AddendumDocument 3 of 21From the GFX 100S II field guide
Focus Stacking —
The Complete Practice
At 102 megapixels, diffraction is not a theoretical concern — it is the most common silent IQ failure in landscape photography. Focus stacking is the solution: shoot at the aperture where the lens is optically perfect, then extend depth of field in post by blending multiple focus planes. This addendum covers the physics, the camera workflow, the field execution, and the complete post-processing chain.
Working architecture details instead of landscape depth ranges? Switch to the C5 Architecture Focus Stacking Addendum for flat-plane geometry decisions, then run the on-location C5 Focus Stacking Field Checklist for capture control.
How to use this document
Use this when field conditions require specialist technique or failure-mode handling.
Show companion links
The Physics of Diffraction at 102 MP
Every photographic lens has two aperture enemies: at wide apertures, optical aberrations (coma, astigmatism, chromatic aberration) soften the image. At small apertures, diffraction does. Between these two extremes is a sweet spot — the aperture range where the lens resolves its maximum detail. On most camera systems, this sweet spot is wide enough to encompass f/8 and f/11 simultaneously, and stopping down to f/16 for depth of field costs relatively little.
The GFX 100S II changes this calculation entirely. At 102 megapixels, the photosite pitch is approximately 3.76 microns. Diffraction — the bending of light as it passes through a small aperture — begins to produce a blur circle larger than a single photosite at f/10 to f/11. By f/16, diffraction softening is visible as a collapse of micro-detail. By f/22, the image is measurably softer than f/8. The camera resolves so much that it resolves the problem too.
The traditional landscape technique — stop down to f/16 for hyperfocal depth of field — trades the sensor's greatest strength (resolution) for depth of field. Focus stacking eliminates this trade entirely.
The aperture map for the GFX system
What diffraction actually looks like — and why it hides
Diffraction on the GFX 100S II does not look like blur. It looks like a loss of texture crispness — foliage that should show individual leaves reads as a slightly smeared mass; rock surfaces that should show grain structure look smooth and slightly plastic; fabric weave collapses from distinct threads to a flat tone. The rear LCD review does not reveal this. It survives scrutiny at 100% on a laptop screen. It fails, definitively, when you open the file on a 27" calibrated display at print resolution.
The correct test: open a landscape frame shot at f/16 and a focus-stacked composite of the same scene shot at f/8, both at 1:1 on your calibrated display. The difference in micro-detail quality is immediate and significant. This is not a marginal improvement — it is the difference between a print that rewards close examination and one that plateaus early.
When to Stack — and When Not To
Focus stacking is not the answer to every landscape. Knowing when the scene calls for it — and when a single frame at f/8 already covers the required depth — is part of the field discipline. Unnecessary stacking adds post-processing time and introduces alignment artefacts where none were needed.
Camera Setup — Focus BKT on the GFX 100S II
The camera has two Focus Bracketing modes: AUTO and MANUAL. AUTO is the correct default for landscape work — you define the near and far limits of the desired sharp zone, and the camera calculates the number of frames and step size automatically. MANUAL gives you explicit control when AUTO produces too many frames, too few, or when you are working in a situation where the depth range cannot be easily defined by pointing at two reference points.
Activating Focus Bracketing
Focus Bracketing is accessed via the DRIVE button on the top plate — the same button used to select Continuous, Bracketing, or Pixel Shift. It does not live in the shooting menu directly; it is a drive mode. The settings for it live at A SHOOTING SETTING → FOCUS BKT SETTING.
