cameras. Therefore only possible way to use the lens is using front filter method.

Filter stack

Option 1 - will not work, it causes vignetting
1. ZM25 lens
2. 46mm - 55mm step up ring
3. Empty (=glass removed) filter - I used Contax 55mm UV filter
4. Proxar 1.0m lens
5. Leica Elpro 3 retention ring

Option 2 - does not vignette, but additional filters can't be added
1. ZM25 lens
2. 46mm - 55mm step up ring
3. Proxar 1.0m lens
4. 55-52mm step down ring
5. 52mm UV filter retention ring


Option 1 - Top view

CarlZeissZM25 ProxarTopView.jpg

Option 1 - Side view

CarlZeissZM25 ProxarSideView.jpg

Option 2 - Side view

CarlZeissZM25 ProxarSideViewOption2.jpg

Option 2 - Side view

CarlZeissZM25 ProxarTopViewOption2.jpg

Shim removal

Image #1

CarlZeissZM25 ShimRemovalRearRings.jpg

Image #2

CarlZeissZM25 ShimRemoval.jpg

  1. Remove ring from rear of the lens indicated by blue arrow in image #1
  2. Lift the rear outer part of the lens (image #2 top left corner)
  3. Remove shims - top of the image #2, make sure you rotate the "holes" in rings so that they can go over the skrew
  4. Reverse the process steps 1 & 2 to assemble the lens


My shims were 0.71mm total thickness, and them removed lens focuses little over infinity with 1.0m Proxar. With the shims it did focus ~1.5m distance or less.

On forum some people needed additional shims removed. See Guy's posts p59 #2 and p59 #6. Personally I would not touch these as they are having effect to how rear lens group positions compared to front lens group => this is not just focus adjustment. This is done by removing rear protection ring (image #1 red arrow), and then removing the shims. According to Guy it should come off by rotating it by hand - mine doesn't rotate by hand, something stronger would be needed to open it.

Vignetting

Option 1

Option 1 filter stack there is almost no mechanical vignetting. However it seems that with this setup it's impossible to use any filters. Below are four pictures shoot at f/8, shooting 0.5m focus and infinity focus.

First two are with filter stack explained above in chapter "Filter stack":

Option 1 @ Infinity

CarlZeissZM25 Vignetting Infinity.jpg

Option 1 @ 0.5m

CarlZeissZM25 Vignetting 0.5m.jpg

Next two are with additional 55mm - 82mm step up ring, which is needed if I want to use polarizer- or ND-filter with the lens. As can be seen there is mechanical vignettin already at 0.5m focus distance, and at infinity it gets really bad:

Option 1 + 55-82mm @ Infinity

CarlZeissZM25 Vignetting Infinity 55-82.jpg

Option 1 + 55-82mm @ 0.5m

CarlZeissZM25 Vignetting 0.5m 55-82.jpg

Option 2

Option 2 filter stack has almost no hard vignetting by itself, but by adding anything to 52mm filter thread hard vignetting appears.

Option 2 @ Infinity

CarlZeissZM25 Option2 Vignetting Infinity.jpg

Option 2 + 52-82mm step-up @ Infinity

CarlZeissZM25 Option2 Vignetting Infinitry 52-82.jpg

Distortion

There is some pincushion distortion in images. I don't shoot architecture, so I'm not best person to answer how easily that can be fixed. In my nature photos this small distortion is 100% irrelevant.

Image quality

Like seen in many examples in forum the infinity performance is very good with this solution.


However boke issues are NOT fully fixed by Proxar f=1m filter. The corners get weirdly in focus and there are remains of astigmatism. This can be caused either by Proxar just not being good for for this purpose, or we should not use edges of proxar filter - or the mechanical vignetting is causing the boke issues (not astigmatism). One possibility is that close focus distances require different correction - I asked this from HaruhikoT and he said simulation should be valid for all distances (on lenses not having floating elements).

This should not be understood wrong; lens is useful and better than many other lenses - boke just does not reach same quality which can be seen from slides shoot with this lens. I'll put here detailed examples IF the final conclusion is that front filter can only help @ infinity. Here are some random shots linked to show the remaining issues on boke :

Conclusion and next steps

Option 1: Image quality @ infinity is OK with Proxar f=1m filter. One of the best lenses I have seen at this focal length @ f/5.6. However as I use filters (polarizer or ND) when shooting landscapes this solution is not going to work very well. Boke has issues and need to test if other front filter solutions or front filter + Kolari combination provides solution.

Option 2: Image quality @infinity TO BE TESTED WHEN WEATHER/SCHEDULES ALLOW. Compared to option 1 boke did not change (or if it did, then not much). So basically only difference compared to option 1 is that there is no vignetting.

