How to use this document
Use this when field conditions require specialist technique or failure-mode handling.
Show companion links
Part I
The Physics — What Bayer Interpolation Actually Is
Every digital sensor — including the GFX 100S II's 102 MP BSI-CMOS — is monochromatic. It measures light intensity, not colour. To produce colour images, each photosite is covered by a coloured filter that passes only one of three wavelength bands: red, green, or blue. The pattern of these filters — two green, one red, one blue per group of four photosites — is called the Bayer pattern, after Bryce Bayer who patented it at Kodak in 1976.
The consequence: at any given pixel location in the final image, only one colour channel is actually measured. The other two are estimated by interpolating from neighbouring photosites of the appropriate colour. This estimation is called demosaicing. Every RAW processor — Lightroom, Capture One, Fujifilm's own software — performs it. The algorithms are sophisticated and generally excellent, but they remain mathematical approximations of values that were never physically measured.
Standard single-shot capture
One channel measured, two estimated
At a red photosite, red is measured. Green and blue are calculated from the nearest green and blue neighbours — typically 1–2 photosites away. At 102 MP with 3.76 micron photosites, the neighbours are extremely close. But they are still from a different physical location. In smooth gradients and most midtone regions, interpolation is essentially invisible. At high-frequency detail — rock texture, fabric weave, bark, feathers — interpolation introduces subtle colour averaging that reduces the perceived sharpness of fine structure. In black-and-white conversion with aggressive channel mixing, it multiplies estimation errors into tonal inaccuracies. The approximation is small. At 102 MP and large print scale, small matters.
Pixel Shift capture
All three channels measured at every pixel
The IBIS mechanism moves the sensor by precisely half a photosite between exposures. After four frames — each capturing the scene with the sensor offset by half a step in X and Y — every pixel location has been covered by a red, a green, and a blue photosite. No interpolation is required. The Fujifilm Pixel Shift Combiner software stacks these four frames and produces a file where every pixel has actual measured RGB values. The result: zero colour moiré (because no frequency aliasing occurs without interpolation), maximum colour accuracy, and perceptibly finer rendering of high-frequency surface detail. At the extreme luminance transitions that define geological and architectural subjects — a dark crack in white rock, a shadow edge on a summit face — Pixel Shift eliminates the microscopic colour fringing that interpolation introduces.
The monochrome argument — why B&W benefits more than colour
In a colour photograph, interpolation errors are partially masked by the colour information itself. The eye is more tolerant of slight colour inaccuracy than of luminance inaccuracy. In a black-and-white conversion, all colour channel information collapses into luminance — and the weighting applied during conversion amplifies any channel errors. A standard Lightroom Black & White Mix with Blues pushed to −50 (darkening a sky) multiplies the interpolation uncertainty in the blue channel across every pixel in that tonal zone. Pixel Shift removes the uncertainty at source. The channel mixer then operates on actual measurements.
For geological subjects — rock, snow, ice, stone — where the tonal rendering is everything and where the photographer is often using aggressive channel mixing to separate tonal zones, Pixel Shift is not a refinement. It is the foundation the tonal work rests on.
Part II
4 vs 16 Frames — Choosing the Right Mode
The GFX 100S II offers two Pixel Shift modes. The choice between them is not about one mode being universally superior — it is about whether your scene and environment can sustain 16 frames of absolute stillness. The 16-frame mode can deliver more detail and lower random noise, but the 4-frame mode is often the more robust option in real locations.
Mode 1
4
PIXEL SHIFT ACCURATE COLOR
Four frames, sensor shifted by half a photosite in each cardinal direction. Provides full RGB measurement at every pixel location — zero colour interpolation, zero colour moiré. Output resolution remains 102 MP; the gain is colour accuracy and micro-detail precision, not pixel count.
The four frames are captured in approximately 1–2 seconds at minimum interval. The shorter capture window means less time for vibration to develop between frames — making this mode more robust in environments that are stable but not perfectly controlled.
Use when: the primary goal is colour accuracy and moiré elimination; the environment is stable but not laboratory-controlled; the subject is architectural or fine art reproduction where colour fidelity matters as much as resolution.
The robust choice
Mode 2 · Preferred
16
PIXEL SHIFT HIGH RESOLUTION + ACCURATE COLOR
Sixteen frames captured across a denser sub-pixel grid. Delivers everything Mode 1 provides, plus: dramatically reduced noise (16-frame averaging suppresses random noise by approximately 4× — equivalent to two full stops of ISO improvement); effectively ~400 MP output resolution after combination, because the sub-pixel sampling creates spatial information beyond the sensor's native photosite pitch.