| # | Setting | Value | Action |
|---|---|---|---|
| 1 | FOCUS BKT MODEAUTO vs MANUAL — the primary choice | AUTO (default) | Set |
| AUTO: You set Point A (near focus limit) and Point B (far focus limit) in the field. The camera calculates the number of frames and the step size between each frame based on your current focal length, aperture, and the distance between A and B. This is the fastest and most reliable method for landscape work — the camera's calculation accounts for depth of field per frame at your chosen aperture better than manual estimation. MANUAL: You specify FRAMES (number of shots), STEP (size of focus change per shot), and INTERVAL (time between shots). Use MANUAL when: (a) AUTO consistently produces more frames than you need for a known scene type; (b) the scene depth extends to infinity and AUTO's frame count seems excessive; (c) you want to precisely reproduce a known bracket sequence from a previous session. | |||
| 2 | INTERVALTime between each frame in the sequence | 0 sec (minimum) | Set |
| Use the shortest interval: Zero or minimum interval fires frames as fast as the camera can process and write them, minimising the time between focus positions. A longer interval increases the chance of environmental movement (wind, water ripple, settling tripod) between frames, creating misalignment that Photoshop or Helicon Focus cannot repair cleanly. The only reason to increase the interval is if you are using flash — give the unit time to recycle. For available-light landscape work, minimum interval is always correct. | |||
| 3 | FRAMESMANUAL mode only — number of shots in the sequence | See lens guide (Part V) | Set |
| In MANUAL mode, FRAMES sets the total number of exposures in the bracket sequence. The camera fires this many frames and stops, regardless of whether focus has reached infinity. If your STEP value is too large, you may have gaps in the focus coverage — zones between frames where nothing is acceptably sharp, which appear as soft bands in the blended output. If STEP is too small, you will generate more frames than needed (though this has no quality penalty, only storage and processing cost). The lens-specific guidance in Part V gives starting FRAMES values for typical landscape scenarios. | |||
| 4 | STEPMANUAL mode only — amount focus shifts per frame | 1–10 · Start at 3–4 | Set |
| Step size is the most consequential MANUAL setting. A STEP value of 1 moves focus very slightly — this is appropriate for extreme macro or for very close foreground elements where depth of field per frame is measured in millimetres. A STEP value of 5–8 moves focus substantially — appropriate for longer focal lengths (100mm) focused at medium distances, where each frame's depth of field zone is large and large steps maintain coverage without overlap. At f/8 on the GF 20-35mm for typical landscape distances (0.5m foreground to infinity), a STEP of 3–4 in AUTO mode produces approximately 4–8 frames. Start there and refine based on results. The key check: the last frame of the sequence must show infinity in acceptable focus. If it doesn't, increase FRAMES or decrease STEP. | |||
Essential accompanying camera settings
| # | Setting | Value | Action |
|---|---|---|---|
| 1 | SHUTTER TYPECritical — change from C1 default for bracket sequences | ELECTRONIC | Set |
| Switch to Electronic Shutter for Focus Bracketing sequences. The mechanical shutter's curtain travel introduces a small but real vibration at the moment of exposure. For a single deliberate frame, the 2-second self-timer eliminates this vibration before it matters. For a 10-frame bracket sequence firing at minimum interval, there is no delay between frames — the mechanical shutter introduces vibration from frame 2 onwards that cannot be corrected in post. The Electronic Shutter is completely silent and vibration-free; it exposes each frame without any mechanical shock. Important: confirm there are no flickering artificial light sources in the scene. In full daylight on a natural landscape, Electronic Shutter is entirely safe. Switch back to Mechanical after the bracket sequence is complete. | |||
| 2 | IS MODE | OFF (on tripod) | Set |
| This is the C1 universal rule — particularly critical for bracket sequences. IBIS moves the sensor between frames. Even a microscopic position difference between frames will misalign the pixels, and at 102 MP those misalignments are visible in the blended output as fine fringing at high-contrast edges. The tripod is providing the stability; IS MODE must be OFF so the sensor stays absolutely fixed at the same position for every frame in the sequence. | |||
| 3 | OIS zoom barrel switchPhysical switch on the lens — must also be OFF | OFF (on tripod) | Set |
| The GF 45-100mm F4 and GF 100-200mm F5.6 both have independent OIS switches on the barrel. Both the barrel switch AND the camera IS Mode must be OFF for tripod stacking. One active and one off is still a failure. Lens OIS stabilises by moving optical groups independently — it will generate inter-frame differences that no alignment algorithm can fully correct. | |||
| 4 | Self-timer or remote release | 2s timer or RR-100 | Set |
| The 2-second timer gives the tripod time to settle after you press the shutter button before the bracket sequence begins. Once the sequence starts, the camera fires all frames automatically without any further button input. The RR-100 remote release is the cleaner solution — it eliminates button vibration entirely. With the remote, the sequence can begin immediately without a 2-second delay. | |||
| 5 | APERTUREThe core decision — set before entering Focus BKT mode | f/5.6 or f/8 | Set |
| 6 | — | Skip | |
| Shoot the entire bracket sequence at your lens's optical sweet spot. The whole purpose of stacking is to avoid stopping down for depth — if you stop down to f/16 to "help" the stacking, you have negated the benefit and introduced diffraction into every frame of the blend. Set f/8 and leave it. | |||
| 7 | ISO | 100 (fixed) | Set |
| 8 | DRIVE MODESet on DRIVE button, not in menu | FOCUS BKT | Set |
Do not touch the zoom ring during a bracket sequence. The manual explicitly states this. Zooming changes the field of view between frames — the frames will not align and the blend will fail. If you are using the GF 45-100mm or GF 100-200mm F5.6, set your focal length before beginning the sequence and do not touch it until the last frame is written.
