1:20 to 1:5 (0.05x to 0.2x)

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

jvanhuys
Posts: 49
Joined: Sun Jan 22, 2023 9:24 pm
Location: Seoul, South Korea

1:20 to 1:5 (0.05x to 0.2x)

Post by jvanhuys »

Hi everyone.

I trust you've had a good week.
When time permits, I'm desperately searching for a lens tuned/ corrected specifically for 1:20 (0.05x) to 1:5 (0.2x), with my preference leaning more towards the 1:20 side and more towards north of 60 mm focal length.

I need it to document flat artworks, ranging from about 0.5x0.5m to 1x1meters in size, with the camera being about 3 or 4 meters behind the artwork. I'm currently doing stitches ranging from about 350MP to 600MP on a monorail 4x5 with Sony A7Rm4 at the back. It's important for me that the lens is flat-field as the opposite affects my stitches badly (see Nikkor 55/2.8 below).

My art is mostly black and white, with tiny accents of color here and there. The result is that my artwork shows severe amounts of chromatic aberration when photographed, especially in the corners. It seems finding a lens that is flat-field and a true apochromat at 1:20 is an almost impossible task.

Below are the lenses I own, that are not up to the task:
  • Nikon 55/2.8 micro-Nikkor AIS: slightly too much CA, not truly flat-field at 1:20, image circle prevents any movements
  • Nikon 105/2.8 micro-Nikkor AIS: far too much CA, image circle prevents any movements
  • Schneider Symmar-S 180mm f/5.6: slightly too much CA
  • Makro-Symmar HM 120 f/5.6: far too much CA as out of mag range
  • Schneider Componon-S 80mm f/4: slightly too much CA and less sharp than Nikon 105/2.8 micro-Nikkor
Lenses I'm considering:
  • Schneider SAPPHIRE 0.007/0.07x V70 16K line scan lens: mag range and MTF curve seems near-perfect
  • Zeiss Orthoplanar 60/ 105mm: mag range and sharpness sounds great on paper, but so little info online
  • Makro Symmar 80 5.6 -0.033x version: mag range sounds promising, but almost impossible to find a copy, even from Schneider themselves
  • Rayfact 63mm f/2.8 N: optimized for 1:20 to 1:1 (perfect) but impossible to find and from lens design papers it seems it's basically an EL 63 Nikkor 2.8 N with a Rayfact price tag... and I'm not convinced that lens is free of CA, judging from sample images on Flickr
  • Super Symmar HM 120: seems promising, but infinity corrected, and judging by Makro Symmar HM, it's not the sharpest, even though its CA control is amazing
I'd appreciate any feedback, pointers or lens sales you guys can provide me with.
Johan.

chris_ma
Posts: 570
Joined: Fri Mar 22, 2019 2:23 pm
Location: Germany

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by chris_ma »

Hi Johan,

how large of an image circle do you need (350 to 600MP on a sony A7r4 soundly like about 130mm IC if you only shift the camera back and leave the lens fixed).
that would mean you'd probably have to look at 4x5" lenses, and it will be hard to find something which matches the best full format lenses.

another option would be to move the whole camera with lens in parallel to the artwork, in this case you could any full format lens with 42mm image circle.
for this case, the Laowa 100mm macro would be my first choice.
chris

jvanhuys
Posts: 49
Joined: Sun Jan 22, 2023 9:24 pm
Location: Seoul, South Korea

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by jvanhuys »

chris_ma wrote:
Tue Jan 24, 2023 3:29 am
Hi Johan,

how large of an image circle do you need (350 to 600MP on a sony A7r4 soundly like about 130mm IC if you only shift the camera back and leave the lens fixed).
that would mean you'd probably have to look at 4x5" lenses, and it will be hard to find something which matches the best full format lenses.

another option would be to move the whole camera with lens in parallel to the artwork, in this case you could any full format lens with 42mm image circle.
for this case, the Laowa 100mm macro would be my first choice.
Hi Chris!

Thanks for pitching in, I appreciate it. The Loawa is certainly a very attractive lens, but with a caveat listed below. If it helps, my entire tripod is also mounted on a camera dolly system, so I can move everything forward or backward.