The sixteen frames take approximately 4–8 seconds at minimum interval. This longer window demands a more controlled environment. Any vibration — road traffic felt through the floor, a passing person, wind on the camera strap — introduces misalignment between frames that the combiner software cannot fully correct.
Use when: conditions are fully controlled (solid tripod, isolated from vibration sources, subject completely static, indoor or windless environment); the output will be printed at large scale where both the noise reduction and the super-resolution gain are visible.
Maximum quality
◈
The 16-frame mode and the Matterhorn. Shooting that scene from a fixed viewpoint — on a solid platform, windless day, no passing vehicles or people — is a 16-frame scenario. The subject will not move in 8 seconds. The sky will not move meaningfully. The output will be approximately 400 MP of colour-accurate data with noise equivalent to ISO 25. That is a different class of image from any single-shot capture at any ISO.
Part III
Subject Selection — What Earns C6
Pixel Shift has one absolute requirement that cannot be worked around: the subject must not move between frames. Any movement — of the subject, of the camera, of the air between lens and subject — causes misalignment that the combiner cannot repair. This constraint defines the subject pool. Within that pool, some subjects benefit far more than others from the colour and resolution gain. The subjects worth reaching for C6 are the intersection of both lists.
← Ideal subjects
Geological — rock face, mountain, cliff, stone
An excellent Pixel Shift subject: completely static, high surface complexity, and strong tonal range in monochrome conversion. The micro-detail that Pixel Shift preserves — individual crystal facets, frost patterns, shadow cracks — is precisely what makes a geological print worth close examination at large scale.
← Ideal subjects
Architectural — façade, interior, structure
Stone, concrete, brick, steel — all completely static, all high spatial frequency. Interior architecture with controlled light is the most reliable Pixel Shift environment: no wind, no vibration, controlled temperature. Colour accuracy in architectural photography is also a deliverable — Pixel Shift's actual RGB measurements eliminate colour casts from interpolation that affect how materials render.
← Ideal subjects
Fine art reproduction — print, artwork, textile
Reproducing your own prints for archive or client proofing; documenting textile, ceramic, or surface work. Flat subject, controlled studio light, absolute stillness. Pixel Shift eliminates the colour moiré that standard Bayer capture can produce when photographing fine woven textures — a failure mode that can contaminate print reproduction quality.
← Ideal subjects
Product and still life — object on table
Any object that can be placed on a stable surface with controlled light. Jewellery, ceramics, books, small objects of material interest. The colour accuracy gain is particularly valuable for products where faithful colour rendering is a commercial requirement.
← Poor candidates
Any landscape with foliage, water, or sky movement
Wind moves foliage between frames. Water surface moves constantly. Fast-moving clouds produce ghosting. Even in calm conditions, slow air currents move leaves. If foliage or water is in the frame, the combiner will show misalignment in those zones. A 4-frame sequence in very still conditions may be attempted with a short interval, but accept that those zones may require patching from a single-shot frame in post.
← Poor candidates
Portraiture, people, animals
Any living subject breathes, blinks, shifts, and moves on timescales shorter than a 4-frame sequence. The GF 80mm's Pixel Shift role is for controlled still-life and product work, not portraits of people. Pixel Shift is not a practical technique for people.
Part IV
The Non-Negotiables — Conditions That Cannot Be Compromised
Unlike focus stacking — which degrades gracefully with imperfect conditions and can be partially salvaged in post — Pixel Shift failure is binary. If a condition is violated, the output is corrupted in a way the combiner cannot repair. These are the conditions. All of them must be met simultaneously.
◉
Absolutely static subject
Zero movement at any part of the subject for the full duration of the sequence. This means no foliage, no water surface, no fabric, no suspended elements. Pure stone, metal, and controlled objects only.
◉
Rock-solid tripod
Ball head locked. Centre column down. Legs spread wide. Tripod feet on solid substrate — stone, concrete, solid wood floor. Not soft earth, not loose gravel, not carpet over a hollow floor above a road.
◉
Remote release or self-timer
Never press the shutter button directly. The RR-100 cable release fires the sequence without touching the camera. The 2-second self-timer achieves the same result. Any physical contact with the camera body during sequence capture contaminates it.
