[Capriquorn]

Hi, I seem to have quickly amassed a collection of lenses for my K-70, none of which I’ve really tried out properly. Now I want to find out which are the sharpest. Can anyone help or point me to a thread that might give some idea of how to test these subjectively. I know glass has a sweet spot, but where to test on the focal range and at which apertures?

[house]

Hi, I seem to have quickly amassed a collection of lenses for my K-70, none of which I’ve really tried out properly. Now I want to find out which are the sharpest. Can anyone help or point me to a thread that might give some idea of how to test these subjectively. I know glass has a sweet spot, but where to test on the focal range and at which apertures?

Did you mean to write "objectively"? (se bold above)

Almost all lenses are best at f5.6 or f4. Exceptionally good lenses are supposed to be best wide open but I'm not sure thats just a theoretical conclusion and it doesn't apply to any Pentax lenses as far as I know.

[marcusBMG]

lens testing eh! Well, philosophies extend from "if I wander around taking enough pics I'll acquire a sufficient impresson of a lens for my purposes" (dubious - see technique below), to rigorously methodical test chart imaging (anal, geekdom and MFT charts beckon).For some perspective, have a look at the reviews posted by these people, all of which have, IMO, considerable merit.

Dan Eurritt codectest.com.

Vintage lens reviews.

Women and dreams blog spot.com. Very good example of a discursive review.

Optical Limits.

Bengt Sandberg BKS

lensqaworks.com Mainly minolta, very well presented reviews.

Ching-Kuang Shene dcview blog, in chinese.

and of course the reviews here, although most posted here are of an anecdotal style, with a significant proportion amounting to little more than "I think I am pleased with my latest purchase".

Some things to think about:

1. Lighting. IMO good lighting is essential, it is pretty pointless shooting away in poor light or backlit conditions unless you are specifically trying to glean something about performance in those conditions eg flare.
2. Subject. There are reasons why brick walls are good. They have both large and small detail. The rectilinear patterns show up distortions. Being flat the whole of the subject is nominally in the plane of focus, or can show up field curvature. Forget about latching onto the bush in the garden, most of it will be out oof the plane of focus, front or back and will reveal nothing.
3. Technique. To focus on eg resolution, you need to minimise all other reasons the image may not be sharp ie focus, shake, vibration. That means: tripod, focus with 10x Live view, mirror lock up and timer shutter.

and then there are the lens variables: F-stop; zoom, focal distance and centre frame vs edge/corner. Personally I only bother with wide open to F11 as the mainly used F stops. Above F11 diffraction is kicking in. It is IMO important to test both close(r) and distant focus. With a zoom, the performance at the ends and at some point(s) in between. And edge vs centre is important with eg wide angles (landscapes see Dan Eurritts reviews), perhaps less so with telephotos used for eg wildlife pics.

You can have a look at what I did re DFA 150-450 vs Sigma 150-500mm here.

And bear in mind the wise aphorism of a grizzled old photographer: "every lens is good at f8"!

[MaineNative]

For testing sharpness, I often take a picture of a bookcase using a tripod and offset flash. The bookcase is kind of like an eye chart where there is a mixture of text sizes. (The publisher’s name is usually smaller than the title and author, for instance.). If you can see the texture, dust and scratches on the book cover, you know you have a sharp lens. You can also compare the text of the titles in the middle with the ones at the edges.

I also like taking pictures of tree bark. The more texture and detail, the sharper the lens. I will also take a picture of a tree branch wide open and look at the bokeh behind the branch.

[Wasp]

A test that I can think of is to put up a page from a newspaper on a wall and take pictures of it, using a tripod. The text will show up any lack of sharpness, especially in the corners. Or print up some test targets from a suitable source.

BTW, I still have a set of Paterson test targets that I bought in the eighties and never used.

[Just1MoreDave]

Shoot in RAW so lens or image adjustments don't fix flaws.

