Tuesday, May 24, 2016

Petzval Lens Sharpness Test - Voigtlander vs. Dallmeyer vs. Scovill Peeless vs. C. C. Harrison vs. Darlot

Petzval Lens Test – C. C. Harrison vs. Dallmeyer 3B vs. Voigtlander vs. Darlot vs. Scovill Peerless

There’s been some online discussions regarding quality of Petzval lenses and it prompted me to do this little test.  Some people were saying that this brand is better than that one based on how many of them were made and how much they were selling back then for (basically that cheaper lenses were worse).  Others were vehemently defending their lenses saying they’ve been happy with images for years.  I’m not in one camp or another, so for the sake of my own knowledge and to confirm or deny my own suspicions about which of my personal lenses are actually sharper than others I devised this little test.
            I would like to say that I’m aware of quite a few factors that make this optical quality test subjective and also of some factors that surely stand true in matters of collecting lenses rather than using them:

  Lenses were hand-made back in the day and thus a piece of glass made on a Monday might have been better than a lens made on Friday because the guy polishing and inspecting it just couldn’t wait to finish up that day and go have a pint of grog at the local pub.

• Glass quality improved through the years and so a Petzval from 1880s should in theory be better than one from 1850s (let’s see about that….)

• Lenses get sharper when stopped down a few stops below wide open, that’s a given.  However these days people like Petzvals for their shallow depth of focus when they are indeed wide open and so I’m choosing to conduct my test with all my lenses devoid of stops.  Lenses in this test are all about f3-3.6 so it should be a fairly fair test in tat regard

• Collectors – they are a different breed of Homo Sapiens.  Nothing is going to convince them that this brand or another is worth less because others are sharper.  This test is geared toward active users of these lenses.

Here are the lenses tested:

Control in front) Fujinon W 210mm f5.6  - this is the sharpest most corrected modern lens I own that is around the same focal length of other lenses in the test.  I’m throwing in this lens as a control to see exactly how sharp I can make my target appear.
Left to right on top:

1) Voigtlander (1865): FL 9.7in, glass diameter 3in, f3.2

2) Dallmeyer 3B (1875): FL 11in, glass diameter 3.325in, f3.3

3) Peerless Tangent dive (1872): FL 12in, glass diameter 3in, f4

4) Peerless Radial drive (~1878): FL 11in, glass diameter 3in, f3.6

5) C.C. Harrison (~1859): FL 10in, glass diameter 3in, f3.3

6) Darlot Full Plate (~1875): FL 10in, glass diameter 3in, f3.3

7) Darlot Extra Full Plate (~1878): FL 13in, glass diameter 3in, f4.6 This one is the odd-bird.  Not  only is it the slowest and longest, but it also has a small crack in the glass smack in the middle of the rearmost element – someone obviously put those elements in wrong and applied too much pressure when screwing the retaining ring back on.  Thankfully they didn’t break the glass completely…  In either case – I wanted to test it because it’s the newest one in my arsenal.

            Above parameters were not taken from each lens’ catalog listing.  Instead I focused each one on infinity (luckily, above the horizon, there were excellent fluffy clouds today!) and measured distance from the ground glass to Waterhouse slot.

            Here’s the shot of the poor rear element of the 13in Darlot with that annoying crack in it…

            I wanted to test the center sharpness of each given lens, not the curvature of focus, bokeh, contrast or other characteristics.  I just wanted to see how sharp my test target will appear under rudimentary magnification and with all things made as equal as possible.

Wet plate collodion technique here for a number of reasons.  First off it’s actually cheaper than film and is a lot more immediate in feedback.  Secondly collodion’s resolution is higher than any film base provides.  With it though come certain uncertainties, which I have tried my best to eliminate or diminish.  The biggest uncertainty being that from my understanding with prolonged development the size of silver crystals increases, thus reducing apparent resolution in my final scans.  I have tried to make my exposures vary as little as possible (while taking into account different speeds of lenses I’m using) and keeping development time consistent and timed via a Gralab timer.

I used an 8x10 Zone VI with a 4x5 reducing back.  As you can see below, while shooting the slightly varied focal lengths lenses I am filling the frame with the test target.  In my mind that is providing us with test of the ‘sweet spot’ of each lens.  The center test is obvious and I’m doing the corner clippings to see how much possible aberration set in there with each lens and how much curvature of field of focus they may have.  Now, some lenses I am testing obviously have a larger field of coverage than other (see lens-by-lens description), so in the ones with a larger image circle we can expect the focus filed to curve less within the test area.  Take that as caution.

I used a ground glass focusing loupe to make each plate appear as sharp as a given lens would allow.

Front and rear standards have been made to be as vertical and parallel as possible by using a bubble-level.  The target is on a wall, which I assume to be of vertical orientation (it’s an old building though, so who really knows, right?).