✗
No vibration in the environment
This is the hardest condition to meet and the most common source of failure. Road traffic vibration transmitted through floors. HVAC systems. Nearby construction. Other people walking on the same floor or building. Test before shooting — see Part V.
✗
IS MODE OFF — both camera and lens
The IBIS mechanism is being used by the Pixel Shift system itself to move the sensor between frames. Any camera-side stabilisation conflicts with and corrupts the pixel-shift motion. Camera IS MODE: OFF. GF 45-100mm / GF 100-200mm F5.6 barrel OIS switch: OFF when either OIS zoom is mounted. Mandatory.
✗
Stable continuous light only
Do not build
C6 around flash recycle. Pixel Shift needs consistent frame-to-frame illumination and electronic-shutter discipline. Use daylight, continuous LED, tungsten, or another stable continuous source. See the Flash and Continuous Light reference for the system-wide rule.
Part V
Vibration Testing — The Protocol Before Every Session
Environmental vibration is invisible to the eye but fatal to Pixel Shift. A building that feels perfectly still has floor resonance from HVAC systems, foot traffic from adjacent spaces, and traffic-induced micro-tremor from roads hundreds of metres away. You cannot assess vibration by standing still and feeling it. You test it with the camera.
The test takes less than five minutes and should be performed at the actual shooting position, with the tripod set up as it will be for the session, before you set up lighting or compose the shot.
01
Set up the camera on tripod — do not touch after setup
Position the tripod at your intended shooting position. Lock the ball head. Attach the remote release. The camera must be in the same physical configuration it will be in for the actual sequence — same lens, same load on the tripod, same extension.
02
Select a high-contrast reference target
Frame a static reference element with a sharp, high-contrast edge — the edge of a white card against a dark background works well; the edge of an object in your scene is equally valid. The edge must be vertical or horizontal, and the frame must include a region of fine detail. This gives you the sharpest possible diagnostic signal for misalignment.
03
Shoot a 16-frame sequence at minimum interval
Set the camera to PIXEL SHIFT HIGH RESOLUTION + ACCURATE COLOR (16 frames), minimum interval. Fire the sequence via remote release. Do not touch anything during the sequence. Do not speak. If other people are in the space, ask them to stand still.
04
Process the test sequence in Pixel Shift Combiner
Open the test frames immediately in Fujifilm Pixel Shift Combiner. Process with default settings. The combiner reports alignment success or failure per frame. If all 16 frames align successfully, the environment is suitable for Pixel Shift. If frames fail to align — particularly frames 8–16, which are captured later and therefore more vulnerable to cumulative vibration — the environment has too much vibration.
05
If the 16-frame test fails — try the 4-frame mode
If the 16-frame sequence shows alignment failures, try PIXEL SHIFT ACCURATE COLOR (4 frames). The shorter capture window (approximately 1–2 seconds) is more robust against environmental vibration. If the 4-frame sequence also fails, the environment is unsuitable for Pixel Shift at this session. This is not a failure of the technique — it is information. Come back when the building is quieter, or work in a more isolated space.
All frames aligned
Proceed with 16-frame
Environment is suitable. Set up final composition and shoot the actual sequence.
Partial alignment — late frames fail
Try 4-frame mode
Low-level vibration building over time. 4-frame sequence within the stable window. Or: wait until vibration source is inactive (late night, closed building).
4-frame also fails
Abort Pixel Shift
Shoot the scene as a standard single-frame at ISO 100 on
C1. Return when conditions allow. Do not attempt to force a result from a vibration-contaminated sequence.
Part VI
Lens Selection — Which GF Lens for Which Subject
Pixel Shift performance is sensitive to lens optics in a way that standard single-shot capture is not. Because the system uses the IBIS mechanism to move the sensor by precisely half a photosite between frames, and the combiner then aligns those frames using the optical image, any lens characteristic that reduces the precision of that alignment — field curvature, strong distortion, optical vignetting — works against the technique. The GF lens lineup behaves very differently in this context.
