Does DXO Handle In-Body Cropping?

I’m new here and am looking to purchase a copy of PL very soon. But before I do, I was hoping to gain some understanding on whether the tool is compatible with in-body cropping. I do a lot of birding and have found that using in-body cropping is a big-time saver for me. For instance, the Canon R5 can be set to crop down to 1.6x as if an APS-C lens was being used. The new Fujifilm X-H2 has a “digital teleconverter” mode to get an effective 1.4x zoom in exchange for throwing away half of the 40MP resolution. There are a number of other camera systems that also have similar implementations (although some only operate in JPEG mode). I found only one other similar (unanswered) question in my forum search (Optics module for (Nikon) APS-C/DX lenses on fullframe camera in crop-mode).

So, to the root of this post: If I have photos coming out of a camera body that are of “medium” resolution, but are also already cropped, will this totally flub any DXO corrections that are applied? Since I haven’t had much firsthand experience with the tool, am I even asking the right question?

Hi, welcome to the forum.

I can’t completely answer to your question regarding the specific Canon R5, but having a Nikon Z7 which has also various modes for Raw and jpeg, ie I can choose full resolution at 48 Mpix, or medium at 28 Mpix, or small at 11 Mpix, and as the raw are real raw in each size (in some previous models it was not the case as already demoisized raw) they are supported by DPL.
They are not really cropping perhaps as you relate here, but the effect should be the same.

So you have to check that the raw produced are real raw.

And to ensure before buying, you can try DPL for 30 days free…

1 Like

My Nikon D850 can shoot full frame, APS-C, 5x4 and square format. each image is complete and not “cropped”.

Thank you for the replies. I guess this is a rare use case and it’s a bit different from both of you. Indeed, DXO should handle the smaller resolution photos without hiccup, but when you crop the photo instead of down sampling to achieve the reduced resolution, I can’t imagine the tool knows what to do about pincushion and other corrections without a special profile for the import tool. I’d expect most would tell me to just do cropping out of the camera body… sigh

I regularly use the 5x4 ratio on my D850, which gives me an image of that is the same height as a full frame 3x2 image, just shorter across the width.

No matter what the in-camera crop, all distortions are handled correctly in PL because they are based on the full frame.

I’m not so sure about the

If reducing the output ratio should make any sense, then it’s saving diskspace and/or adapting better to a given layout (of magazines or web-productions), so the size of the RAW actually is no longer 8256 × 5504 but 6880 × 5504. Therefore I’m asking how can a distortion map for 3 : 2 ratio being mapped on a 5 :4 or 1:1 ratio? No need to answer, I gonna check it for myself with the super distorting 14-30/4 on the Z 7. I simply have no lens for the D850 as bad distorting as that Z crap.

I don’t know so much Joachim. I would think Joanna is using the 5 x 4 format in her Nikon to soft proof for the film camera. Much cheaper than wasting a sheet of film in case the customer doesn’t like it.

The “cropped” sizes are not made by manipulating the image, simply by only recording part of the whole sensor area.

So, for a full frame shot, the pixel dimensions are 8256 x 5504, for the 1.2x crop they are 6880 x4584, for the DX crop they are 5408 x 3600, for the 5:4 crop they are 6880 x 5504 and, finally, for the square crop they are 5504 x 5504.

But, whatever the crop, the image is made up of a straight one to one centred crop of the sensor area - in other words, there is no pixel scaling. Possibly better described as a simple masking of the sensor area.

Of course, this means, for something like the DX crop, only the centre part of the sensor gets exposed, thus cropping out the outer parts of the image that can be susceptible to more distortion.

Well, there is that :stuck_out_tongue_winking_eye: But, in fact, I just like the 5 x 4 format for certain images, especially portraits and, since I can easily get a 30" x 24" print from the file with virtually the same level of detail as scanning a 5" x 4" neg… :roll_eyes:

Which reminds me - I really must get the Ebony out of the cupboard and start using it again. After all, what’s the point of a chest freezer half full of film when it could be used to make stunning prints?

No, they are not handled correctly, @Joanna. I just made a series with Z 7+ 14-30/4 and D850 + 14/1.8 + 11/4

So far I can say as long as the height is identical (D850: 8256 / 6880 / 5504×5504 or 3:2 / 5:4 / 1:1), the differences in divergence of the distortion are negligible. But for 6880×4584 and 5408×3600 (DX) it’s a different story. It gets worse when I look at the 14-30/4 Z: Difference from FX to DX is massive. Yes, optical modules were switched on, and I used the trial version of PL6:




And as I was already in crashtest-mode, I also wanted to see how C1 deals with their “Manufacturer Profiles” of all lenses. All lenses? No, only the (terribly distorting) 14-30/4 had one, for the 14/1.8 and 11/4 (Sigma and Irix) in front of the D850, there was only a “generic profile”.
▼FX Z7

▼DX Z7

I could add the other two pairs of the 14/1.8 and 11/4, they are pretty straight, even at DX size.

