Saturday, June 27, 2015

135mm Lens Comparison

Finding myself with a large number of 135mm lenses, I decided to satisfy my curiosity and test the following lenses: Canon 135mm f/2 L, Pentacon 135mm f/2.8 "Bokeh monster," Pentax Super Takumar 135mm f/3.5, and Carl Zeiss Jena Sonnar 135mm f/3.5. All images are at infinity (the Zeiss infinity stop was inaccurate, I used Live View for best focus) at ISO 100 on a tripod. The Canon and Pentacon were shot at f/2.8, f/4, and f/5.6. The Pentax and Carl Zeiss Jena at f/4 and f/5.6. All images were shot with a Canon 6D.

As an aside, the old M42 lenses had been sitting around in a closet for years, and I discovered that both the Pentacon and Carl Zeiss Jena wouldn't stop down properly. I had to do some impromptu CLA to get them working. The Pentax, even though it is probably older, had no such problem and is much better constructed.

There are a couple of far more complete 135mm lens reviews here and here.

I've cropped the center and (close to) edge of the following image (the cropped areas are highlighted):

Now for center sharpness:

At f/2.8 (wide open for the Pentacon, one stop down for the Canon L lens). The first image is from the Canon, the second image from the Pentacon.

Canon 135mm f/2 at f/2.8 
Pentacon 135mm f/2.8 at f/2.8

Here are the results at f/4. The Pentax and Carl Zeiss are f/3.5 lenses and so are close to wide-open, while the Canon is stopped down two stops (and is therefore optimal) and the Pentacon is stopped down one stop:

Canon 135mm f/2 at f/4 
Pentacon 135mm f/2.8 at f/4

Pentax 135mm f/3.5 at f/4

Carl Zeiss Jena Sonnar 135mm f/3.5 at f/4
Here are the results at f/5.6:

Canon 135mm f/2 at f/5.6

Pentacon 135mm f/2.8 at f/5.6

Pentax 135mm f/3.5 at f/5.6

Carl Zeiss Jena Sonnar 135mm f/3.5 at f/5.6
Now for the corner images, first the Canon and Pentacon at f/2.8. There was no image correction on the Canon as I did not have the lens correction profile loaded.

Canon 135mm f/2 at f/2.8

Pentacon 135mm f/2.8 at f/2.8
The corner results at f/4:

Canon 135mm f/2 at f/4

Pentacon 135mm f/2.8 at f/4

Pentax 135mm f/3.5 at f/4

Carl Zeiss Jena Sonnar 135mm f/3.5 at f/4
Finally, the corner results at f/5.6

Canon 135mm f/2 at f/5.6

Pentacon 135mm f/2.8 at f/5.6

Pentax 135mm f/3.5 at f/5.6

Carl Zeiss Jena Sonnar 135mm f/3.5 at f/5.6
The Pentacon is obviously soft wide-open, even in the center, and edge performance is poor. The Pentacon only really sharpens up at f/5.6, and the slower Pentax and Carl Zeiss Jena match or beat it at equivalent apertures. The only use case for the Pentacon is its mythical bokeh at f/2.8. I suppose for portraits it would do great because the softness would mask skin imperfections.

The Canon beats all the other lenses at all apertures, and its performance is almost constant from f/2.8 to f/5.6 - I expect poorer performance at f/2 but as none of the other lenses are that fast, there is no point of comparison. This is expected given its price point.

Among the two slower lenses, in my personal opinion, the Zeiss is more pleasing to the eye than the Pentax, but the Pentax seems to have better edge definition.

At f/4 and f/5.6 all of the lenses perform well at the center, and slightly less so at the edge, although the Canon is still visibly the best both at center and edge, and the Pentacon is clearly the worst. In terms of contrast, the Canon is best, followed by the Zeiss which has a multi-coating.  The Pentacon and Super-Takumar are both single-coated and have obviously lower contrast.

In my opinion, the Canon is best, the Carl Zeiss Jena is second-best (beating the Pentax only because of its better coatings), the Pentax is third, and the Pentacon dead last. Prices of the Zeiss Sonnar have come up a bit since I bought mine in 2001 for $80.  Seems that a "red MC" Sonnar like the one I bought is about $110 plus S/H on eBay nowadays. Still a great price, almost 10% of the price of the Canon. You can still get the Pentax for $20 (plus shipping) so the Pentax has got to be the best bang for the buck. A Super-Multi-Coated Takumar is $50 (plus shipping) and would have better contrast than the one I tested, so it might beat the Zeiss and would therefore be a better choice.

However, at the end of the day, a 135mm lens is rather simple by design, and at middle apertures all of them perform acceptably or better. So for general telephoto purposes at long distances and modest apertures, all of these lenses, even the cheap Pentax, would perform equivalently. The Canon and Pentacon are probably better for portraits because they can blow out backgrounds better. Also the Canon is auto-focus which makes it much more usable; it also costs ten times any of the other lenses.