Optimal · Primary
C6 lens
GF 80mm F1.7
Fujifilm's own guidance identifies the 80mm as the Pixel Shift prime. The reasoning is specific: flat field at working apertures — the focal plane is flat across the frame, not curved, which means the combiner's pixel-level alignment works correctly from centre to corner. Optical symmetry — the lens is well-corrected with low distortion, so there is no geometric warping that could cause misalignment between the shifted frames. The focal length — at 80mm, the relative parallax shift between near and far elements from the half-photosite sensor movement is very small. At 20mm, that same physical movement produces a proportionally larger angular shift, causing near-field elements to shift relative to the background. Shoot at f/8 for peak sharpness and maximum LMO correction benefit. Avoid f/1.7 for Pixel Shift — the field flatness is best at f/4–f/8.
Selective — use from 60mm upward
GF 45–100mm F4 OIS
The 45-100mm is usable for Pixel Shift at longer focal lengths on flat or distant subjects. At 60–100mm on a flat architectural façade or distant mountain face — where there is no near-field foreground that could shift relative to the background — it performs well. At 45–55mm with any near-field element in the frame, the parallax from the sensor shift is large enough to cause misalignment in foreground elements while the background aligns correctly. The combiner cannot resolve two different alignment solutions for the same frame. Critical: confirm both the barrel OIS switch AND camera IS MODE are OFF before every sequence with this lens. Neither OIS system can be active when IBIS is managing the Pixel Shift motion.
Prudent — distant static subjects
GF 100-200mm F5.6
The native 100-200mm earns its place in
C6 when the subject is distant, static, and benefits from stronger compression than the 45-100 can provide.
Use it for remote façades, mountain planes, isolated ridgelines, or compressed landscape details where every important element sits far enough away that parallax between frames remains negligible. It is not a near-field Pixel Shift lens — if there is foreground depth close to the camera, the longer focal length does not rescue a bad geometry problem.
Critical: as with the 45-100mm, confirm both the barrel OIS switch AND camera IS MODE are OFF before every sequence.
Not recommended
GF 20–35mm F4
The wide-angle nature of this lens makes Pixel Shift unreliable for any scene with depth. At 20mm, the half-photosite sensor movement produces a proportionally large angular displacement of near-field elements against the background. The combiner sees two different geometric relationships between foreground and background across frames and cannot align both simultaneously. The only valid Pixel Shift use for the 20-35mm is
entirely flat subjects at a single optical distance — artwork reproduction, a flat wall, a map. For any landscape scene with any depth, this lens is a
C1 instrument, not a
C6 one. Reach for the 80mm, the 45-100mm at longer focal lengths, or the GF 100-200mm F5.6 for genuinely distant static compression instead.
Part VII
C6 Bank Configuration — Complete Camera Setup
The C6 bank stores the complete camera state for Pixel Shift sessions. The Drive Mode (Pixel Shift Multi Shot) is set on the DRIVE button and saved to the bank. The mode dial sits in M (Manual). All settings below must be confirmed before saving.
H · IQ pages 1-4/4
H IMAGE QUALITY SETTING — C6 values · STILL PHOTO DISPLAY → MENU/OK → H IMAGE QUALITY SETTING
6 settings
| # | Setting | Value | Action |
|---|
| 1 | IMAGE QUALITY | RAW | Set |
| Universal. Each of the 4 or 16 individual frames is written as a full RAF file. The combiner software processes these RAF files directly. |
| 2 | RAW RECORDING
Two sub-settings: compression + bit depth
| LOSSLESS · 16-bit | Set |
| Storage planning required. A single 16-bit lossless RAF from the GFX 100S II is approximately 100–130 MB. A 16-frame Pixel Shift sequence therefore produces 1.6–2 GB of source files before combination. Budget accordingly — fast, high-capacity media is essential. Use fast UHS-II cards; V90 is recommended headroom rather than a hard camera requirement. |
| 3 | FILM SIMULATION | PROVIA / STANDARD | Set |
| 4 | TONE CURVE | H: 0 · S: 0 | Set |
