Frozen Projection Plot

Three things to settle before anything is ordered, and one that is already fine. The booth head‑clearance survey has been walked and passes; the procedure is kept in §05 for the record.

The dimension drawing gives distances but not their endpoints. Four of the five labels resolve to one consistent geometry, so that is the geometry used throughout.

Lineset schedule — measured

Battens are on 8″ centres with a handful of exceptions, all of them around the electrics. Everything in this document is now built on these numbers rather than on the nominal 1′‑per‑lineset assumption the drawings imply.

With these two lenses the assignment is not a preference, it is arithmetic. Throw ratio is throw distance divided by image width, so each lens can only live in one part of the room.

Projector A — tech booth

Projector B — spot booth

• 89′‑0″ from the plaster line, lens at +18′‑0″. Zoom to full tele (1.32:1) and the raster lands at 67′‑5″ × 42′‑2″ — there is no tighter setting on this lens.
• Let the warp throw the extra width away — do not correct it with the lens. 4′‑9″ falls outside the portal on each side. Its QLab region maps the same content into a smaller quad inside the raster and leaves the surplus black (§07), so there is nothing to author and nothing to mask. Never try to fix it with zoom or projector geometry — you would only lose more pixels.
• Lens shift down 6′ to put the raster centre at +12′, about 14% of image height — trivial for this lens.
• Its own dedicated 120 V, 20 A circuit in the spot booth. Do not spur it off the followspot supply.
• Confirm the spot booth has the depth and the port for a 77 lb, 22″‑deep body alongside whatever followspot lives there. This is the practical question that decides whether the stack happens at all.

Booth installation

• Project through an open port, never glass. Booth glass costs 10–15% of output and produces a faint ghost image offset from the main one. If the port must be glazed, use a single sheet of optical‑grade glass tilted 5–7° off perpendicular.
• The body is 77.2 lb, 21.7″ W × 22.4″ D. Build a shelf that carries it with the lens centreline at the height you settle on in §05, and that leaves 6″ of free air on every side for the intakes.
• Leave the projector square to the room and use lens shift, not keystone, to place the image. The lens sits at +5′ and the beam centre wants to be at +12′, so the raster shifts up 7′ — 19% of the image height, comfortably inside the ±52% the ET‑D75LE6 allows on this body. If it runs out, 7′ over 75′ is only 5.3° of tilt and the ±28° keystone covers it easily.
• Set Lens Position Lock and save a lens memory once aligned. Photograph every menu page.

Projector menu settings

Rigging

• Lineset 14 — 14′‑4″ upstage of plaster, lens face 13′‑6″ from the cyc, lens centre at 21′‑0″. Position the body on its yoke so the lens face lands at 13′‑6″; a few inches of fore‑and‑aft adjustment on a sidearm lets you trim it exactly. This is a bare batten, so it carries only the projector — but it also has none of an electric’s stiffness. Bridle it to Electric 3 with short trim chains, or dead‑hang a stub pipe at the same plane, so the map does not move when someone brushes the set. The mapping is only as repeatable as the trim, and a counterweight set that comes in a few inches low re‑breaks every warp point.
• Lineset 14 is a counterweight set, so the mapping is only as repeatable as its trim. Add a hard trim stop or trim chain at 21′‑0″ and a witness mark on the operating line, and re‑check alignment against the grid cue before every performance (§11).
• Both Panasonic beams pass right over this position, but the content there is black (only the leg panels carry image inside the opening), so the projector body shadows nothing. Keep it matte black anyway so it does not catch stray light.
• Mount with a safety cable to an independent point, and get the lens centre at 21′‑0″ rather than slung below the pipe. Have the electrician confirm the set’s capacity against ~80 lb concentrated at centre, on top of the fixtures already there.
• Plan the service position. At 21′ a genie or a tall ladder reaches the focus ring, which is one of the quieter advantages of this hang over the higher positions.
• Power: one dedicated 20 A / 120 V non‑dim, non‑switched circuit at the pipe. Never a circuit that dies with house power or gets swept by the lighting console.