I’m only talking about distortion correction here - every other lens correction I can’t compare (and I also don’t care much about vignetting, and extra sharpening is cheating for me, don’t want over sharpened images), but after this little experiment DxO’s lens profiles lost their glory for me. They are not bad, but none of them was a reason to wait for it or switch to PL. C1 does it better here, sorry. :confused: I didn’t expect such a difference.


Can you share one of your RAW files that was cropped in camera? Use a sharing service and post the link here, if the file is too large to be attached directly.

To the original question I forgot to add: While I was trying to find out what DxO’s lens profiles are doing when a smaller than FF area of the sensor was selected, I run the same series on a Sigma fp-L. That one has 7 different ratios to set up, wow. And delivers the same FF output for all of them either in PL or C1, booooh! :smiley:

Whereas both Nikons did what I was expecting: the smaller the frame, the smaller the file size, cool, which was what @Joanna already said before.

However, I could not compare what DxO’s lens profile would do as DxO doesn’t exactly spoil users of L-mount :face_with_raised_eyebrow: there was simply still no profile for 24/2 available. And there’s also nothing to compare, if the camera output always shows max. sensor area, no profile would make a difference.

However 2, using Sigma’s “slow to the max!” “de-accelerating” RAW converter PhotoPro, the ratio is used on the edit screens and - magically - the formerly PL-curved curtain rail appears straight. I doubt that anyone would find a RAW-converter as slow as Sigma’s, but at least their lenses and cameras are great for me.

After all these trials, I’d say it also depends on how the camera manufacturer thinks about RAW. Sigma apparently doesn’t want to limit later decisions, Nikon thinks if a photog chooses to go the path of smaller ratio it’s a decision worth to reward with reduced file-size.

At first glance, I would say that you are using a zoom lens and changing the focal length to capture the same subject area. In which case, you may well not get perfect correction at different crop dimensions.

As far as I can tell, lens modules are only truly optimised for something like the two extremes of the zoom plus a couple of lengths in between, hoping that will be enough. I am of the opinion that isn’t sufficient but, unless the modules are measured for every conceivable focal length, you are always going to get some imperfections. I can imagine individual zoom lenses might not even give the same results, due to manufacturing tolerances.

And, if the camera creates the same pixel-size images from the different crops, then all bets are off as that would be the equivalent of digital zoom and it could well be the in-camera software that gets it wrong.

It would be interesting to test your theory on a fixed focal lens at different crop ratios.

And which of the lenses are zooms? Only one. Two others are primes and also show various distortions at various ratios. Whereas C1 keeps the lines really straight, no matter which ratio.

@Joanna, don’t mislead yourself riding a “zoom-lens” horse: The zoom ring of the 14-30 remained all time in the14 mm setting, no matter which ratio I chose in the camera menu.

It’s a side-track here, I just jumped on your idea, one lens profile is good for all ratios. Because I think it depends how it is mathematically defined and also, if the lens distortions are regular spherical or show some “moustache” patterns due to special lens elements. It makes a difference if a circular or diagonal correction is applied, I could imagine, but I was lousy in maths. :frowning: Although it depended very much on the teacher, some I could understand better when I had the impression they themselves understood what they were talking about.

To understand better: The term “in-body cropping” actually means the same as “different ratios”? At first glance I was tempted to reply “why would you use the camera’s abilities to crop an image if you can do it better on a big screen?” For me, the term cropping always means “free position inside the frame and free ratio of the final output size”. I guess, it was choosing different ratios in the menu.

When I originally posted the message I honestly wasn’t thinking about the different aspect ratio cases, but those are totally valid. What brought this on for me was the new Fujifilm X-H2 at 40MP and my bird photography use case. Birders will almost always take more focal length if you give it to them. The X-H2 has a “special” 1.4x or 2.0x crop mode which is supposed to emulate using a teleconverter (and this is on top of the APS-C 1.5x crop factor!). So, if the user (me) is willing to throw away some of the 40MP to get more bird in the frame, then the camera will crop for you. The novel thing is that the viewfinder and display all change to this 1.4x or 2.0x crop mode so you see what you’re actually going to get. Having the distant bird fill more of the screen allows for me to more precisely apply focusing. With a 1.4x crop, I’d still get a 20MP photo which is plenty good for me. So that’s really why I’d rather do it “in-body” that in post. With @JoJu 's results, I can see that DXO would never know what to do with this mode unless it was added as a completely separate lens profile. After all, it is basically a teleconverter and there are separate lens profiles for the glass versions…

I also use DX crop mode when I know my lens will not have enough focal length to catch little birds. And I too, don’t see at this moment why I should crop in post and store 50% more MP just to avoid in-camera-cropping. On the other side: Birds are usually not sitting on entirely straight structures, and long FL also have less distortions than wide angle zooms. In that way (and in my opinion) it doesn’t matter much if there’s a lens correction profile available or not. But I might be wrong.

I just remember how much troubles I had to do focus-adjustments (for DSLR) with a long tele-zoom, which leads me to think it’s a nightmare to do lens profiles for long lenses.

1 Like