| Zero tone curve for Pixel Shift, same as C1. The combiner software processes the RAW data with neutral preview parameters — any in-camera tone modification affects the JPEG preview reference used for frame alignment checking. Keep it neutral. |
| 5 | HIGH ISO NR | −4 | Set |
| 6 | AUTO UPDATE CUSTOM SETTING | DISABLE | Critical |
| Universal critical. DISABLE protects the C6 bank from accidental overwriting. |
G · AF/MF pages 1-3/3
G AF/MF SETTING — C6 values · STILL PHOTO DISPLAY → MENU/OK → G AF/MF SETTING
8 settings
| # | Setting | Value | Action |
|---|
| 1 | AF MODE | SINGLE POINT | Set |
| 2 | NUMBER OF FOCUS POINTS | 425 POINTS | Set |
| 3 | PRE-AF | OFF | Set |
| 4 | FACE/EYE DETECTION | OFF | Set |
| 5 | SUBJECT DETECTION | OFF | Set |
| 6 | FOCUS CHECK | ON | Set |
| For Pixel Shift work at f/8, confirm focus precision before entering Pixel Shift drive mode. Focus, verify at 1:1, then switch drive mode to Pixel Shift and fire. Do not rely on AF confirmation alone — at 102 MP the depth of field at f/8 is generous, but for Pixel Shift you want the sharpest possible focus on the critical surface. |
| 7 | RELEASE / FOCUS PRIORITY (AF-S) | FOCUS | Set |
| Focus priority for Pixel Shift, same as C1. The shutter must not release until focus is confirmed. A soft Pixel Shift sequence is uniquely wasteful — 16 frames of an unsharp subject cannot be recovered. |
| 8 | TOUCH SCREEN MODE | OFF | Set |
A · Shooting pages 1-3/3
A SHOOTING SETTING — C6 values · STILL PHOTO DISPLAY → MENU/OK → A SHOOTING SETTING
7 settings
| # | Setting | Value | Action |
|---|
| 1 | SHUTTER TYPE | ELECTRONIC — always | Set |
| Electronic Shutter is mandatory for Pixel Shift. The mechanical shutter's curtain travel introduces vibration at the moment of exposure. For a single frame, this is managed with a self-timer. For a 16-frame sequence at minimum interval, the mechanical shutter fires 16 times — 16 moments of mechanical shock, each contaminating the frame that follows. The Electronic Shutter fires by reading the sensor progressively with no moving parts — zero mechanical input throughout the sequence. This is the correct and only choice. |
| 2 | SELF-TIMER | 2s · or use RR-100 | Set |
| The 2-second self-timer allows camera vibration from the shutter button press to fully dissipate before the first frame fires. The RR-100 remote release is superior — it eliminates the button press entirely. In either case, once the sequence starts, the camera fires all remaining frames automatically without further input. |
| 3 | SAVE SELF-TIMER SETTING | ON | Set |
| 4 | ISO | 100 (fixed) | Set |
| 5 | IS MODE | OFF — mandatory | Set |
| The IBIS system is occupied managing the Pixel Shift sensor movement. If IS MODE is ON simultaneously, two systems are sending conflicting instructions to the same IBIS actuator. The sensor cannot move to the correct half-photosite position while also trying to compensate for camera motion. The frames will not align. The sequence is destroyed. IS MODE must be OFF and confirmed OFF before every C6 session. Also confirm the barrel OIS switch is OFF if the GF 45-100mm or GF 100-200mm F5.6 is mounted. |
| 6 | FLICKER REDUCTION | OFF (studio) / ALL FRAMES (artificial light) | Set |
| 7 | 35mm FORMAT MODE | OFF | Set |
DRIVE button
DRIVE BUTTON — set before saving bank · DRIVE BUTTON (top plate) → PIXEL SHIFT MULTI SHOT
1 settings
| # | Setting | Value | Action |
|---|
| 1 | PIXEL SHIFT MULTI SHOT
Choose mode within this option
| Select mode, then set INTERVAL | Set |
| After selecting PIXEL SHIFT MULTI SHOT from the drive menu, you will be prompted to select between ACCURATE COLOR (4 frames) and HIGH RESOLUTION + ACCURATE COLOR (16 frames). Choose your mode for this session. Then set INTERVAL to the shortest available option for still-subject available-light or continuous-light work. This drive mode state is saved to the C6 bank when you perform EDIT/SAVE CUSTOM SETTING. |
◈
Save the bank: with all settings configured and the mode dial at M with your working exposure (ISO 100, f/8, shutter determined by scene), navigate to H IMAGE QUALITY SETTING → Page 3/4 → EDIT/SAVE CUSTOM SETTING → C6 → SAVE CURRENT SETTINGS. Then EDIT CUSTOM NAME → C6 → PIXEL SHIFT. Verify with EDIT/CHECK: confirm IS MODE = OFF, SHUTTER TYPE = ELECTRONIC, AUTO UPDATE CUSTOM SETTING = DISABLE.
