Ilford SFX 200

Datasheet-derived original, tier: datasheet-validated, partial-fidelity. Primary source: Ilford SFX 200 technical information, Nov 2018. Archived locally at X-T5/_reference-sources/datasheets/ilford/sfx-200.pdf, 6 pages, read in full. Settings in the notes.

What the datasheet says

Filter factors, from the sheet:

Note what the 89B row implies: at four stops of filter factor on an ISO 200 film, Ilford's own worked example is "1/30th second at f5.6" in bright sunlight, and it recommends a tripod. Real SFX at the dramatic end is a slow, deliberate process. The digital translation is not, which is worth knowing when judging whether a frame "looks like SFX."

The finding: what ACROS+R can and cannot reproduce

This is the substantive research result of this derivation, and it corrects a piece of shorthand the bank was already carrying, the notes previously logged SFX 200 as "❌ (Acros+R approximates)" That is half right, and the half it gets wrong is the more famous half.

The sky: reproducible ✅

A blue sky is strongly blue-reflecting and weakly red-reflecting. Any red filter. Optical or computational. Suppresses it. Fujifilm's manual describes the ACROS colour filters as deepening "shades of gray corresponding to hues complementary to the selected color", and ACROS+R specifically as "Enhances contrast and darkens skies considerably." This lands in the same place as SFX-plus-red-filter. No caveat needed.

The foliage: not reproducible, and the recipe goes the opposite way ❌

Ilford's second stated effect, "most green vegetation almost white" Is a near-infrared phenomenon, not a red-filter phenomenon.

1. The mechanism is the vegetation red edge. Chlorophyll absorbs strongly across the visible spectrum, including deep into the red, and then becomes almost transparent above roughly 700 nm; with nothing left to absorb the light, it scatters back out and leaf reflectance rises sharply, by a factor of five or more. Across approximately 680–750 nm. This is standard remote-sensing physics; it is the same effect that makes foliage glow white in true infrared photography (the Wood effect, named for Robert Wood).
2. SFX catches the bottom of that ramp. Deliberately, and only just. Ilford's 740 nm ceiling reaches into the rising edge but not far up it. This is exactly why SFX is described everywhere as producing an infrared look rather than true infrared, and why its foliage renders pale-grey rather than the luminous white of a genuine IR film. Ilford's own sheet is careful about this too, "unlike true infra-red film," it can be loaded in subdued light.
3. The X-T5 cannot record that band at all. Every unconverted digital camera has an IR-cut / hot-mirror filter bonded over the sensor, whose transmission falls away progressively through the 500–700 nm region and is effectively closed before the red edge begins. This is not a Fujifilm quirk: infrared-conversion services exist precisely to remove that filter, and offer the service for Fujifilm X-mount bodies.
4. So ACROS+R operates entirely on visible red, where foliage is dark. Chlorophyll's strongest absorption sits around 660 nm, right in the visible red. Healthy leaves reflect little red light. A red filter therefore darkens foliage, which is precisely what Fujifilm's manual says the filters do to complementary hues. The recipe fails to lighten the leaves and pushes them the other way.

What that means in practice

The only routes to the foliage effect are an infrared-converted body or shooting the actual film. There is no in-camera setting, no WB shift and no tone-curve move that recovers it, because the information was removed by a piece of glass in front of the sensor before any processing happened.

This is a clean, concrete instance of the bank's founding argument in the notes, that a film simulation is a spectral-sensitivity-aware conversion of the raw sensor signal, and cannot recover what the sensor never received. Usually that argument is made about colour dye layers and is hard to demonstrate. Here it reduces to a single number: 740 nm, on one side of the hot mirror or the other.

Sourcing note

The Ilford figures above are primary and archived. The red-edge physics is well-established published science, cited here from general remote-sensing and infrared-photography literature (Wikipedia's Red edge; standard treatments of the Wood effect) reached via web search. Full-text fetches of those pages returned HTTP 403 from this environment, so they are recorded at search-snippet level rather than as verified full-page reads. The hot-mirror claim comes from infrared-conversion vendors' published descriptions of stock camera filters (Kolari, Life Pixel), also at snippet level. No manufacturer publishes the X-T5's exact IR-cut transmission curve, so the recipe and this file describe the cutoff qualitatively rather than quoting a number Fujifilm has not stated. Anyone wanting to harden this: an X-T5 frame of foliage through a physical R72 filter, at a long exposure, would settle empirically how much leaks through.

The tone-curve decision

The obvious move would be to copy the bank's attributed Acros Red High Contrast, which runs Highlight —/ Shadow —on the same base. It is a good recipe. It is not this film.

Ilford describes SFX twice as having a wide tonal range and a wide exposure latitude, and calls its rendition "unusual" rather than "contrasty." SFX's drama is spectral. It comes from recording a band the eye does not see, then filtering it. Stacking a hard tone curve on top would be modelling the reference recipe's aesthetic rather than the film's data, and would flatten the very tonal range the sheet advertises.

So: Highlight —/ Shadow —. The +R filter supplies the separation; the curve stays long. The "maximum drama" variant exists for anyone who wants the 89B end, and it is labelled as a choice rather than as fidelity.

The — decision

Unusual for this bank, and worth its own note. The sheet makes no acutance or sharpness claim whatsoever. Contrast that with XP2 Super, whose product title is literally "SHARP, FINE GRAIN." What SFX's sheet does instead is warn about a focus shift: red light comes to focus at a slightly different point than the rest of the visible spectrum, and Ilford's remedy is to stop down to the smallest working aperture, noting only apochromatic designs may need no correction.

That means a genuine SFX-plus-deep-red frame is frequently marginally soft by construction. — records that. The architecture variant lifts it to +1, on the reasoning that the digital "filter" is applied after capture and therefore causes no focus shift, so on stone and glass the datasheet's own reason for restraint does not apply.

The weakest value in the recipe

Grain. Ilford publishes no granularity figure for SFX, no RMS, no grain index, nothing comparable to Acros 100's RMS 7 or the T-grain construction language in the Delta sheets. Weak, Small is placed purely by speed class: a 200-speed conventional emulsion, sitting between Delta 100 and the 400-speed originals. That is an interpolation, not evidence, and it is the first thing to change if a real scan says otherwise.

What the digital version fixes

Worth recording because it cuts the other way for once. Normally these derivations are a list of what is lost.

Shooting SFX for real means compounding two exposure problems: a filter factor of 1½ to 4 stops depending on how deep the red is, and a TTL meter that Ilford warns "can under expose by up to 1½ stops with deep red or orange filters in place." Hence the sheet's advice to bracket ±2 stops and to run a calibration roll before starting.

On the X-T5 the +R "filter" is part of the film simulation, applied to the signal after metering. The filter factor and the metering error both vanish. Meter normally, shoot at 0 to +1/3, keep the shutter speed you would otherwise have had, no tripod required for what Ilford calls a 1/30 at f5.6 in bright sun. That is a genuine, concrete advantage of the digital translation, and one of very few in this bank.

Validation status

Datasheet-validated, partial-fidelity. Every setting traces to the archived Ilford sheet. The partial-fidelity label is not hedging. It names a specific, physically-grounded, permanently-unreachable gap (foliage), while the rest of the film's documented behaviour is reproduced.

No real-film scan yet. The single most valuable frame for references/ is a red-filtered SFX exposure containing both blue sky and green foliage, which would let the caveat above be measured rather than argued, and would also settle the grain question in the same frame.

Derived: 2026-07-25