You're exactly right. I have more freedom to stitch on my 4x5 lenses for sure, but I'm not necessarily a big fan of large IC. If the artwork is really large, then I can rise the back standard in order to avoid seeing the front of my rail. The Loawa's image circle wouldn't allow any movements at all, whereas something like a Componon-S 80 f/4 does leave some room for movements on smaller artworks. So ideally an IC of 60-120mm is preferred. What I've noticed with my lenses listed above, is that the larger the image circle, especially when going north of 300mm everything from centre of frame to corners start going very soft (exception being my Symmar-S 300mm, which is very usable, but won't win awards. My Nikkor Apo 360 f/9 is rubbish in terms of sharpness.

The exception to this larger IC = softer image in my limited exeperience is my Symmar-S 180mm f/5.6 (single-coated, 1970s model, but it is flocked), which is according to my tests, is sharper than my Componon-S 80mm f/4 (maybe I got a dud?). The Makro Symmar HM 120 is unusable at this magnification, roughly 1:20, so I'm curious if Super Symmar HM 120 could be the sweet spot. Like most LF taking lenses, it's corrected for infinity, so it makes me doubt whether it will be as free from CA or as sharp as Schneider's Sapphire 16k line-scan lens, which offers a model that caters specifically for (-0.17 ... -0.03) magnification range.

Thanks for suggesting the Loawa. How's your experience with it? Is it truly flat field between 1:20 and 1:5 magnifications?

chris_ma
Posts: 570
Joined: Fri Mar 22, 2019 2:23 pm
Location: Germany

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by chris_ma »

hi Johan,

I don't have much first hand experience with large format lenses, but from what I've read you'd need to get one of the latest greatest expensive ones to get anywhere near to what a 10 times cheaper FF lens gives you.

the laowa has been discussed here many times and considered excellent, mostly at 1x and 2x, but I'd expect it performs great at 0.05 to 0.2x too. it's so cheap that it's easy to buy and test it and resell if it doesn't work for your use case.

in the 60mm IC area there are some interesting possibilities in the machine vision specialized area, but it's not going to be cheap either, and ideally you'd want different lenses for different magnification.
if you have the funds, the most suitable one is probably the inspec.x L 5.6/120 float since it has a correction ring for changing magnification and also covers a large 82mm image circle:
https://www.excelitas.com/product/inspe ... -and-56120
chris

chris_ma
Posts: 570
Joined: Fri Mar 22, 2019 2:23 pm
Location: Germany

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by chris_ma »

ps: if you haven't done so, it's worth looking at robert's website since he has many detailed tests.
he recently wrote a lot about the HR-M line which covers the range you're interested in:
https://www.closeuphotography.com/blog/ ... by-qioptiq
https://www.closeuphotography.com/blog/ ... -announced
https://www.closeuphotography.com/blog/ ... m-mtf-data

the test is not quite finished I think but he posted some of his findings here on the forum:
http://www.photomacrography.net/forum/v ... 7d50665bce
chris

chris_ma
Posts: 570
Joined: Fri Mar 22, 2019 2:23 pm
Location: Germany

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by chris_ma »

... and if you need even higher quality and don't mind going broke, the HR Digaron SW f6.5 138mm FLOAT might also be a great option (about 14000USD though):
https://www.linhofstudio.com/catalogue/ ... _a4152.pdf
chris

rjlittlefield
Site Admin
Posts: 23564
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by rjlittlefield »

Johan, I realize that you asked specifically about lenses, but I'm more of a systems guy so I'm thinking about the bigger picture of how to create 350MP to 600MP images of flat art.

As background, I was working in the bowels of panorama stitching software twenty years ago when the first gigapixel panos were being made. At that time we routinely corrected for lens distortion (barrel/pincushion/gullwing) inside the stitching software. CA correction was commonly done by other programs. These days focus stacking is also common (and I'm the fellow who wrote Zerene Stacker).

So, from my standpoint the critical aspect of your problem is to get a lens that is uniformly sharp across your camera's sensor, letting geometry, CA, and field flatness be addressed in software.

I assume that I have overlooked something which makes this approach not work for your application.

But that makes me curious: what is it that I've overlooked?

