Edge Blending Projection: A Two-Projector Workflow
Key takeaways
- Plan a two-projector edge blend with 15-30% overlap of image width, landing around 20-25% as a working default.
- Use a gamma or custom blend curve instead of linear to keep summed luminance flat across the overlap band.
- Calibrate in a fixed order: physical alignment, then geometry and keystone, then warp, then blend.
- Built-in projector blending only works for identical models on flat walls; curved surfaces and mixed units need media server warp and blend.
- Test every blend with full-white fields and moving grey ramps, since dark show content hides seams.
Edge blending projection for two projectors comes down to three decisions: an overlap of 15-25% of the image width, a blend curve matched to your projectors' gamma, and a calibration order that fixes geometry before warp and blend. This guide walks through each step for flat, curved, and dome surfaces, then hands the result off to your media server.
What is edge blending projection and when do you need it?
Edge blending projection is the technique of overlapping two or more projected images and fading their edges into each other, so the array reads as one continuous canvas. The overlap zone gets a brightness ramp on each projector so the combined output stays even across the seam.
You need it when no single projector covers the surface: a wide flat wall, a curved set piece, a dome segment, or an immersive room where the required throw distance or lens ratio can't be met by one unit. If a single projector covers the surface with margin to spare, adding a second is extra cost, extra alignment, and extra failure points.
The alternative, hard-edged tiling, places images flush with no overlap. It's cheaper and faster, but any drift in alignment, warp, or lamp output shows as a visible line. Built-in blending on projectors exists for this reason; Epson's projector documentation covers edge blending and overlap setup on models that support it directly (Epson support). For anything beyond a simple flat wall, most operators plan the geometry and blend in software and hand the result to a media server. More on that workflow is in the THROXEL blog.
How much overlap should two projectors have for a clean blend?
For a two-projector edge blend, aim for 15-30% overlap of each image width. Below about 15%, the blend zone is too narrow to hide luminance and colour differences between panels, and you get a visible band. Above 30%, you are stacking two images over a wide strip, which wastes lumens and complicates warp alignment for marginal gain. Many operators land around 20-25% as a working default, then adjust after checking brightness uniformity on site.
Calculating overlap from throw distance and screen width
Work from throw ratio, not guesswork. If your screen is 8 m wide and your projector's throw ratio is 1.2:1, throw distance is 9.6 m for the full image. For a 25% overlap, each projector covers 6.25 m of screen width, so each one's throw distance is 6.25 × 1.2 = 7.5 m, positioned accordingly. Check lens shift limits at that position: if you need shift to hit the overlap zone and the projector is near the end of its shift range, reposition or swap the lens before load-in. Planning this in 3D with a tool like THROXEL lets you verify throw distances, shift, and the resulting overlap against the actual geometry before the truck leaves.
Why overlap costs brightness and how to compensate
In the overlap zone, both projectors contribute light, so that strip runs brighter than the rest of the screen unless the blend curve compensates. With a 25% overlap, a quarter of your screen area is double-lit. The blend ramp scales each projector's contribution down across the zone so the sum stays constant, but the curve only works if the overlap is wide enough for a smooth ramp. Narrow overlap forces a steep ramp, which is exactly where seams show. If brightness at the seam still reads hot, drop each projector's output slightly and let the blend zone carry the balance, or widen the overlap toward 30% if throw distance allows.
How do you choose between linear, gamma, and custom blend curves?
A linear curve fades both projectors at a constant rate across the overlap. It works on bright, flat content with even luminance, but on most real installs it produces a visible dip because the two projectors' outputs don't sum to full brightness in the middle of the band.
Gamma and power-law curves shape that fade so the summed luminance stays flat. A gamma curve is usually the right default for two-projector blends on dark or video content; a custom curve is what you reach for when the blend band sits over uneven content, a curved surface, or projectors with mismatched lenses or lamp ages.
| Curve | Best for | Watch out for |
|---|---|---|
| Linear | Flat walls, even content, quick rigs | Visible luminance dip mid-band |
| Gamma | Video/dark content, standard blends | Needs matching projector gamma |
| Custom | Curved surfaces, mismatched projectors | Time-consuming to tune on site |
VIOSO's ProjectionTools documentation covers blend width and overlap guidance in detail here. Whatever curve you pick, verify it against your actual overlap before showtime: THROXEL simulates blend geometry and coverage per projector in the array, so you can test curve behaviour before load-in (features).
How do you align projectors before blending: geometry, keystone, and warp order?
The order that avoids rework: physical alignment, then geometry and keystone, then warp, then blend. Each stage locks in the assumptions of the next, so fixing a blend on top of an unresolved keystone means doing it twice.
Physical alignment checklist
Before touching any software, get the optics right. Match throw distance and lens shift to the plan, square the rig to the surface, and level the images so the overlap zone is even across its full height. On a flat wall, two well-rigged projectors should need only minor keystone; if you are leaning heavily on digital correction to fix a sloppy hang, every later step inherits that error.
Warp and blend order in media servers
In Disguise Designer, PIXERA, or Resolume, the workflow is the same shape: set output resolution and placement first, then geometry correction, then warp to the surface, then edge blend. Warp changes pixel positions, so blending before warping guarantees a curved seam. On curved surfaces and domes, expect iterative passes: warp, check the overlap width on the test grid, adjust, then re-trim the blend curve. Budget real on-site time for that loop, because a dome rarely accepts the pre-build warp unchanged.