Projector menu settings

This is the section the TD, the LD and the choreographer all need. A projector beam is a physical object in the air; anything inside it is either a screen or a shadow.

The Epson beam, lineset by lineset

With the lens on lineset 14 at 21′‑0″, 13′‑6″ from the cyc, the beam envelope at a point x feet upstage of the lens is:

lower edge (ft)  =  21 − 1.556 × x        upper edge (ft)  =  21          <- dead level
half-width (ft)  =  1.429 × x

Maximum scenery height under the Epson beam

For any point D feet upstage of the plaster line, between 15′‑2″ and the cyc at 28′‑8″:

max unobstructed height (ft)  =  21 − 1.556 × (D − 15.17)

The two front‑of‑house beams over the house

Both Panasonics cross the entire auditorium, from opposite ends of the vertical range: the tech‑booth lens sits at +5′‑0″, 75′ out; the spot‑booth lens at +18′‑0″, 89′ out. That 13′ of height and 14′ of depth is the whole reason for the stack, and it is also what makes each unit behave differently in the two paragraphs below.

Head clearance

The lowest useful ray of each unit runs from its lens to the bottom of the portal legs at deck level. Height above stage datum at distance x from each lens:

tech booth   h(x)  =  5  × (1 − x / 75)
spot booth   h(x)  =  18 × (1 − x / 89)

The spot booth is not a concern — it is above the audience by design and stays over 7′ for the first 60′ of the room. The tech booth is the one that matters, and it starts at chest height: its lowest ray is at 4′ only 15′ out of the booth and reaches the deck at the portal. In practice the tech‑booth beam grazes the heads of the back third of the house, which is exactly why its shadows are life‑size (below) and why the spot booth earns its keep.

Performer shadows on the portal

Anyone downstage of the plaster line — on the apron, on the pit cover — casts a shadow onto the bottom of the portal. For a performer of height h standing d feet downstage of the plaster line, the shadow reaches up the portal to:

tech booth   =  5  + (h − 5)  × 75 / (75 − d)
spot booth   =  18 + (h − 18) × 89 / (89 − d)

The stack does not remove the shadow, it grades it. A performer pressed against the portal blocks both lenses at once and the silhouette is solid — that is geometry, not a fault of the rig, and it is true of any number of projectors. Move them downstage and the two shadows separate vertically, so the upper part of the silhouette fills back in to about 58% of full brightness while the lower part stays dark.

Treat the bottom 6′‑0″ of the portal legs as a shadow band — dark, low‑contrast, low‑detail texture only, with the bottom 2′‑9″ the darkest of it. Do not put a light source, a face, or anything with a hard edge down there.

Performer shadows on the wall legs

The wall legs are lit from front of house, so anyone between a booth and a panel shadows it. With a full company on stage that is most of the ensemble. This is the question the stack was built to answer, so the numbers below are given twice — once for each unit — and then combined.

1 · The tech booth sets the height of every shadow; the spot booth fills the top of it

Because the tech‑booth lens is at chest height rather than up in the air, its rays run almost horizontally and its shadows barely foreshorten: a 5′‑6″ performer casts a 5′‑6″ to 5′‑8″ shadow from anywhere on stage, and a 6′‑2″ performer about 6′‑6″. The spot‑booth lens at +18′ looks down on the same performer, so its shadow is squashed — and the further upstage the panel, the shorter it gets.

2 · Each body still blacks out about half the panel’s width

Magnification runs from 1.0× (standing beside the panel) to 1.32× (at the plaster line, shadowing pair 3), so 20″ shoulders cast a 20–26″ shadow on a 48″ panel — 42% to 55% of its width. Two people abreast in the same corridor still black the panel out.

The stack helps less here than it does vertically. Both lenses are on centreline, so the two shadows of the same body land within 6–8″ of each other laterally — against a 26″ shadow that softens the edges but does not open a gap in the middle. The fill you get from this rig is vertical, not horizontal. Plan for it that way: the top of a silhouette dissolves, the waist does not.