--Rik

chris_ma
Posts: 570
Joined: Fri Mar 22, 2019 2:23 pm
Location: Germany

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by chris_ma »

rjlittlefield wrote:
Tue Jan 24, 2023 4:03 pm
So, from my standpoint the critical aspect of your problem is to get a lens that is uniformly sharp across your camera's sensor, letting geometry, CA, and field flatness be addressed in software.
I understood the question about a lens with "flat-field" as in no field curvature in focus, ie the uniformity of sharpness across the sensor.
If the question was about light falloff I agree it's not much of a concern with modern software (same with distortion. never made test with CA to see how much difference it makes since I use APO lenses).

last month I made a panorama of 7x10 images for a 7.5 Gigapixel image (45GB at 16bit) using PTGui and I was surprised how it worked perfectly on the first attempt without manual intervention.
chris

jvanhuys
Posts: 49
Joined: Sun Jan 22, 2023 9:24 pm
Location: Seoul, South Korea

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by jvanhuys »

chris_ma wrote:
Tue Jan 24, 2023 1:48 pm
ps: if you haven't done so, it's worth looking at robert's website since he has many detailed tests.
he recently wrote a lot about the HR-M line which covers the range you're interested in:
https://www.closeuphotography.com/blog/ ... by-qioptiq
https://www.closeuphotography.com/blog/ ... -announced
https://www.closeuphotography.com/blog/ ... m-mtf-data

the test is not quite finished I think but he posted some of his findings here on the forum:
http://www.photomacrography.net/forum/v ... 7d50665bce
Fantastic information all round. Really appreciated. Let me go study all the links first before responding.

jvanhuys
Posts: 49
Joined: Sun Jan 22, 2023 9:24 pm
Location: Seoul, South Korea

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by jvanhuys »

rjlittlefield wrote:
Tue Jan 24, 2023 4:03 pm
Johan, I realize that you asked specifically about lenses, but I'm more of a systems guy so I'm thinking about the bigger picture of how to create 350MP to 600MP images of flat art.

As background, I was working in the bowels of panorama stitching software twenty years ago when the first gigapixel panos were being made. At that time we routinely corrected for lens distortion (barrel/pincushion/gullwing) inside the stitching software. CA correction was commonly done by other programs. These days focus stacking is also common (and I'm the fellow who wrote Zerene Stacker).

So, from my standpoint the critical aspect of your problem is to get a lens that is uniformly sharp across your camera's sensor, letting geometry, CA, and field flatness be addressed in software.

I assume that I have overlooked something which makes this approach not work for your application.

But that makes me curious: what is it that I've overlooked?

--Rik
Hi Rik. Thanks so much for your advice. I appreciate it. You make some great points here.

Also, I should try your stacking software!

I slightly disagree over CA and geometry correction though, as there is definitely a filtering hit when either of those are applied. I'm actually a VFX compositor in the daytime, and we have to run something called 'difference' mattes as part of our QC process comparing a movie's 'plate' exr sequence versus the 'composite' exr sequence. You can of course set two layers in Photoshop to 'difference' blending mode too. It is subtle, but it's definitely there if you add some sort of lookup-curve or gamma correction after the difference matte, to better see the filtering hit effects.

Does it make any real-world difference? Probably not, but I joined your site to see what lenses are truly capable of and to learn from folks like yourself, Robert O'Toole and many others. My point is I'd like to get CA and distortion removed from the image in-camera if possible, even if it's just for personal kicks.

Would love to hear back from you,
Johan.

jvanhuys
Posts: 49
Joined: Sun Jan 22, 2023 9:24 pm
Location: Seoul, South Korea

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by jvanhuys »

chris_ma wrote:
Tue Jan 24, 2023 1:48 pm
ps: if you haven't done so, it's worth looking at robert's website since he has many detailed tests.
he recently wrote a lot about the HR-M line which covers the range you're interested in:
https://www.closeuphotography.com/blog/ ... by-qioptiq
https://www.closeuphotography.com/blog/ ... -announced
https://www.closeuphotography.com/blog/ ... m-mtf-data

the test is not quite finished I think but he posted some of his findings here on the forum:
http://www.photomacrography.net/forum/v ... 7d50665bce
Hey Chris,

So I had a look at that 'd.fine HR-M 2.8/80 0.09x' and it seems literally perfect for my needs. I contacted the distributor I'm hoping it'll be around $1000USD or less. The Rodenstock you mentioned is out of the question due to the price though. Thanks again.

