Hasselblad 20-35 Test Images

[xoda]

I bought a new 20-35 lens at a local Hasselblad store, but didn't bring it home nor open it since I wanted to test the store's sample copy first. They don't allow returns or exchanges after opening the box.

I forgot to bring a tripod today, so the images were taken handheld with IBIS on. I was disappointed in the sharpness of the 20-35mm, especially on the left side at 20mm. It doesn't even seem to improve much stepped down to f/8. I basically only shoot landscapes, so corner to corner sharpness is very important to me.

But today, I was also very disappointed by the sharpness of the XCD 21mm & 30mm lenses that I'd brought to test & compare. Perhaps my lenses have gotten misaligned or decentered since I last tested them? The buildings I shot at all more or less on the same plane at least 50 meters away, so they should be all very much beyond the hyperfocal distance for any of these lenses even wide open.


Anyways, what do you guys thinks? Is the 20-35 lens worth getting? Or should I return it, and just send my own 21mm back for service / calibration? Here are the raw files:

https://1drv.ms/f/s!AoyxytEAaZ1JjodV2pln0mLsJ2VYnw?e=BsxqIN

 

[Paul2660]

Your crops IMO show the 21mm is better. Look at the railing in 2nd crop and windows towards the corner on 1st crop. As much as I hate to say it, I would also look at Phocus to see what the images look like. I have found that Lr does not have the best lens corrections for Hasselblad V glass.
Paul

 

[glennedens]

IBIS might be impacting the tests as well? I'd do the test on a tripod with IBIS off, I have seen another camera with an IBIS issue making all lenses look bad. After a repair all was fine. YMMV

 

[rdeloe]

I downloaded your images and had a look. The first thing I'll say is if you're going to make images to evaluate a lens and you want to compare lenses, it's not worth looking at any that you didn't make with the camera on a tripod, IBIS off, and wind and vibration under control. I also didn't find it particularly helpful that you pointed up as you did. If you're going to compare performance against a building, as you've done, it's much better to be level and square to the building. Ideally you want to be high enough in a neighbouring building so that you're not just showing sky. Also, you should keep in mind that it is exceptionally difficult to focus in exactly the same spot between sequences of tests with the same lens, let alone different lenses.

All that aside, these look good to me. A better test might reveal issues that are masked in these samples by the way you did the test. But what I'm seeing in these files looks quite good. I was expecting a hot mess from the 20-35 given all the moaning I've been reading, but I was pleasantly surprised.

Were I in your shoes, I'd find a test scene where I could keep the camera level and square. I'd then make several sets of images from wide open to whatever your consider the smallest aperture you'd use. Re-focus on the same thing between each set. Then compare each set to find the "best" one. You'll discover that it may not be clear cut which is best overall because of the way slight differences in focus translate across the frame.

 

[hcubell]

Your crops IMO show the 21mm is better. Look at the railing in 2nd crop and windows towards the corner on 1st crop. As much as I hate to say it, I would also look at Phocus to see what the images look like. I have found that Lr does not have the best lens corrections for Hasselblad V glass.
Paul

As of today, LR has no lens corrections for the new 20-35 XCD zoom.

 

[vjbelle]

A quick look at the files and I have to agree with Rob that everything looks pretty much OK. I see no ill effects from IBIS - if there were the entire image would be affected. I do not see a bias from left to right and the zoom you shot looks fine.

But..... for sure get the camera on a tripod and re-shoot. If possible get up a little higher so that you can fill the frame from top to bottom with the subject in the same plane of focus. If you can't to that then try to take different images that would concentrate with bottom of the frame vs. top of the frame with the camera level.

Too bad you can't shoot the zoom you are buying....

VIctor B.

 

[Paul2660]

As of today, LR has no lens corrections for the new 20-35 XCD zoom.

I realize this. I just don’t find the accuracy of the corrections to be that good in general. Finding issues with strange banding depends on the image content. And the vignette correction is not that good on the longer lenses.
Paul

 

[JaapD]

These days lens designs heavily rely on lens corrections. Some aspects of the total lens design will be carried out through the optics construction itself and some of it through lens corrections. One cannot conclude on the lens performance without taking both aspects into account.

Cheers!

 

[jotloob]

These days lens designs heavily rely on lens corrections. Some aspects of the total lens design will be carried out through the optics construction itself and some of it through lens corrections. One cannot conclude on the lens performance without taking both aspects into account.

Cheers!

If this is true , there should not be any variations between the lenses produced . This can be seen in the resulting images . Then they should all be the same . What is wrong in my thinking ? ? ?

 

[JaapD]

If this is true , there should not be any variations between the lenses produced . This can be seen in the resulting images . Then they should all be the same . What is wrong in my thinking ? ? ?

Nothing wrong with your thinking. They should be all the same optically, meaning 'within the test-spec limits of the manufacturer'. Keep in mind that there will always be variations from sample to sample, especially on multi-glass constructions like zoom lenses.

Cheers!