Final plates have been dried, not varnished, and photographed on a Polaroid MP4 copy stand using Canon 5D Mark II camera with a 100mm macro lens.  Fill 4x5 plates were shot at JPEG-small setting to make them manageable for import.  1:1 macro copies out of the center were shot using RAW setting.  All plates were gives same exposure while re-photographing.  JPEGs of full plates were imported into Photoshop and given an Unsharp Mask filter (100%, 1 pixel). RAW files were imported into Lightroom and sharpened 100%, radius 1, detail 25.  Then they were exported as JPEG at 100%.  Then (back in Lightroom) a tight center crop was made and again exported as 100% JPEG.
I know I'm gonna hear ALL SORTS of feedback on this post.  "Oh, you should have done this" and "Oh, you must have done this wrong".  Honestly - I did my very best to focus and let the camera settle after taking off the cap before exposure.  I used the same plate holder for all plates and plate holder was indeed in all the way on every shot.  If any of my critics would like to present their own test that would be great.  On the other hand that will really make no sense - as I said above, one Dallmeyer may be better than another.  So that eager critic might want to come over to San Diego and do the test on my lenses - I'm completely open to it.

            Here are the images of the middle section:


3)Peerless Tangent

4)Peerless Radial

5)C. C. Harrison

6)Darlot 10in

7)Darlot 13in

8) Fujinon 210W

            Here are 1:1 center images:

1) Voigtlander

2) Dallmeyer

3) Peerless Tangent

4) Peerless Radial

5) C. C. Harrison

6) Darlot 10in

7) Darlot 13in

8) Fujinon 210W

            Here are crops from 1:1 center images:

1) Voigtlander

2) Dallmeyer

3) Peerless Tangent

4) Peerless Radial

5) C. C. Harrison

6) Darlot 10in

7) Darlot 13in

8) Fujinon 210W

            And it seems that the winner in the center-sharpness category is the 10in Darlot!  It seemed to even outperfrom the Fujinon control lens.  Followinc closely in second place and matching the Fujinon is the Dallmeyer 3B.  I was honestly expecting the Harrison to come out on top, but it actually seems like it’s in the 4th place behind the 13in Darlot (the one with a crack in the glass!).

            I don’t know what this proves to anyone.  My conclusion is that my little Darlot is just as good as any other Petzval and I love it even more now.  It was actually my very first brass lens – this beauty was rescued literally from a dumpster by my college friend who heard that a photographer passed away and “all grandpa’s junk went to the dump”.  I’ll continue shooting them all because they all have a different bokeh, a test of which may or may not come in the future.  One thing is certain though – not all Voigtlanders are as sharp as others and not all Darlots are bad either like some people would have you believe based on their original cost and the mass-produced factor.

EDIT - 5-28-2016 

  So I thought that something must have been awry there in the original test - it seemed very odd that a Darlot would outperform a Harrison...  Some people pointed out that chemical focus may have been involved (this is when blue end of the spectrum is focused by a lens behind the plane of visual focus - I don't think that's the case with these lenses as I have never had images come out softer than they looked on the ground glass..., still, possible), others noted that at such close distance from lens to target any slight movement of the focus plane can throw off the results by a lot, someone else suggested that  I could have saved myself a lot of time and effort by doing the test digitally.  Armed with these suggections I decided to redo the test.
  This time I strapped my friend's Sony A7r to a 4x5 reducing back on my Kodak 2D in a following crude yet efficient manner.

  Using the nifty little 'focus assist' feature I focused on the right side of the electrical post you see in my crops.  The camera was set to ISO 50 and all exposure times were the same (I think about 1/200sec), which actually lets you see rather clearly how much (or how slightly) one lens is brighter than another.  Resolution was at finest JPEG setting. I also used a self-timer on 2sec setting, so after I pressed the button the camera had time to settle.  It was rather windless today, so I don't THINK camera shake was an issue.  COULD it have been an issue with a couple of themTotally possible!
  After shooting I exported the files into Photoshop, cropped to the point where the focus was and gave them all 100% unsharp mask filter at 1pixel.
  Since I could not move closer to the subject the pole did appear slightly bigger in the frame with longer lenses, so more pixels were devoted to it, but differences were pretty minor except with the last Darlot.   Also I did add a lens that I just received in the mail a couple days ago and was able to mount today - a 16in Wollensak Vitax.  It's also a Petzval design and of a wonderful f3.8 speed, really stoked to have this lens.

  With this method the test took about 15min instead of the previous  2 hours or so. Yes, digital IS faster than wet plate, go figure....

  One thing I totally missed - I forgot my control lens at home, so maybe some day I'll once again use that A7r and just shoot with a 210 Fujinon.  

  Here's the full frame as seen by the Sony through the Voigtlander.

  And here are the crops.
 1.  Voigtlander
2. Dallmeyer 3B

3.  Scovill Peerless Tangent Drive 

4. Scovill Peerless Radial Drive

5. C. C. Harrison

6. Darlot 10in

7. Darlot 13in

8. Wollensak Vitax 16in

  This time, as I suspected - Harrison seems to take the cake (take a look at the detail in the High Voltage sign).  Followed closely by Dallmeyer and Peerless Radial (which was made by R. Morrison who worked closely with Harrison himself, so no surprise there).  The 13in Darlot also looks great (and remember, that's the one with a crack in the rear glass).  Surprisingly my favorite 10in Darlot looks like dreck - I'm gonna write it off to camera shake, I mean c'mon - take a look at the first test results!
  As suggested to me online after I posted the first results - unless a given Petzval lens has been severely messed with throughout the ages - they are ALL GOOD.  What this test proves to me personally is that there's really not THAT much of a difference in sharpness between a coveted Dallmeyer 3B and a good Darlot.  
  Anyway, I feel rather tested-out and will now concentrate on more fun things to do.