chris_ma
Posts: 570
Joined: Fri Mar 22, 2019 2:23 pm
Location: Germany

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by chris_ma »

jvanhuys wrote:
Tue Jan 24, 2023 5:48 pm
I slightly disagree over CA and geometry correction though, as there is definitely a filtering hit when either of those are applied. I'm actually a VFX compositor in the daytime, and we have to run something called 'difference' mattes as part of our QC process comparing a movie's 'plate' exr sequence versus the 'composite' exr sequence. You can of course set two layers in Photoshop to 'difference' blending mode too. It is subtle, but it's definitely there if you add some sort of lookup-curve or gamma correction after the difference matte, to better see the filtering hit effects.
cool, I'm also sometimes doing VFX compositing jobs :)

as for filtering or undistortion:
obviously there's a some filter softening, as with any change of image geometry. but the point is that unless we use the most perfect lens in it's most perfect design point, we'll always have some undistortion to do. I have specialty lenses which have 0.05% distortion in the corners, but that's still 5 pixels on my camera, and the filter hit is similar for 5pixels or for 50pixels.
what's more, I'm mainly seeing the filtering on noise and debayer artefacts and not on real image detail.
if you apply some sharpening and reapply some noise the difference between the original and the undistorted image should be pretty much impossible to spot.

what program are you using for stitching?
chris

chris_ma
Posts: 570
Joined: Fri Mar 22, 2019 2:23 pm
Location: Germany

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by chris_ma »

jvanhuys wrote:
Tue Jan 24, 2023 5:59 pm
So I had a look at that 'd.fine HR-M 2.8/80 0.09x' and it seems literally perfect for my needs. I contacted the distributor I'm hoping it'll be around $1000USD or less. The Rodenstock you mentioned is out of the question due to the price though. Thanks again.
great, that should be an excellent fit. Let us know if you get any pricing info.
from the predecessor lenses in their line up I'd expect something in the 2500-3000USD range.
the 120 float is about 5000USD here in europe.
the xenon sapphires are also in the 4000-5000USD range new if I recall properly (I got one second hand for a bit over 1000).
chris

Lou Jost
Posts: 5948
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by Lou Jost »

I've just finished a large test of enlarger lenses at about 1/4x-1/5x. After testing a wide range of enlarger lenses, including Apo-Symmars, Fuji EP and EX series, and all the usual (non-Apo EL) Nikon enlarger lenses, I found that the very best was not an enlarger lens at all but rather the 105 Sigma Art lens for mirrorless cameras. No CA, much higher contrast, better corners than virtually all enlarger lenses. However I generally used small stacks for these tests so I can't speak about flatness.

Pixel shifting can bring you up to very rich clean detailed images.

jvanhuys
Posts: 49
Joined: Sun Jan 22, 2023 9:24 pm
Location: Seoul, South Korea

Re: 1:20 to 1:5 (0.05x to 0.2x)

Post by jvanhuys »

Lou Jost wrote:
Tue Jan 24, 2023 7:04 pm
I've just finished a large test of enlarger lenses at about 1/4x-1/5x. After testing a wide range of enlarger lenses, including Apo-Symmars, Fuji EP and EX series, and all the usual (non-Apo EL) Nikon enlarger lenses, I found that the very best was not an enlarger lens at all but rather the 105 Sigma Art lens for mirrorless cameras. No CA, much higher contrast, better corners than virtually all enlarger lenses. However I generally used small stacks for these tests so I can't speak about flatness.

Pixel shifting can bring you up to very rich clean detailed images.
Hi Lou this is awesome info thanks. Regarding field-flatness...
Not sure how scientific this is, but what happens when you bring objects in the extreme corners out of focus? From my totally non-technical standpoint, the bokeh shapes look very different, more diagonal in the corner, vs something roughly similar to a circle in the centre (obviously dependent on shape of the aperture blades), but I'm sure you know what I mean.

Also, it seems the published IC for the Sigma is 35mm, which doesn't sound plausible coz you'd need a 42mm image circle for full-frame no? Is it an APS-C lens?
https://www.imaging-resource.com/lenses ... fications/

As I mentioned in the initial post, I'd need at least 60mm image circle. Let me know your thoughts?

Post Reply Previous topicNext topic