TODO:

  • DONE: Try with 46-55mm -> 55-52mm combo (a.k.a. "Option 2")
    • Test how it performs infinity focused
  • DONE: Order Hoya +1 HMC lens - ordered 46mm, 58mm, and 67mm
    • Remove +1 lens from 46mm and install to shorter 46mm filter frame and test
    • Test step up + 58mm/67mm options, if they work put the +1 to other frame which has front thead
  • Experiment with OptoSigma Plano Convex lenses (1500mm reversed? or 3000mm with A7 Kolari modified cameras)

Carl Zeiss Biogon T* 2/35 ZM

35mm Biogon has so steep ray angle to sensor that it does not work even with thin filter stack modified cameras. Therefore only possible way to use the lens is using front filter method. The group of people who only test infinity and shoot landscapes may claim it's perfect on Kolari modified thin filter camera. However in real life shooting where boke is usually involved, this lens really doesn't work at all, not even on Kolari modified camera. Here is example 1280px wide and same in really big size.

Filter stack

HaruhikoT calculated this for me and PCX 3000mm lens should be OK for standard thick sensor stack camera. Luckily I ordered one 3000mm :)

  1. ZM2/35 lens
  2. 43mm - 55mm step up ring
  3. Empty (=glass removed) filter
  4. OptoSigma SLB-50-3000PM
  5. Leica Elpro 3 retention ring OR 55-52 step down + 52mm retention ring
Top view

250px

Side view

250px

Shim adjustment

1. Remove C the rear part of the lens (the part which is against table surface if one puts lens without rear cap to desk).
2. After C is removed you can open the lens by rotating the part circulating part C with lens spanner (this cannot be done before removing part C).
3. Lift part B, the middle part of the lens from A, front part of the lens.
4. Remove/add shims to location indicated by green arrow.
5. Reverse steps 4,3,2 and 1 to assemble the lens.

Under any circumstance DO NOT TOUCH the shims indicated by red arrow.

CarlZeissZM2 35 LensExplodedShims.jpg

Memo of shimming process:

  • Adapter used was "Fotka" brand
  • No filter: Infinity hard stop or lens doesn't go to infinity, depends on adapter and camera body.
  • SLB-50-3000PM: Infinity with all original shims = ~4m
  • SLB-50-3000PM: Infinity with 1 thick shim removed = ~7m
  • SLB-50-3000PM: with 2 thick + 1 thin shims removed lens goes very little over infinity (room temperature)

Vignetting

Should not be issue with 2/35 - but needs to be tested.

Distortion

Should not be issue with 2/35 - but needs to be tested.

Image quality

Infinity test...

Boke test...

Conclusion and next steps

xxx


Carl Zeiss Distagon T* 4/18 ZM

18mm Biogon has so steep ray angle to sensor that it does not work even with thin filter stack modified cameras. Therefore only possible way to use the lens is using front filter method. The group of people who only test infinity and shoot landscapes may claim it works.

Filter stack

HaruhikoT calculated p51 #6 and PCX 3000mm lens should be OK for standard thick filter stack camera. As front lens element comes out from the surface around the lens, method from p7 #14 can't be used (Proxar is concave from the side which should be facing the lens and will not hit front lens). It is possible that there is no easy and good way to attach plano convex lens to ZM18.

  1. ZM4/18 lens
  2. xxx
  3. xxx
  4. xxx
  5. xxx
Top view

250px

Side view

250px

Shim removal

Refer to video on forum.

My lens had 5 shims, of which total thickness is 0.72mm. 2 of them have "bronze" color and 3 "gold".

Vignetting

xxx

Edge/corner color issues

xxx

Distortion

xxx

Image quality

Infinity test...

Boke test...

Conclusion and next steps

xxx

Carl Zeiss Biogon T* 2.8/21 G

xxx

Filter stack

xxx

Top view

250px

Side view

250px

Shim removal

xxx

Vignetting

xxx

Edge/corner color issues

xxx

Distortion

xxx

Image quality

Infinity test...

Boke test...

Conclusion and next steps

xxx

Carl Zeiss Biogon T* 2.8/28 G

xxx

Filter stack

xxx

Top view

250px

Side view

250px

Shim removal

xxx

Vignetting

xxx

Edge/corner color issues

xxx

Distortion

xxx

Image quality

Infinity test...

Boke test...

Conclusion and next steps

xxx

Links to useful resources

Notes

  1. 1.0 1.1 Cameras having large moving mirro between lens mount and film/sensor => large lens mount flange distance => generally ray angles are "gentle" to digital sensors and their thick filter stacks. https://en.wikipedia.org/wiki/Single-lens_reflex_camera
  2. 2.0 2.1 http://toothwalker.org/optics/astigmatism.html
  3. https://en.wikipedia.org/wiki/Petzval_field_curvature
  4. https://en.wikipedia.org/wiki/Astigmatism