For a sharpness test, focus is ridiculously important. I spend a lot of time on focusing, then repeat the test three times and pick the sharpest example. I have used a brick wall and newspaper clippings taped to a wall. A lot of lenses are sharp enough in the center but fall off in the corners, so the wall provides a good way to check the corners. You need to match the wall distance to focal length so there isn't too much distance difference between the center of the frame and the corners. Consistent lighting is important. Once you get the setup and focus right, take some shots to get a good histogram at say f8 for one of the lenses. That is my base exposure. Then it's easy to take a series of shots at each full stop from wide open to f11 - fixed ISO, every time you open up the aperture up a full stop, raise the shutter speed a full stop. For KA or autofocus lenses, you can use AE-L to lock the exposure, then every time you change aperture in Av mode, the shutter speed adjusts to match. I turn off autofocus on AF lenses so each series has the same focus point. When I'm done with a series, I put my hand in front of the lens and take a shot, so it's easier to see one series when I'm processing.

On the computer, I take crops from one corner and the center, maybe 150x200 pixels. Then I make two grid composite images, one for those center crops and one for the corners. One axis of the grid is apertures. One axis is the different lenses. When it's complete, it's very easy to see all the centers in one image, all the corners in another.

For AF lenses, I use CDAF. For manual focus, magnified live view. I put hoods on the lenses and block the viewfinder for the test shots. Two second delay. Try not to kick the tripod.

I have tested lenses for flare by setting up a bright LED flashlight at one end of a dark hall, tripod at the other end, lens wide open, framed so the light hits off-center. I test for colors and bokeh by taking the lenses to a botanic garden. They are OK with tripods and have plenty of subjects.

[DWS1]

I've found all my lenses are consistently best when stopped downed 2-3 stops from the maximum aperture, regardless if it's a f:1.4 or f:5.6 maximum, and zoom lens are consistently best in the middle half of the zoom range. I too normally use shots of a bookcase, as MaineNative suggested above, as good photos to evaluate lens characteristics.

[Capriquorn]

Did you mean to write "objectively"? (se bold above)

Almost all lenses are best at f5.6 or f4. Exceptionally good lenses are supposed to be best wide open but I'm not sure thats just a theoretical conclusion and it doesn't apply to any Pentax lenses as far as I know.

Yes! Objectively. 🤦

[house]

Yes! Objectively. 🤦

If you have access to a printer, a tripod and a computer you can run MTF mapper / Home / Home to get mtf curves and 3d grid resolution charts similar to those imaging resource use. I tried it out and it helped med understand the performance of some of my lenses better. It's quite hard to figure out what is happening by inspecting photographs only. You can see something is happening but it's hard to figure out what it is.

Mtf mapper involves shooting a chart (supplied in separate zip file) in dng format. Setting up pixel pitch in the settings then opening up all your shots and it will generate a variety of charts depending on your selection.

[AstroDave]

Here’s my lens testing/comparing regimen.

It doesn’t qualify for real quantitative results, but I think it tells me how my lenses perform at the edges and corners of field of view, as well as how sharp they are in the middle.

This is based on the (old - year 2000) ISO standard resolution test chart ISO 12233. You can buy a very nice version of this for more than $100 (!), or you can do something based on the efforts of Stephen Westin, which is more-or-less what I describe below. His version of the (old!) ISO chart can be found at ISO 12233 Test Chart .

If you have a very high resolution printer (around 2400 dpi or so), you can just print out his chart and use it to fill the image frame of your camera. However, most of us don’t have a printer of that quality (and, you won't be testing at the highest resolution around the edges of the frame). Modern, low-end laser printers typically print at 600 dpi. Many ink jet printers claim a higher pixel density, but I don’t have much experience with them (but am rather skeptical that they really can print at claimed resolutions of 1200-1440 dpi).

I have taken a subsection of Westin's chart, which gives information on lines per frame height for the range from 500 to 2000 (the 5 to 20 scale), and blown it up to where it gives a reasonable printout with my equally venerable HP Laserjet 2100 printer, which actually prints at 1200 dpi.

Here is a link to my jpeg file which is at 2400 dpi: http://photodave.us/ResTests/5-20-LinePairs-2400-137mm.jpg (if that ain't enough resolution there is also a 4800 dpi version!). If you open this file with e.g. Photoshop and print it at actual size, you should get an image which is about 137 mm long. Use the best printer resolution you can, and examine your printout closely to make sure that the finest details are adequately resolved. It should look like this:




Print out 9 of these at as high a resolution as you can, and paste them in the center and at the edges and in the corners of a piece of cardboard that is 18 x 32 inches in size. This size is critical if you want accurate results!!