Monday, June 15, 2015

Canon 10-22mm EF-S and 16-35mm L IS Compared

The results of this comparison are pretty much fore-ordained, but I was curious as to how obvious the differences were to an amateur and non-critical photography user such as myself.

I've used the 10-22mm f/3.5-4.5 EF-S lens for about a year, through a couple of extended holidays, and it's a very satisfying lens on a crop sensor body. The 16mm equivalent focal length is quite striking and gets pretty addictive, and the ring-type USM, full-time manual, focusing window, and well-damped mechanicals are a pleasure to use (not far off from L build quality, to be honest).

However, we got a Canon 6D body (a full three years after its introduction - we've been using an ancient Canon 40D on its second shutter for the past 6+ years) and the 10-22mm won't mount on a full frame body.  After some analysis paralysis, I decided that I wanted to retain the 16mm wide end and so went for the 16-35mm f/4L IS (instead of the somewhat cheaper 24-70mm f/4L IS).

I also have a Canon 550D which I was planning to IR-mod for astrophotography (the elderly 40D has been through so many holidays and memories and has a brand-new shutter, so I didn't want to butcher it). The 550D has almost the same pixel count as the 6D, so I figured it would be interesting to compare the 550D and the 10-22mm, with the 6D and the 16-35mm.

Here's the sample image that I used, which is 1/500 second, f/4, ISO 100, "natural" picture style, straight JPEG from the camera, tripod-mounted.  f/4 is wide-open for the 16-35mm, and stopped down a smidgen for the 10-22mm.  The 10-22mm had lens correction enabled, but the 16-35mm did not (the 6D's firmware didn't have the profile for the 16-35mm loaded).

The full image from the 6D and 16-35mm (it got nudged a bit and is a bit narrower than 16mm here). Note the vignetting in the corners, due to the lack of a lens correction profile.

and from the 550D and 10-22mm at 10mm. Note that there's no vignetting in the corners! the 550D has a less-sophisticated lens correction system than the 6D (only correcting for vignetting, not for CA) but it shows its magic here:

This is the center at 100% with the 6D and 16-35mm:

with the 550D and 10-22mm:

I also used a $150 Tokina 20-35mm f/3.5-4.5 on the 6D. The Tokina is a bit narrower so I scaled down the image by 20% (to "convert" the 20mm to 16mm equivalent) and here is the center:

This is the far left edge with the 6D and 16-35mm:

with the 550D and 10-22mm:

and with the 6D and Tokina 20-35mm:

I then tilted the cameras upward to put the house at the far left edge (in the immediate images above) at the bottom-left corner. Here's the bottom corner with the 6D and 16-35mm:

with the 550D and 10-22mm:

and with the 6D and Tokina 20-35mm:

The results should be self-evident, but some observations:

The Tokina is not bad at all in absolute terms, and is even more amazing considering the low, low price of $150 on eBay. In the sample images above, I would guess that the Tokina might almost be giving the 16-35mm f/4L IS a run for its money. Of course the correct comparison of the Tokina would be against the 10-22mm at 12.5mm and the 16-35mm at 20mm. In that case I suspect the Tokina would fare worst, but this post isn't about the Tokina, I just threw it in since I had it.

I hadn't realized how mediocre the 10-22mm was in the corners until I did this little test. Which only goes to show - I carried the 10-22mm through two countries, had a great deal of fun, and came back with a lot of memorable images. Never once did I wonder if I was getting critically sharp images. Do I feel that those holidays were compromised by the not-so-great 10-22mm lens? not at all. As an engineer, I certainly feel better if I take technically better images, and there are worse vices than accumulating L glass. But absolute image quality has very little to do with the personal satisfaction and memories captured with any set of camera gear.

The 16-35mm is obviously the best, as it should be at over double the price of the 10-22mm and ten times the price of the Tokina 20-35mm. In addition to its high price, the 16-35mm is also heavy, and large, about an inch longer than the 10-22mm.

Of course if you want a moderately-priced ultra-wide zoom on a full-frame Canon body, the 16-35mm f/4L IS is probably the best game in town right now, as it's sharper than the f/2.8 version, is lighter, cheaper, and has IS. In my opinion, if you want a practical ultra-wide zoom for a full frame Canon, this lens is the no-brainer choice. It's even a pretty good deal for an L lens, considering it is Canon's latest and greatest not-so-expensive ultra-wide.

Friday, April 10, 2015

Astro-Tech 8" Imaging Newtonian, Part VIII

Final refinement on the Astro-Tech Imaging Newtonian: rather than paint, powder-coat, or anodize the tube, I simply wrapped it with fake carbon fiber wrapper (used to make regular cars look more ricey).

A 1m x 1.7m cut cost me S$25 (about US $20) and I only ended up using half of it.

A useful tip: putting the holes in the wrap for the screw holes is much easier with a hot soldering iron!

The large hole for the focuser must be cut out with an X-Acto knife, box cutter, or similar implement.