3 · The corridors are narrow, and the centre of the stage is free

A performer only shadows a panel if they stand in the wedge between a lens and that panel. Those wedges are surprisingly tight. Because the spot booth is 14′ further out, its wedges sit a few inches outboard of the tech booth’s — so the union of the two is where any shadow at all falls, and the overlap is where a solid one falls:

Nine mapped surfaces in three planes: the portal frame at the plaster line, six leg panels at three depths upstage, and the cyc at the back wall.

Where the leg panels land in the tech‑booth raster

Now that the six positions are surveyed, the starting quads can be computed. A physical point X feet off centreline and Y feet above the deck, sitting D feet upstage of the plaster line, appears in the plaster‑plane raster at:

s  = 900 / (900 + D)                     <- D in INCHES, booth at 900"
X′ = X × s                               <- horizontal, independent of lens height
Y′ = H_lens + (Y − H_lens) × s          <- vertical, needs the surveyed lens height

px = (X′ + 29) × 66.2                    py = (30.125 − Y′) × 66.2

Are all six panels fully lit?

Short answer: horizontally yes, vertically no. Three separate things can shade a leg panel, and each one bites a different pair. The panel positions below already account for all three — pairs 2 and 3 were moved inboard from the original survey precisely to keep them out of each other’s shadows.

1 · The portal header shades the top of every panel — and a border covers it

Both Panasonics see the stage through the 50′ × 17′‑6″ working opening. That aperture is a physical mask at the plaster line, so nothing upstage of it can be lit above the ray that grazes the header. Where that ray lands depends entirely on lens height, and the two booths are at opposite ends of it. The tech booth at +5′ sets the ceiling — sitting below the header, its ray climbs steeply as it travels upstage, so a low lens lights the upstage panels higher, not lower:

Each border trims to the lit line of the panel behind it. That is the whole rule, and it works because the border sits only 8″ downstage of its panel and the lit line barely moves between them — the border ends up masking to within an inch of where the light stops, from every seat in the house. Set the bottoms at 18′‑7″, 19′‑8″ and 21′‑4″.

The borders cannot go lower than those numbers — a border below the ray shaves light off every panel upstage of it, so each is a trim to hit, not a range. And note that usable canvas now varies by pair: 18′‑8″ on pair 1, 19′‑9″ on pair 2, and 21′‑5″ on pair 3 — the upstage pair is lit almost to its full 22′. Each leg tile is authored the same, and each pair’s border decides where it stops.

1a · The stack puts a brightness step at 17′‑5″ — grade for it

Because the two lenses cut the panels at different heights, every wall leg has two zones. Below 17′‑5″ both units land; above it, only the tech booth. That is a real, visible step in brightness, and it is at a fixed height you can author against.

2 · The aperture edge grazes pair 1

Horizontally the opening spreads at ±25′ × (900+D)/900. Pair 1 is the tightest: at 7′‑0″ upstage the aperture has reached ±27′‑4″ and the panel’s outer edge is at 26′‑0″ — 1′‑4″ of margin. Pairs 2 and 3 have 7′ and 13′. Moving pair 1 upstage from 5′ to 7′ doubled this margin, which is a quiet win.

3 · The panels shade each other — this is the real one

Every panel is an opaque object between the booth and the panels behind it, and the beam keeps diverging as it travels — so an upstage pair has to step inboard faster than the beam spreads, or it lands in the shadow of the pair in front. At the adopted positions:

Architecture

One stage, fifteen regions, three output routes. QLab 5 nests these as stage → region → output route → device. Several regions can share one route — that is how one projector covers seven separate physical surfaces — and, just as importantly here, two regions can cover the same rectangle of stage and go to different routes. That second fact is what makes the front‑of‑house stack cost you almost nothing in the workspace.

The stage is a tile atlas: each surface gets its own rectangle of undistorted content space in the Layout tab, and the Warping tab pushes that rectangle out onto its physical target. Designers author flat, rectangular files; the warp does the geometry.

* If your cyc is a flat drop, Perspective is correct. If it wraps or curves at the ends, switch that region to Bézier and raise the mesh split to 4 or 8.