Field Execution — AUTO Mode Step by Step
The AUTO mode workflow is the recommended default for landscape focus stacking. The sequence below assumes you are on a tripod with the scene composed and the exposure set in Manual mode at f/8, ISO 100.
A SHOOTING SETTING → FOCUS BKT SETTING. Select AUTO. Set INTERVAL to 0 (minimum). Exit the menu. Press the DRIVE button and select FOCUS BKT from the drive mode list. The camera is now in Focus Bracketing mode — you will see the bracket indicator in the EVF and on the LCD.Make a single reference frame at the same exposure before the bracket sequence. Before activating Focus BKT, take one frame at your intended focus point in normal shooting mode. This gives you a reference file in Lightroom with full metadata that you can use to identify the start of the sequence in your catalogue, and it confirms the exposure is correct before you commit to a multi-frame sequence.
Lens by Lens — When and How Each GF Lens Stacks
GF 20–35mm F4 — the landscape primary
The 20-35mm is the C1 primary lens precisely because of its relationship with depth of field. At 20mm f/8, the hyperfocal distance is approximately 1.4-1.8 metres depending on the viewing assumptions behind the calculation — meaning practical near sharpness begins somewhere around 0.7-0.9m and extends to infinity in a single frame. This is a remarkably generous working depth of field for a 102 MP sensor.
When stacking is needed at 20mm: only when you have a specific foreground element closer than 80–90cm from the lens that must be simultaneously sharp with the distance. A mushroom at 30cm, a tide pool at 50cm, a rock texture at 60cm directly below the lens. Outside of this close-foreground scenario, a single frame at f/8 focused at 1.5–2m provides everything you need. Do not stack unnecessarily at 20mm — the added post-processing complexity is only justified by a genuine near-field focus requirement.
When stacking is needed at 35mm: more frequently. At 35mm f/8, hyperfocal distance is roughly 4.2-5.5 metres depending on the output assumptions behind the calculation, so a foreground element at 1.5m remains outside practical depth of field and usually requires stacking. At 35mm, any subject closer than 2-3m warrants a bracket sequence if simultaneous sharpness with the distance is required.
| Focal length | Foreground distance | Action | Approx. frames at f/8 AUTO |
|---|---|---|---|
| 20mm | 90cm – ∞ | Single frame, focus at 1.5m | — |
| 20mm | 50–90cm | Stack: A at subject, B at ∞ | 3–5 frames |
| 20mm | <50cm | Stack: A at subject, B at ∞ | 5–10 frames |
| 35mm | 2m – ∞ | Single frame, focus at 4m | — |
| 35mm | 1–2m | Stack: A at subject, B at ∞ | 4–7 frames |
| 35mm | <1m | Stack: A at subject, B at ∞ | 7–14 frames |
GF 45–100mm F4 OIS — the compression lens
The 45-100mm changes the stacking calculation fundamentally. At 100mm f/8, depth of field at medium distances is measured in metres, not tens of metres — a subject at 10m has a depth of field of roughly ±0.5m. The GF 100-200mm F5.6 extends that same compression logic when reach becomes the composition. For any long-lens landscape scene with meaningful depth, stacking is essentially mandatory at the longer end of these zoom ranges.
The 45-100mm and 100-200mm generate the most stacking work in this system — but they also produce the most compelling stacked results, because tonal compression creates images where every element in the frame reads with equal pictorial weight. When the near and far elements are simultaneously sharp under long-lens compression, the spatial relationship between them is entirely different from what 20mm wide-angle renders. It is worth the additional frames.
The GF 100-200mm F5.6 enters the same territory when reach becomes the picture. It is not the default C1 lens, but it is the prudent choice when the composition depends on distant ridge layers, woodland compression, or a subject across water or open ground that still feels too loose at 100mm. Treat it as an extension of the 45-100 logic: greater compression, shallower depth of field at distance, and an even stronger need to stack deliberately if multiple planes must hold. Use it when the scene asks for long-lens order rather than trying to crop the 45-100 into a role it did not quite reach.