Part VIII
Field and Studio Execution — The Complete Sequence
01
Assess and prepare the environment
Before placing the tripod, consider the floor or ground substrate. Solid stone, reinforced concrete, and solid wood floor are good. Suspended wooden floors, carpet, soft earth, and loose gravel are problematic. If shooting outdoors on rock — a mountainside position, a stone platform — the substrate is excellent. If shooting indoors, identify and mentally note the vibration sources: is there an HVAC unit nearby? What floor level are you on? Is there traffic below? These questions inform your confidence before the vibration test.
02
Set up tripod with maximum stability
Spread legs to the maximum stable width. Centre column fully retracted — never extended for Pixel Shift. Ball head locked with zero play. If the tripod has a hook below the centre column, hang a weight from it (your camera bag, a sandbag) to lower the centre of gravity and damp resonance. Ensure the camera strap is not hanging free where wind could move it against the camera body during the sequence.
03
Compose and focus — before entering Pixel Shift drive mode
Do all your compositional work in normal shooting mode (
C1 or just M mode directly). Set your aperture (f/8), focus precisely using AF-S Single Point, verify with Focus Check magnifier, confirm IS MODE OFF and (if using the GF 45-100mm or GF 100-200mm F5.6) barrel OIS switch OFF. Check the RGB histogram for exposure — expose for ETTR, confirming no channel clipping. Do everything while the drive mode is still SINGLE FRAME. The act of switching to Pixel Shift drive mode is the final step before firing.
04
Run the vibration test
Fire a test sequence as described in Part V. Process in Pixel Shift Combiner. Confirm all frames align. If they do, proceed. If not, diagnose and resolve before committing to the session sequence.
05
Switch to Pixel Shift drive mode
Press the DRIVE button → PIXEL SHIFT MULTI SHOT → select your mode (4 or 16 frames) → confirm INTERVAL is set to shortest. The camera is now in Pixel Shift mode. The EVF will show a Pixel Shift indicator.
06
Eliminate all human-caused vibration
If other people are in the space: ask them to stand completely still during the sequence. If you are in a shared environment where this is not possible, wait for a naturally quiet moment. In a studio session: inform assistants before firing. In an outdoor location: wait for any wind gust to pass. In a building: if you hear footsteps on the same floor level, wait until they have moved further away. These are not excessive precautions — they are the practical discipline that separates clean Pixel Shift output from a failed sequence.
07
Fire via remote release or self-timer
Using the RR-100 remote or the 2-second self-timer, initiate the sequence. Do not touch the camera, the tripod legs, the table or floor the tripod is on, or anything physically connected to the setup during the sequence. With the 2-second timer: press the button, step back, do not breathe heavily in the direction of the camera. The camera fires all frames in the sequence and completes automatically. With the remote: hold the release button until the sequence completes, or use a locking cable release that holds the button depressed for the duration.
08
Verify on the camera before moving anything
After the sequence completes, review the individual frames on the camera back. Check that all frames have the same exposure, that no frame appears blurred, and that the frame count matches your expected sequence length. Do not move the tripod or the subject until you have processed at least the 4-frame sequence in Pixel Shift Combiner and confirmed alignment success. If the test fails, reshoot from the same position before the setup is disturbed.
Part IX
Fujifilm Pixel Shift Combiner — The Software Round-Trip
The Pixel Shift Combiner is a separate, free application from Fujifilm. It is not Lightroom, Photoshop, or Capture One — it is a dedicated processing tool that reads the individual RAF files, aligns them at the sub-pixel level, performs the colour-accurate combination, and outputs a single combined DNG RAW file. Lightroom cannot open the individual Pixel Shift RAFs and produce a correct combined output; the workflow requires this dedicated round-trip. Download from: fujifilm-x.com/products/software/pixel-shift-combiner/.
Step 1
Import the sequence into Pixel Shift Combiner
Open Fujifilm Pixel Shift Combiner. Drag the folder containing your Pixel Shift RAF files onto the application window, or use File → Open. The combiner identifies the sequence automatically based on the embedded sequence metadata in each RAF file — you do not need to manually specify which files belong to which sequence. Important: do not move, rename, or reorganise the RAF files from their original folder structure before importing. The combiner relies on sequential file naming and metadata to group frames correctly.
