Hasselblad X2D II RAWs Record Focus Distance. No Editor I Tested Reads It

Your camera knows exactly where you focused. It has to: autofocus is exactly the act of measuring and moving until the subject is sharp. On July 2, 2026, Capture One and Hasselblad announced a partnership letting Capture One open Hasselblad RAW files natively for the first time, and buried in the release notes was a claim that Hasselblad files carry no focus-distance information for lens corrections. That left a question I could not let go of: does the X2D II write that number down?

The short answer is yes, quietly, since a firmware update last November. The longer answer involves a claim in Capture One's release notes, three test frames shot at marked lens distances, a number that came back almost exactly right, and a doctored test file that fooled me for an afternoon.

Key finding: Every 3FR from an X2D II on firmware 1.2.7.11 or later (November 27, 2025) records data from which real focus distance can be recovered. Focused at the 3 m barrel mark, the file works out to 2.99 m. Yet none of the three RAW editors I tested reads it, and ExifTool cannot even parse it.

Why Your Software Wants to Know Where You Focused

A lens does not behave the same at every focus setting. Focus a 55mm close and it behaves measurably differently than the same glass at infinity: distortion shifts, chromatic aberration changes, and in many designs the focal length itself drifts as elements move inside the barrel, the difference sharpest in close work.¹ That is why serious correction engines are built around focus distance, not just lens identity. Adobe's lens profiles are constructed from test shots at combinations of focal length, aperture, and focus distance.² DxO calibrates its modules the same way, since distortion depends "firstly on the focal length" and "secondly on the focus distance."³ Canon, Nikon, Olympus, and Panasonic all cooperate by writing some version of that distance into their RAW metadata.⁴

Which brings us to the line in Capture One's 16.8.3 release notes that started this investigation. Explaining why its brand-new Hasselblad support applies a default lens correction rather than a precise one, Capture One wrote that the RAW file "doesn't contain this information," with a fix waiting on a future Hasselblad firmware update.⁵ That sentence sounded final. It also turned out to be more complicated than it looks.

What I Found Inside the Files

I compared the metadata of two of my own 3FR files: one shot in November 2025, one shot in June 2026. The June file contains three numbers the November file simply does not have, added without a word in any release note by firmware 1.2.7.11, whose official changelog talks about shutter sounds and rating modes, not metadata.⁶

Those three numbers appear to belong to the image stabilization system: to steady a shot properly, IBIS needs to know how the lens is focused, since the closer you focus, the more the image shifts for the same wobble of your hands. Read together, they describe the lens's focus state at capture: the effective focal length right now (which drifts as you focus, the behavior photographers call focus breathing), the magnification, and an internal measure of the optics' position.

Two of those numbers, focal length and magnification, are all you need. A formula that has lived in optics textbooks for two centuries turns them into the distance you focused at.⁷

The Test: Three Frames at Marked Distances

Claims like this need substantiation. Three frames with the XCD 2,5/55V, all at f/8, focused at three of the distances the barrel marks: 1.5 m, 3 m, and infinity.

The 3 m frame came back at 2.99 m, a third of one percent off the mark. The 1.5 m frame landed at 1.38 m, further off but in the direction close-focus optics predict, and barrel markings are themselves approximate. The quiet showstopper is infinity: recorded magnification isn't "very small," it's exactly zero, the mathematically correct value for infinity focus. (Full arithmetic in the test-data appendix.)

These are not numbers that happen to correlate with focus distance. They are focus distance, wearing a thin disguise.

What Actually Reads These Numbers?

None of the three editors I tested reads these fields. The proof is a doctored file built to fool at least one of them, and it very nearly did.

I ran the bluntest experiment I could think of: byte-identical copies of the 3 m test file, differing only in the recorded focus numbers (one doctored to claim an extreme close-up), each exported with untouched default settings and checked against a same-file rerun to establish each converter's own render-to-render noise floor.⁸ If an editor's lens correction responded to those numbers, the copies would render differently by more than that floor.

Phocus fooled me for an afternoon. My first pass showed the doctored copy rendering visibly different, vignetting off, white balance shifted, and for a while it looked like Hasselblad's own software was reading these numbers after all. It wasn't: a stray setting had attached itself to one file in that session, not the metadata, and a clean rerun with fresh copies rendered identically no matter what focus numbers each file carried. Phocus ignores these fields too. The lesson I keep relearning: one test is never enough. Full protocol and hashes are on the test-data appendix.