The OIS zooms in this system have two stabilisation points that must both be OFF on a tripod. The physical barrel OIS switch AND the camera IS Mode. Check both before every sequence. With either active, the glass elements and sensor are both moving independently between frames — the resulting misalignment at 102 MP produces fringing artefacts in the blend that are impossible to remove cleanly.
| Focal length | Scene depth scenario | Action | Approx. frames at f/8 AUTO |
|---|---|---|---|
| 45mm | Subject at 3m, infinity | Stack | 5–8 frames |
| 60mm | Subject at 5m, infinity | Stack | 7–12 frames |
| 80mm | Subject at 8m, infinity | Stack | 9–16 frames |
| 100mm | Subject at 10m, infinity | Stack | 12–22 frames |
| 100mm | Subject at 5m, 30m background | Stack — finite range | 6–10 frames |
| GF 100-200mm at 150mm | Distant planes with separated ridges/trees | Stack — compressed range | 10–18 frames |
| GF 100-200mm at 200mm | Remote subject layers / across water | Stack only if multiple planes must hold | 12–24 frames |
GF 80mm F1.7 — landscape stacking is niche
The 80mm f/1.7 is not a landscape lens — its DC motor AF and its identity as a portrait prime mean it is rarely mounted in a C1 session. However, it is the Pixel Shift prime and the finest resolution lens in this system when the subject is flat or at a single distance.
Where the 80mm focus stacking makes sense: flat-subject fine art reproduction (a tiled floor, an aged wall surface, artwork) where you need both maximum resolution and maximum depth within a flat plane that isn't perfectly parallel to the sensor. Also useful for intimate landscape details — a section of lichen-covered rock at close range, a small pool of water with surface texture and depth, botanical close-ups in natural settings. These are specific, deliberate applications rather than general landscape use.
When the session objective is architectural texture or relief detail rather than near-to-infinity landscape depth, continue in the C5 Architecture Focus Stacking Addendum and execute from the C5 field checklist to keep the capture sequence aligned with the flatter geometry model.
Post-Processing — From RAW Files to Blended Master
Focus stacking post-processing has two stages: preparing the RAW files in Lightroom, and blending them in Photoshop (or dedicated software). The RAW files must be prepared identically before blending — any difference in exposure, colour temperature, or tone curve between frames will create visible seams in the output at the transition zones between depth planes.
Stage 1 — Lightroom preparation
Settings → Sync Settings. Sync everything except Crop, Spot Removal, and Graduated/Radial Filters. Every frame in the sequence now has identical processing.Photo → Edit In → Open as Layers in Photoshop. Lightroom renders each RAW as a 16-bit PSD or TIFF and opens all of them as stacked layers in a single Photoshop document. This is the complete, processed dataset — every frame is a full 16-bit, Lightroom-corrected render. Alternative for dedicated software users: export all frames as 16-bit TIFF files to a folder, then open them in Helicon Focus or Zerene Stacker directly. See Part VII.Stage 2 — Photoshop blending
Ctrl/Cmd + A to select all, or click the top layer, shift-click the bottom). Navigate to Edit → Auto-Align Layers. Select Auto projection. Click OK. Photoshop analyses each layer for matching content and shifts them to align pixel-for-pixel. This corrects for any tripod micro-movement between frames, and for focus breathing (the slight change in magnification as focus shifts). The alignment process may crop a thin border from the image — this is normal and expected at 102 MP, the resulting image is still enormous. The alignment step takes 30–120 seconds depending on frame count and machine speed.Edit → Auto-Blend Layers. A dialogue appears with three options. Select Stack Images (not Panorama). Check Seamless Tones and Colors — this allows Photoshop to make minor exposure adjustments at blend boundaries to reduce visible seaming. Uncheck Content Aware Fill Transparent Areas unless you need it for edge fills. Click OK. Photoshop analyses the sharpness of each layer at every pixel position, selects the sharpest source for each pixel, and creates layer masks to blend between them. The result is a single layer (after merging) with complete depth of field across the full scene at f/8 optical quality. Processing time: 1–5 minutes for 6–20 layers at 102 MP.Layer → Flatten Image. Save the file (File → Save) — Lightroom has been watching the file and will automatically import the flattened composite back into the catalogue as a new file adjacent to the source RAW frames. The composite is a 16-bit TIFF or PSD at full 102 MP resolution (minus the thin edge crop from alignment). Apply any final global adjustments in Lightroom — output sharpening, final tone, vignette — and export for print.The output file will be very large. A 16-bit TIFF from a blended 102 MP stack is approximately 600 MB. A PSD with layers intact before flattening can reach 3–6 GB for a 15-frame sequence. Ensure your working drive has adequate space before beginning the blend, and archive the flattened master TIFF rather than the layered working file.