Step 2
Review the alignment result
After loading, the Combiner performs alignment analysis on all frames. The result is displayed per-frame — each frame is shown as either successfully aligned or failed. A single failed frame in a 16-frame sequence does not necessarily abort the output — the Combiner can exclude failed frames and produce a result from the remaining aligned frames. However: if more than 3–4 frames fail alignment, the sequence is contaminated and should be reshot. The output quality degrades progressively with excluded frames, and a 10-frame aligned output from a 16-frame sequence is meaningfully worse than a clean 16-frame result. Re-examine your vibration environment before the next attempt.
Step 3
Process the sequence to a combined DNG
The Combiner's RAW output is a combined DNG file. That DNG can be opened in Lightroom Classic and handled like any other high-bit-depth RAW master for tone, color, sharpening, noise reduction, and output work. If you need TIFF, generate it later from the DNG in your RAW software rather than treating TIFF as the primary archive master.
Step 4
Process and export the combined file
Click Process (or the equivalent in your Combiner version). Processing time for a 16-frame 102 MP sequence is typically several minutes on a modern workstation with sufficient RAM. The Combiner is RAM-intensive — 32 GB or more is recommended for 16-frame sequences. The combined DNG is written to your specified output folder, and derivative TIFFs can be exported later if needed.
◈
Keep the source RAFs permanently. The combined DNG is derived from the source RAFs — if Fujifilm updates the Pixel Shift Combiner with improved algorithms, you can re-process the original source files and potentially achieve better output. Treat the source RAF sequence as the archive master, not the derived DNG. Archive at least one set of source RAFs per subject on a separate drive.
Part X
Back into Lightroom — Processing the Combined File
The combined DNG from Pixel Shift Combiner is processed in Lightroom much like any other RAW file, with one important difference. Because the source frames were combined from 4 or 16 exposures with measured RGB data across the sequence, the file has different characteristics from a standard single-shot RAF: lower random noise, higher colour accuracy, and finer renderable detail at fine spatial frequencies. Start with lighter sharpening and noise reduction than you would use for a normal RAF.
Menu
Lightroom Processing Adjustments Specific to Pixel Shift Output · Develop module — adjustments relative to single-shot defaults
4 settings
| # | Setting | Value | Action |
|---|
| 1 | Noise Reduction — Luminance | Start at 0 | Set |
| The 16-frame combination produces noise equivalent to ISO 25 — you will see almost no luminance noise at 1:1 review at ISO 100. Applying default or aggressive Luminance NR to a 16-frame Pixel Shift output is smoothing a file that does not need it, at the cost of micro-detail. Start at 0 and add only what inspection at 1:1 shows is necessary. The 4-frame output has slightly more noise — start at 5–10 rather than 0. |
| 2 | Sharpening — Amount and Detail | Reduce from defaults | Set |
| Lightroom's default sharpening (Amount 40, Radius 1.0, Detail 25) is aimed at normal single-shot files that have already been through interpolation. The Pixel Shift combined file has inherently sharper micro-detail because it is not interpolated. Applying the same sharpening can introduce visible over-sharpening halos at fine edges. Start at Amount 20–25, Radius 0.8, and increase only if the 1:1 review shows the image needs more. Output sharpening for print can remain at your standard settings. |
| 3 | Black & White Mix — channel confidence | Work with confidence | Set |
| This is where the Pixel Shift investment pays its return. The channel values in the combined DNG are measurements derived from the full multi-shot sequence rather than ordinary single-frame Bayer estimation. When you push Blues to darken a sky, you are working with materially cleaner color data. The tonal transition from deep black sky to lit rock is rendered from richer source information than a standard single-shot frame can provide. |
| 4 | Colour Rendering (if not converting to B&W) | Use Fujifilm profile | Set |
| For colour output, ensure the Lightroom camera profile matches your Film Simulation setting in C6 (Provia/Standard by default). The combined DNG behaves like a RAW master inside Lightroom. The visual difference in colour output from a Pixel Shift combined file versus a single shot is most visible in saturated subjects with high spatial frequency — fabric weave, fine painted surfaces, glazed ceramics — where colour moiré in standard capture produces subtle false colour that Pixel Shift largely suppresses. |
Part XI
Failure Modes — What Goes Wrong, Why, and Whether It Can Be Fixed
Pixel Shift failures are almost always total rather than partial. Unlike focus stacking — where a problematic zone can often be corrected by painting an alternative source frame — Pixel Shift misalignment typically corrupts the output file in a way that is not correctable in post. The remedies are almost all "reshoot." Understanding the specific failure signature for each mode helps you diagnose immediately and decide whether a reshoot is worth attempting at the same session.