 

[jotloob]

Ok. But then , if you put the lens correction on top of it , all lenses should have (almost) the same output quality . And you would not see any faults in the output .
I tend to compare this with a calibration of displays .

 

[mristuccia]

Ok. But then , if you put the lens correction on top of it , all lenses should have (almost) the same output quality . And you would not see any faults in the output .
I tend to compare this with a calibration of displays .

Adding a constant to a variable will still give you a variable.

 

[Ben730]

I bought a new 20-35 lens at a local Hasselblad store, but didn't bring it home nor open it since I wanted to test the store's sample copy first. They don't allow returns or exchanges after opening the box.

I forgot to bring a tripod today, so the images were taken handheld with IBIS on. I was disappointed in the sharpness of the 20-35mm, especially on the left side at 20mm. It doesn't even seem to improve much stepped down to f/8. I basically only shoot landscapes, so corner to corner sharpness is very important to me.

But today, I was also very disappointed by the sharpness of the XCD 21mm & 30mm lenses that I'd brought to test & compare. Perhaps my lenses have gotten misaligned or decentered since I last tested them? The buildings I shot at all more or less on the same plane at least 50 meters away, so they should be all very much beyond the hyperfocal distance for any of these lenses even wide open.


Anyways, what do you guys thinks? Is the 20-35 lens worth getting? Or should I return it, and just send my own 21mm back for service / calibration? Here are the raw files:

I only looked at the pictures 20 mm at F8. They look ok, a little CA and a little bit smeary in the corners, but that's to be expected with a 20mm medium format zoom lens. However, I think the Fuji GF 20-35 performs much better at 20mm.

 

[jotloob]

Adding a constant to a variable will still give you a variable.

I do not know , how lens correction works .
But my imagination is , that you replace the "bad" IS value by the "good" SHOULD BE value .
But that would also mean , that you would have a different corretion value for each single produced lens .

So I must admit , that I did not understand that LENS CORRECTION properly .

 

[JaapD]

I do not know , how lens correction works .
But my imagination is , that you replace the "bad" IS value by the "good" SHOULD BE value .
But that would also mean , that you would have a different corretion value for each single produced lens .

So I must admit , that I did not understand that LENS CORRECTION properly .

To put it simply, with lens corrections you re-align all three/four RGB/RGBG color channels on top of each other in such a way that a square in the image becomes a square again. => correction of distortion and chromatic errors. Additionally, there is also a correction on vignetting. Some software (DXO) can also correct for a reduction of sharpness near the corners and edges.

This is all based on the data of one set of lens parameters, no correction for each individually produced lens. So you’ll remain having (small?) sample to sample variations.

Cheers.

 

[Geoff]

Put another way, the corrections fix issues stemming from design, not manufacturing variation.

 

[Doppler9000]

These days lens designs heavily rely on lens corrections. Some aspects of the total lens design will be carried out through the optics construction itself and some of it through lens corrections. One cannot conclude on the lens performance without taking both aspects into account.

Cheers!

Some lenses are designed to rely heavily on post-processing correction, while some, see the big, fast Sigmas, for example, are highly corrected optically.

The latter have inherent image quality advantages - post processing is costly, with regard to data. The former can be lighter, smaller, and cheaper, all else equal.

 

[JaapD]

Post-processing costly w.r.t. data? Are we talking about lens corrections? It doesn’t require any more data, as in ‘data storage’. Lens corrections are also carried out on the fly in the RAW converter, not requiring much data-processing. And in case of jpg it’s all done in-camera. Maybe I miss something here.

The lens designer has increased freedom of choice to perform some lens corrections / aberrations in his optical design while correcting others during image processing, something impossible in the time when we were still using film. One example of this is Fuji’s 14 mm f/2.8. Optically perfectly corrected on edge to edge sharpness, distortion and chromatic aberrations but with a truckload vignetting. So the vignetting part had to be corrected through lens corrections. On the other hand some of my Fuji zoom lenses require to be corrected on barral distortion and chromatic aberrations, less on vignetting.

What’s in it for you or me? => in the end cost-wise a better performing lens (i.e. lesser need to use costly APO glass).

I hope not to be off-topic here

On-topic: Suggestion to OP; shoot some in-camera jpg’s instead of RAWs. I expect the jpg’s to be lens-corrected. Don’t know if with Hasselblad you’ll first need to manually upload the lens corrections from the HB website to the camera.Then evaluate and decide whether you’ll appreciate its quality.
UPDATE: forget this as Matt points out that in-camera JPG’s do NOT have lens corrections applied.

Cheers!

 

[MGrayson]

This has been reported before, but I just tested it again and in-camera JPEG's do NOT have lens corrections applied. (X2D, XCD 25/2.5 @ f/4, but you get the idea)

No corrections applied to FFF file. Matches the JPEG on vignetting and distortion.


Lens corrections applied to FFF file. Vignetting and distortion now different from JPEG.


Matt

 

[JaapD]

Thanks Matt for pointing this out. That’s pretty poor that in-camera JPG’s do not have lens corrections applied. What were they thinking at HB Headquarters?

Cheers!