Capture One 16.8.3 rendered both copies identically, within its own render-to-render noise, exactly what its release notes imply. Lightroom Classic did the same, with Adobe's own 55V profile confirmed active on every export via embedded develop settings. ExifTool has no Hasselblad module at all and shows the fields as anonymous, unnamed numbers. Three editors, three confirmations: nobody reads this data yet.

Was Capture One Wrong?

Not in any way that matters. My render test backs Capture One up on its own behavior: it does exactly what its documentation says it does. The numbers exist in the file, but in a private, undocumented corner of Hasselblad's metadata, with no public specification, no field names, no guarantee they mean the same thing after the next firmware update. No vendor can build corrections on numbers like that. "The file doesn't contain this information" is misstated rather than false, and the promised firmware update is presumably about moving the number from an unlabeled drawer to a documented shelf.

What the finding does change is the timeline. Hasselblad didn't start solving this problem when the Capture One partnership was announced on July 2. The camera has been writing down focus state since November. The plumbing came first; the partnership announcement described the faucet.

What This Means for Your Photographs

Today: nothing changes, and nothing needs doing. Your lens corrections work exactly as they did yesterday. The one practical read is for close-focus shooters: macro, product, tight portraits, foreground-heavy landscapes. Capture One and Lightroom now confirm what you'd suspect: those corrections are tuned per lens but blind to distance, best at ordinary working distances and weakest up close.¹ At infinity and normal shooting distances the difference is negligible, and if you mostly shoot at a distance, this changes nothing for you.

Phocus is the interesting unknown. My test proves it doesn't read these two fields, but the camera bakes other per-shot data into the file, and I can't rule out Phocus reading some of that by another route. Whether it corrects for focus distance where the others don't is a question for a later post.

The future is more interesting. When the promised firmware update lands, and if third-party software starts consuming focus distance properly, corrections should improve most exactly where they're currently weakest. And because the X2D II has already been recording focus state for eight months, some files you've already shot could benefit retroactively.

I haven't tested whether the X2D 100C or CFV 100C write the same fields, since I don't own either; if you shoot one and want to look, the firmware changelog post covers how the X2D II versions map, and I'd like to know what you find. The color comparison I promised in the Capture One post, Phocus versus Capture One versus Lightroom on identical files with measured differences, is on my ToDo list.

References

1. Things You Should Know About Your Lenses, But May Not - Roger Cicala, LensRentals, on close-focus performance, floating elements, and focal length change with focus
2. Adobe Lens Profile Creator User Guide - Adobe's profiles are built from calibration sets at fixed focal length, aperture, and focus distance
3. Geometric Lens Distortion: How DxO corrects optical flaws - DxO on distortion's dependence on focal length and focus distance
4. ExifTool Tag Names index - per-brand maker-note documentation (Canon FocusDistanceUpper/Lower, Nikon, Olympus, Panasonic focus-distance tags; no Hasselblad module exists)
5. Capture One 16.8.3 release notes - the lens correction paragraph: default correction applied because the file "doesn't contain this information," pending future Hasselblad firmware
6. Hasselblad X2D II Firmware Updates Since Launch - the running changelog; firmware 1.2.7.11's public notes make no mention of new metadata
7. For the technically curious: distance = focal length × (1 + 1/magnification), the thin-lens magnification relation. In the test files the fields live in Hasselblad's private maker note as 8-byte floating point values (tags 0x0081, effective focal length in mm, and 0x0082, magnification); a third field (0x0083) tracks internal lens position. Values verified by controlled capture; the mapping to Hasselblad's internal field names is inferred from firmware analysis and flagged as such.
8. Behavioral test methodology in brief: copies of the same 3FR differing only in the focus-distance fields (a 16-byte edit, verified with a full-file byte comparison), each exported to 16-bit TIFF with untouched default settings, compared against a same-file rerun control to establish each editor's own render-to-render noise. The full test-data appendix (exact field values, ExifTool dumps, every hash, and the per-editor pixel-difference numbers) is published here: test-data appendix.