Every ‑C region is a duplicate of its ‑A twin at the same stage origin and size. Build the seven route‑A regions first, get them warped and signed off, then duplicate them as a block, rename the copies, switch the copies to route C and re‑warp only those. Nothing else in the workspace changes — not a cue, not a file, not the mask.

Route settings

• Route A — 3840 × 2400, device: Panasonic, tech booth. Scaling mode Center (1:1, no resampling). This is the reference route: it is aligned first and everything else is aligned to it.
• Route B — 1920 × 1200, device: Epson. Scaling mode Center. The CYC region occupies output rows 76–1124; everything above and below is black.
• Route C — 3840 × 2400, device: Panasonic, spot booth. Scaling mode Center. Its content sits inset in the raster with roughly 270 px of black down each side and a smaller band top and bottom — that is the 9′‑5″ of surplus width, correctly thrown away.
• If the Panasonics will not accept 3840 × 2400 from the Mac, fall back to 3840 × 2160 on both. Nothing in the map changes except the portal region’s vertical extent, and the physical image width stays 58′.
• Name the devices in macOS Displays so route A and route C cannot be confused after a reboot. A swapped stack is invisible in a still frame and obvious in a shadow.

The stage mask

One greyscale PNG, FROZEN_MASK_v01.png, at 5760 × 2440. White passes light, black blocks it. It is your hard spill control — it keeps both Panasonics off the proscenium wall above the canvas, off the apron below it, and out of the working opening. One mask serves the whole stage, so it serves routes A and C at once.

Everything else is black. Build the real mask by tracing a photograph of the projected alignment grid on the actual canvas — the numbers above are the starting geometry, not a substitute for tracing. Add a 3–5 px Gaussian blur to the mask edges so they do not shimmer.

MacBook Pro specification

Signal chain

• Fit an EDID minder on both outputs. When a projector powers down or drops out, macOS renumbers displays and QLab loses its routes mid‑show. An EDID emulator holds the display present. This is the single cheapest reliability upgrade in the whole rig.
• Spot‑booth run — survey this before you buy cable. The two booths are usually adjacent or stacked, but the cable has to leave the tech booth, get to the spot booth, and arrive with 4K/60 intact. Passive HDMI will not do it past about 25′ at this bandwidth. Use active‑optical (fibre) HDMI — it is directional, so check the source and display ends before pulling it — or HDBaseT if there is Cat6A between the two rooms. Do not daisy‑chain or split from the tech‑booth projector: the two units need independent signals so either can fail alone.
• Epson run: booth to plaster 75′ + 14′‑4″ upstage + 21′ up + dress and service loop ≈ 145′. HDBaseT is rated to 100 m (328′) at this resolution; active‑optical HDMI at 200′ is equally reliable and needs no power at the far end. Either works — pick HDBaseT if the building already has Cat6A to the grid.
• Label both ends of every cable. Coil a 15′ service loop at the projector.

Power

The three projectors draw 31.5 A between them, so no two can share a circuit — and none should share with dimmers or anything on the lighting console’s sweep. The spot‑booth circuit is the new item on the electrician’s list, and it is the one most likely to be missing: spot booths are often fed a single 20 A circuit that the followspot already lives on. An 11.1 A projector plus a followspot on one 20 A breaker will hold until the followspot strikes, and then it will not. Check the Panasonic’s cordset when it arrives: a unit configured for 200–240 V may ship with the wrong plug, and you want a NEMA 5‑20P. All three projectors should have no‑signal shutdown and sleep disabled, so a momentary source dropout never turns the show dark.

macOS setup

• Displays → Extended (never Mirrored). Internal display is primary and carries the menu bar and the QLab editor.
• Turn off: screen saver, display sleep, system sleep, Night Shift, True Tone, auto‑brightness, automatic software updates, Time Machine, iCloud sync of the show folder.
• Focus mode on — no notification banners will land on the projector output.
• Freeze the macOS version for the run. Do not update between tech and closing.
• Create a dedicated user account with nothing else installed. Log in to it for every performance.

Everything here is authored in‑house, so these are the templates. Hand this table to whoever is building content and hold the line on it — mismatched canvases are the most common reason a mapped show falls apart in tech.