The finished telescope actually looks pretty good, and no one's the wiser that it isn't real carbon fiber.
One caveat though, the wrap is rather soft. I dinged the telescope tube against an aluminum door frame, which cut into the wrap and exposed the metal underneath. I was able to push back the wrap over the exposed metal, but the finish is no longer pristine.  I would expect that for larger dings, it might be necessary to overcoat the damaged portion with clear varnish or similar to prevent the wrap from peeling.

Sunday, March 22, 2015

Berlebach Dovetail Clamp with Shoe for Takahashi EM-11

One problem with the Takahashi EM-11 (and EM-200) is that they are designed for Takahashi tube clamps, and don't come with a Vixen or Losmandy dovetail saddle.

ADM Accessories is probably the only company left making saddles (since Robin Casady no longer does).  I was not interested in a Losmandy saddle because it is large and requires large Losmandy dovetails which drives up the weight.

Vixen makes a compatible clamp, but frankly, it's ugly, and is of the wrong diameter. The photo below shows the (white) Vixen dovetail and how it's obviously of the wrong diameter and finish. It also uses two set screws which ding your dovetail and aren't terribly secure.  This was what I was using when I got the EM-11 because it is what came with the mount.
Luckily I found this Berlebach dovetail clamp with shoe.  It is not specifically made for Takahashi mounts (it's probably designed for Vixen Porta type mounts) but it has the 35mm hole spacing for the two central bolts.

Unfortunately, I discovered that the holes are 6mm diameter, not the 8mm needed for Takahashi mounts.  So I had to drill them out.  The Berlebach clamp uses countersunk machine screws, not the counterbored ones that the Vixen uses. Hence the Berlebach cannot use M8 SHCS machine screws, but rather M8 x 25mm countersunk ones.

A nice bonus of the Berlebach clamp is that its diameter is only slightly larger than the Takahashi mount top (at least for the EM-11) so it doesn't look too out of place.  The clamping shoe is also a nice touch, avoids marring your Vixen dovetails.

In my case, I used M8 countersunk machine screws with socket heads. The socket heads made the machine screw tops a bit too large, as is obvious in the first picture. I will have to find replacement machine screws (probably Philips) with slightly smaller heads that will fit exactly in the dovetail clamp's countersunk holes.  On the other hand maybe M8 bolts will never fit - the clamp was designed for M6 bolts.

Sunday, March 15, 2015

CCD Inspector Results

Here are CCD Inspector results from several of the telescopes and flatteners/reducers which I've owned or had the chance to use.

Astro-Tech AT90EDT FPL-53 triplet, 600mm focal length, with William-Optics 0.8X Reducer/Flattener II, APS-C chip (QHY8). Curvature is 22% which is not too bad, there is a hot spot in the center.
The same AT90EDT APO triplet, with the Orion 1.0X non-reducing flattener (which is very similar to the AT2FF). Curvature is actually slightly worse than the William-Optics reducer.
William-Optics ZenithStar 70ED, 420mm focal length, with William-Optics 0.8X flattener/reducer II.  Curvature is rather high at 34% on the QHY8 APS-C chip, this seems to indicate that the WO Flat II is optimized for higher focal length (and thus smaller field curvature).
My current imaging refractor, TMB design Lomo 80mm f/6 OK-4 APO triplet, with Televue TRF2008 0.8X flattener/reducer (400mm - 600mm focal length) and Canon EOS 40D.  The Lomo 80mm has a native focal length of 480mm, so within the specification of the Televue flattener. Curvature is 23% which is about the same as the AT90EDT with the WO Flat II. Stars are smaller, although that could be due to better guiding. I believe these results are comparable to the above two, since the Canon DSLR has an APS-C chip, same size as the QHY8 (although smaller pixels).
The same Lomo 80mm f/6 but with the much smaller Kodak KAF8300 chip (4/3 chip size, half the area of APS-C and one-fourth the area of 35mm full frame). Curvature is much less but minimum FWHM is only slightly better than the same scope and reducer with the Canon APS-C chip.
I've seen CCD Inspector results for the Televue NP-101, and frankly they don't look much better (and some reports are considerably worse) than my Lomo 80mm  with the Televue flattener. So I should probably not consider the Televue going forward.
Now here's a slightly different setup: Astro-Tech 8" f/4 imaging newtonian, with a Televue Paracorr Type 1.  It actually has better numbers than the much more expensive Lomo 80mm which is quite embarrassing.
Now for the absolute best case, I dug out some raw FITS from my 2012 telescope rental at GRAS/iTelescopes in New Mexico.  This is the curvature data for a Takahashi FSQ-106ED, KAF8300 sensor.  It looks bad because there's a major difference in the FWHM in the center and at the edge, but the FWHM's are very, very small. I'm not sure if I'm measuring it correctly.
If the measurements are correct, then all of my current results are basically invalid due to poor seeing and atmospheric conditions.