Here’s what that looks like. I put a cross hairs target in the middle to help with focusing, as well as a color reference chart, and a length scale. (I made 3 of these for a photo course I taught, for the students to photograph so I could understand their cameras, hence the hand-written letter near the middle.) This is a low quality jpeg, so the resolution isn't as good as in the original.




Here's a 1-to-1 cut out of the top left corner, at full quality. This is with the SMC Pentax-D FA 100mm F2.8 Macro WR on my K-1, at f/8 . This combination readily resolves to much better than 2000 lines per height, all over my test chart.



Set up your camera (on a tripod!) at a distance from the chart so that the rectangle fills your camera image frame (generally across the frame for a standard DLSR 1.5:1 ratio). Then, the grid scales properly in lines/frame height. Take some effort to make sure that your camera is centered on, and orthogonal to, the test chart. Uniform illumination is nice, along with a proper color balance.

I take frames for just about every f-stop across the range of a given lens (use the Av mode and mirror lockup). For a zoom lens, you probably want to test at 3 to 5 focal lengths across the zoom range. Make sure to adjust the separation between the chart and the camera so the chart files the frame at all focal lengths. Feel free to check more focal lengths, especially if you have a favorite or two.

Then, just use your favorite image browser to look at the results, and see how many lines you can discern. I made a little table to print out where I write down the frame number and lens/focal length and f-stop and the limiting lines count for the center and all the edges and corners.

For modern sensors (say 15/16 Mpixel and up) and good lenses, you may find that you can resolve to 20(00) for almost all frames. If this is the case, you can always move the chart farther away - if you arrange things so that the chart fills only half the frame across (it's twice as far away), then all the chart numbers would be doubled - i.e. you could check lines from 1000 to 4000 per frame height.

Not only can you check your camera/lens resolution, but you will probably discover a host of other lens frailties, such as coma and chromatic aberration. The arrays of lines are quite unforgiving of such things! I usually just shoot jpegs for quick testing, but if you shoot RAW, you may be able to invoke lens corrections to see if they improve your results.


How do I get my inserted images to show without having to click on them?!?! Does it matter that they are not in a secure web site (i.e. not a https prefix?)

Looks like you need https! I put these in google photos and created an imbeddable link (what a PITA!), so now you can see my images directly.

[marcusBMG]

How do I get my inserted images to show without having to click on them?

A lot of your pics exceed the site size limits of ~1270px x 8xx (can't remember the exact number) pixels, server automatically does a link. And https not http is also a rule I think.

[AstroDave]

Your pics exceed the site size limits of ~1270px x 8xx (can't remember the exact number) pixels, server automatically does a link.

Nope - I don't think that's it - the last one (the cutout) the original is only about 400x400. I think the displays are automatically resized to fit (within the limits you mention).

Using https seems to be the trick.

[Just1MoreDave]

...
Set up your camera (on a tripod!) at a distance from the chart so that the rectangle fills your camera image frame (generally across the frame for a standard DLSR 1.5:1 ratio). Then, the grid scales properly in lines/frame height. Take some effort to make sure that your camera is centered on, and orthogonal to, the test chart. Uniform illumination is nice, along with a proper color balance...

Some mason's string attached to each corner of the cardboard would make that easy. Run the strings to the camera and when they are all equal, the camera is centered. Then just level it and make the cardboard vertical. Mason's string doesn't stretch so it's good for measurement. Home Depot has it around the cement and rebar area.

You have plenty of room on the cardboard for a post-it note that can label each image. If you test a large group of lenses together, it is very useful to quickly group the shots taken with each lens.

[Capriquorn]

Wow! Thank you for all the helpful replies. I hadn’t considered things like distortion or CA. What I might do is a general ‘brick wall/bookcase’ test for all, then make a shortlist from those results and go in depth Using one of the Daves methods? Raw shooting, even lighting and mirror up shooting seem to be the consensus too.
Thanks again,
Jeff.