✗ Environmental vibration — frame misalignment
Signature: the Pixel Shift Combiner reports one or more frames as failed alignment; the combined output shows fine doubling or fringing at all high-contrast edges throughout the frame — it looks like a slightly out-of-register print at the sub-pixel scale. This failure is often invisible at screen zoom and only visible at 1:1 or in print.
It is one of the reasons the vibration test protocol exists — this failure is undetectable on the camera back review.
Fix: Always the vibration test first. If a sequence fails post-session: reshoot when the environment is quieter (late night, building closed). Try the 4-frame mode as a fallback. There is no post-processing remedy for vibration-induced misalignment.
✗ IS MODE active during sequence
Signature: systematic misalignment across all frames — the Combiner fails to align any of them, not just some. The combined output (if the Combiner produces one at all) shows severe doubling at every edge in the frame, not localised to one area. The failure is uniform because every frame was affected by the IS/IBIS conflict.
Fix: Reshoot with IS MODE confirmed OFF. Also check the barrel switch if using the GF 45-100mm or GF 100-200mm F5.6. Add IS MODE verification to your physical pre-shoot checklist so this failure cannot occur. It is entirely preventable.
✗ Subject movement — ghosting or doubling in the moving zone
Signature: the combined output shows clean alignment everywhere except one specific zone — where the subject moved. Foliage shows doubled leaves. A flag reads as a ghost. Distant water has smeared texture. The Combiner aligns the static parts of the frame correctly and shows the conflict between frames in the moving zones.
Fix: If the moving zone is small and peripheral: use the Combiner's retouching tool (if available in your version) to exclude the conflicting frames from the moving zone and source from the single cleanest frame there. If the moving zone is central or large: the sequence is unusable for Pixel Shift. Shoot a single frame at ISO 100 at the best available moment of stillness.
✗ Near-field parallax at wide angles — foreground misalignment
Signature: the background aligns perfectly; the foreground elements show fringing and doubling. The Combiner cannot resolve two different geometric relationships in the same frame — it chooses the background alignment and the foreground is left misaligned. Most common with the GF 20-35mm at wide angles with any near-field subject.
Fix: Switch to the GF 80mm f/1.7, the GF 45-100mm at 60mm+, or the GF 100-200mm F5.6 for distant static subjects with depth. Restrict the GF 20-35mm's Pixel Shift use to flat subjects at a single optical distance. The parallax from the half-photosite shift is a physical consequence of the focal length and subject distance — it cannot be corrected in software.
✗ Inconsistent illumination across frames
Signature: individual frames in the sequence have visibly different exposures or color balance. The Combiner produces a combined file with uneven tonal rendering because it is blending frames that were not illuminated consistently.
Fix: Use stable continuous light: daylight, continuous LED, tungsten, or another source that remains consistent across the full sequence. Do not treat flash recycle as the normal
C6 solution; if the subject requires supplementary light, light it continuously or leave Pixel Shift.
✗ Combiner not recognising the sequence — file management error
Signature: the Combiner opens the files but cannot identify them as a Pixel Shift sequence, or identifies only some frames, or reports an error. Usually happens after files have been moved, renamed, or imported through software that strips metadata.
Fix: Feed the original, unmodified RAFs directly to Pixel Shift Combiner from a clean copied folder. Avoid renaming, exporting, or running the source frames through other RAW software before combining. The safest workflow is: card → copied source folder → Combiner → combined DNG → Lightroom.
◈
The correct import workflow. Memory card → copy source RAFs to working folder → open in Pixel Shift Combiner → process → output combined DNG to output folder → import that output into Lightroom. Do not rename source RAFs before Combiner processing, and do not run the source frames through any RAW converter before combining. The source files are inputs to the Combiner first, then the combined DNG becomes the Lightroom master.
⚠
The summary constraint. Pixel Shift is a studio-grade precision instrument that occasionally escapes into the field when conditions are right — a perfectly still mountain face from a solid position on a windless day is a legitimate field scenario, as the Matterhorn argument demonstrates. But it should be entered with eyes open: the failure modes are binary, the required conditions are strict, and the software round-trip adds workflow complexity that standard shooting does not. When the conditions are right and the subject earns it, the output is beyond what any single-shot capture can produce. That gap — between what Pixel Shift delivers and what single-shot delivers on the subjects where it matters — is the reason C6 exists in the bank architecture.