Rates are worst‑case; Snappy chunk compression typically saves another 20–40% on real content. Budget 1 TB of media for the show and buy 2 TB.

House rules

• 30 fps for everything except particle work — snowfall and blizzard go to 60 fps.
• Rec. 709, 8‑bit, square pixels, no interlacing, no variable frame rate.
• Wrap as .mov. Name as ACT-CUE_SceneName_TILE_vNN.mov, e.g. 1-08_LetItGo_PORTAL_v03.mov.
• Loop points matter. Any background that holds under dialogue must loop seamlessly — match first and last frame exactly, and test the loop in QLab, not in QuickTime.
• Never exceed two simultaneous 4K HAP layers plus one cyc layer. A cross‑fade counts as two layers on the same tile.

Encoding

The friendliest route for students is the free AVF Batch Converter. For a scripted pipeline:

ffmpeg -i source.mov -c:v hap   -format hap    -chunks 4 -pix_fmt rgba out_HAP.mov
ffmpeg -i source.mov -c:v hap   -format hap_q  -chunks 4 -pix_fmt rgba out_HAPQ.mov
ffmpeg -i source.mov -c:v hap   -format hap_alpha -chunks 4 -pix_fmt rgba out_HAPA.mov

-chunks 4 lets QLab decode across cores. Keep it at 4 for the 4K tiles and 1–2 for the small leg tiles.

Budget a full day with the stage clear, and a second full session for the stack — aligning two projectors onto one surface to sub‑pixel accuracy is not a task you squeeze into the end of a load‑in day. Do it in this order; skipping optical alignment and going straight to digital warping is how a map ends up soft and unstable.