How do you plan an edge blend before the projectors are installed?
Model the surface in 3D, place the projector array in software, and check throw ratio, lens shift, and overlap against the actual lens spec before anything ships. A two-projector blend planned in 3D tells you the overlap width in metres, whether the lens ratio covers the surface from the rigging position, and where warp zones fall on curved geometry. That is the difference between a load-in that runs to plan and one spent repositioning projectors that were never going to cover the surface.
The honest workflow: build the surface geometry, drop in projector positions at realistic mounting heights, set the throw ratio from the manufacturer's lens data, and simulate coverage and blend zones from every projector's viewpoint. If a projector needs more than a few degrees of keystone or its lens shift is maxed out to reach the surface, fix that in the plan, not on the ladder. Tools like Disguise Mapping Matter support this kind of pre-visualization for warp and blend workflows, and THROXEL does the same for projection planning and simulation at EUR 29/month. The limitation to accept: simulation gets you close, but final alignment still happens on site against the real surface.
Pre-build planning checklist
- Surface dimensions and geometry (flat, curved, or dome) measured on site or from CAD
- Projector positions, mounting heights, and rigging constraints
- Throw ratio and lens shift range from the manufacturer's lens spec sheet
- Overlap width in metres, and the blend curve you will apply
- Export format your media server accepts for warp and blend
Comparing planning tools: THROXEL, VIOSO, Disguise Mapping Matter
THROXEL, VIOSO ProjectionTools, and Disguise Mapping Matter all model geometry and simulate projector coverage; they differ in ecosystem and price. THROXEL is priced for individuals at EUR 29/month with a 7-day trial and machine-bound licence, aimed at handing planned setups to Disguise, PIXERA, or Resolume. VIOSO ProjectionTools and Mapping Matter sit closer to their respective product families. No independent user reviews of THROXEL exist yet, so treat the trial as your evaluation.
What causes visible seams in edge blends and how do you fix them?
Most seams come down to five causes. Brightness mismatch: two units of the same model still drift after lamp/LED ageing, so re-run projector matching or set per-output gain in your media server. Colour shift: different firmware, lamp hours, or a swapped lens can shift white point; match colour settings in the projector OSD before touching the blend. Wrong blend width: too narrow and the overlap edge reads as a hard line, too wide and you lose brightness across the band. Gamma mismatch: the blend curve that looked right on one projector pair is wrong on another, so adjust the curve per output. Content-dependent visibility: dark scenes hide everything, bright gradients expose a seam you missed, so always test with full-white and moving grey ramps.
PIXERA's blend and warp tools let you tune overlap width and blend curves per output during calibration, which is where most of these fixes happen (https://www.pixera.tv/). Fix geometry and colour first, blend curve last.
How does edge blending differ between hardware built-in tools and media server software?
Built-in projector blending is limited to what the manufacturer supports: usually two units of the same model, fixed overlap percentages, and simple linear or gamma curves. It works when your array is symmetrical, the geometry is flat, and you are not warping onto a curved surface. The moment you need custom warp on a dome, an architectural corner, or an asymmetric overlap, you are out of range.
Media-server warp and blend in Disguise, PIXERA, or Resolume gives you per-output mesh warping, arbitrary blend widths and curves, and colour matching across mixed projector models. The trade is setup time and a machine in the signal chain that must stay healthy before showtime.
Either way, the planning problem is identical: throw distance, overlap, and coverage must be worked out before load-in. That is where THROXEL fits: you project the geometry in 3D, fix the overlap and blend zone numerically, then hand the validated numbers to whichever tool does the final warp on site. Where it falls short: it does not replace the server's live calibration, and there are no public user reviews yet to pressure-test it against VIOSO ProjectionTools or Mapping Matter.
FAQ
Can you edge blend projectors with different lens ratios or models?
Yes, but only through media server software. Built-in projector blending usually requires two identical models with matching lenses, and mismatched throw ratios make the overlap widths differ across the image height. In Disguise, PIXERA, or Resolume you can warp each output independently and match colour and gain across mixed units, which absorbs most of the mismatch.
Does edge blending reduce overall brightness, and by how much in practice?
The blend band itself loses brightness because both projectors are faded down inside the overlap; the rest of the canvas runs at full output. With a linear curve the summed luminance dips noticeably mid-band, which is why gamma or custom curves are the default. A badly tuned curve can make the seam visible even on full-white test patterns.
How wide should the blend zone be on a curved screen or dome?
Plan for 15-25% of the image width, then verify against the geometry. On curved surfaces the effective overlap changes with viewing angle and warp, so a width that works at the centre can read too narrow at the edges. Simulate the blend zone per projector before load-in, then expect to re-trim the curve during on-site warp passes.
Can projector built-in edge blending replace a media server for a two-projector wall?
For a flat wall with two identical projector models and no warping, yes. Built-in blending handles symmetrical arrays with fixed overlap percentages and simple curves. It fails once you need custom warp, an asymmetric overlap, curved geometry, or mixed projector models. It also puts the blend inside the projector menu tree, which makes on-site tuning slower than a media server.
How do you test an edge blend without the real content?
Use full-white fields and moving grey ramps. Dark scenes hide seams, so testing only with show-like content lets a brightness dip or colour shift survive to opening night. Full white exposes luminance mismatch, grey ramps reveal banding and gamma problems across the blend band, and a test grid shows whether warp left the overlap width uneven across the height.