Dedicated Software — When Photoshop Isn't Enough
Photoshop's Auto-Blend Layers is capable and convenient for clean sequences with static subjects. For complex scenes — irregular focus transitions, near-field subjects with complex geometry, partially moving elements, or sequences with more than 15 frames — dedicated focus stacking software produces measurably better results. The algorithms are purpose-built for the task rather than being a feature of a general-purpose compositor.
Method A — Weighted Average: averages pixel values across frames, weighted by sharpness. Produces smooth, gradual transitions. Best for: subjects with fine, intricate detail (botanical, geological texture). Risk: slight detail reduction in complex areas where frame weights are close.
Method B — Depth Map: builds a depth map of the scene and assigns each pixel to its optimal frame based on calculated distance. Fastest method. Best for: most landscape scenarios with clear separation between depth planes. Risk: can produce visible transitions ("stepping") in scenes where depth varies continuously without clear boundaries.
Method C — Pyramid: decomposes each image into frequency bands and blends the sharpest version of each frequency band independently. Highest quality. Best for: complex scenes where Methods A and B show artefacts; any sequence where the result will be output at maximum print size. Slowest — allow 5–15 minutes for a 15-frame 102 MP sequence. This is the method to use when the image matters most.
Retouching tool: Helicon Focus includes a retouching interface where you can paint source-frame assignments directly onto the blend output — select which frame contributes at any specific zone. Essential for correcting halo artefacts or movement ghosting.
PMax (Point Maximum): selects the maximum contrast pixel at each position across all frames. Extremely sharp; tends to accentuate edge definition. Best for: scenes where sharpness is the absolute priority and smooth blending is secondary; macro and close-focus botanical. Risk: can produce slightly "crunchy" halos at transitions when contrast is high.
DMap (Depth Map): similar to Helicon's Method B — builds a depth map and assigns pixel sources accordingly. Produces smoother transitions than PMax. Best for: general landscape where smooth blending is as important as sharpness.
Retouching: Zerene includes a layer-based retouching interface similar to Helicon's, allowing manual correction of problem areas.
When to choose Zerene over Helicon: for very close-focus subjects (within 30cm) with intricate surface detail, Zerene PMax often produces sharper results than Helicon Method C. For general landscape (subjects at 0.5m+), the difference between the two tools is smaller and workflow preference becomes the deciding factor.
Which tool for which situation
| Scenario | Recommended tool | Method / Mode |
|---|---|---|
| General landscape, 4–8 frames, static scene | Photoshop | Auto-Blend → Stack Images |
| General landscape, 10+ frames, print output | Helicon Focus | Method C (Pyramid) |
| Close foreground detail (rock texture, 30–80cm) | Helicon Focus | Method A or C |
| Botanical / macro (10–30cm subject) | Zerene Stacker | PMax |
| Scene with halos or blending artefacts in PS output | Helicon Focus | Method C + retouch |
| Scene with partially moving elements (wind) | Helicon or Zerene | Manual retouch of ghosted zones |
Failure Modes — What Goes Wrong and Why
Focus stacking failures are almost always preventable. Each one has a specific cause and a specific remedy. Understanding them before the session means you can avoid them in the field and diagnose them accurately when they appear in post.
The client proposition. A focus-stacked landscape from the GFX 100S II at f/8 delivers something that no single-frame approach can match: 102 MP of resolution at the optical peak of the lens, with depth of field from centimetres to infinity. At large print scale — 100×150cm or larger — the difference between a diffraction-compromised f/16 single frame and a clean f/8 stack is visible to any viewer who spends time with the print. It is not a technical subtlety. It is the difference between a photograph that holds up at close distance and one that reveals its limits. That is the value of the practice.
Support the work
This guide is free to read
The complete GFX 100S II suite — six shooting streams, field references, and the linked editing guides — is available here. If a document has saved you time in the field or at the edit, a Ko-fi contribution helps keep it going.