1. Survey and record. Tape‑measure the booth‑to‑plaster distance, the cyc distance, and the six leg‑panel positions. Fill in the worksheet in §15 before anything is hung.
2. Head‑clearance laser check from the tech‑booth lens position (§05). Resolve this before the shelf is built.
3. Confirm the spot booth physically works before anything else is ordered: 22″ of depth for the body, a port that clears the lens at +18′, room alongside the followspot, and its own 20 A circuit. If any of those fails, the stack is off and the rest of this section reverts to one Panasonic.
4. Hang all three projectors. Epson on lineset 14, lens face 13′‑6″ from the cyc and lens centre at 21′‑0″, level and on centre, inverted, safety cabled. Both Panasonics square to their rooms and on centreline. Do not touch a menu yet.
5. Full white field from each, one at a time. Confirm coverage, confirm no clipping on either booth port, and walk the Epson beam edges against the fly plot. Correct the fly plot now, not in tech.
6. Optical alignment only, tech booth first. Tech‑booth Panasonic: zoom to 58′‑0″ measured on the deck, then lens‑shift the beam centre to +12′‑0″. Spot‑booth Panasonic: zoom to full tele and stop — do not chase 58′, you cannot get there — then lens‑shift down about 6′ to bring its raster centre to +12′. Epson: shift the image to sit 0′–21′ on the cyc, centred (it will run 38′‑7″ wide, leaving 5′‑9″ unlit at each cyc edge — that is correct, not an alignment error; the leg panels mask all of it from every seat). Focus each. No digital keystone yet.
7. Tape the corners. Put retro‑reflective tape on all four corners of the portal canvas, all four of the cyc, and the four corners of each of the six leg panels. Twenty‑eight marks. They stay for the run.
8. Project the alignment grid. Build a 1‑foot grid image per tile with numbered axes and a 50%‑grey field. Store it as a cue named GRID in the workspace — you will use it every night.
9. Warp the route‑A regions, region by region. Portal first (it is the reference plane), then the cyc, then the leg pairs from downstage to upstage. Pull each region’s corners to the tape. Use Perspective warp; raise mesh splits only if a surface is genuinely curved. Finish route A completely and sign it off before route C exists — the tech booth is the reference and nothing about it should move again.
10. Duplicate the seven route‑A regions, rename them ‑C, and switch them to route C (§07). Then align the stack:
    1. Kill route A. Warp each route‑C region roughly to the tape, as you did for A. This gets you within an inch or two.
    2. Bring route A back up and stop looking at the tape. From here you align the spot booth to the projected image, not to the physical marks — the eye reads a doubled edge far more finely than a tape measure reads a corner.
    3. Project the grid on both at once and alternate A/B on a fine control‑point grid: nudge, look, nudge back, look. Work one surface at a time and finish it before moving on.
    4. The spot booth is the softer image — 57 px/ft against the tech booth’s 66. Set it so its edges sit just inside the sharper one’s. A soft edge peeking outside a sharp one reads as a halo; tucked inside, it is invisible.
    5. Expect this to take a full session, and expect the last 10% to take half of it. Stop when a 1‑pixel line in the grid looks like one line rather than two, from the centre of the house and from both extreme side seats.
11. Re‑focus the spot booth after warping. It is throwing 14′ further and its depth of field on the leg panels is shallower than the tech booth’s. Focus it on leg pair 2, the middle depth, not on the portal.
12. Focus the Epson at two‑thirds down the cyc and accept the top and bottom. Re‑check after any trim change.
13. Photograph the grid on the real surfaces and trace the mask in Photoshop from that photograph. Export the 5760 × 2400 greyscale PNG, blur the edges 3–5 px, load it into the stage.
14. Measure actual output. Full white field, meter at the portal face, centre and both ends — take three readings: tech booth alone, spot booth alone, and both. Record all three. The stacked figure is what the Epson trims to, and the two singles tell you what you still have if one unit dies mid‑run.
15. Colour match, three ways. Project 50% white, then full R, G and B. Match the two Panasonics to each other first — they are the same model, so this should be small, but a mismatch here shows as a colour cast that moves with every shadow, which is far uglier than a static one. Then match the pair to the Epson using RGB gain and colour temperature. The cyc fabric is your reference white, not a test chart.
16. Shadow‑test the stack with bodies. Put four people on stage at the plaster line, at 7′ and at 14′ upstage, in and out of the corridors in §05, with the grid running. Photograph it. This is the only way to know what you actually bought, and it takes ten minutes.
17. Verify with motion. Run a slow diagonal wipe across every surface. Straight lines must stay straight across a warp boundary; a kink means a control point is off.
18. Lock everything. Lens Position Lock on both Panasonics, Epson Lens Lock on, lens memories saved, every menu page photographed, QLab workspace exported to two places. Photograph the stacked grid as your reference image — if the stack drifts later, that photo is how you know.
19. Verify with the lighting rig live. Bring up the LD’s brightest front‑light look and confirm the portal and cyc still read. This is where you find out whether front light needs shuttering off the canvas.

Cue structure

• Every look is a Group cue set to start all children, containing one Video cue per tile in use plus one Fade cue that takes the outgoing look to zero.
• Number projection cues to match the stage manager’s book: 1.08 = Act 1, cue 8. Never renumber mid‑run.
• Fixed layer plan, so nothing ever fights for stacking order:

• Keep a PANIC cue — a 0.5 s fade of every layer to black — on a dedicated hotkey the operator can find without looking.
• Keep the GRID cue and a HOUSE OPEN preset in a separate, non‑sequenced cue list.
• End each act with an explicit projection blackout group. Do not rely on the last video cue ending.

Pre‑show checklist

• All three projectors on, shutters open, no error lamps. Filter hours logged. Confirm both Panasonics are live — a single unit still makes a complete‑looking picture at half brightness, so this is easy to miss until the house is in.
• Media drive mounted; QLab workspace opens with zero broken‑file warnings.
• All three outputs present in Displays, in the right order; routes A, B and C green in Workspace Settings → Video.
• Fire the GRID cue. Walk downstage centre and confirm every grid line lands on its tape, and that each line reads as one line rather than two — a doubled edge means the stack has drifted. If the Epson is on a lineset, this catches trim drift before the audience does.
• Fly plot set to the projection state: Electric 4 and Back Traveler out; borders in at their trim marks — LS 5 at 18′‑7″, LS 12 at 19′‑8″, LS 22 (split) at 21′‑4″; Electric 3 at 17′‑6″ or its centre 5′‑6″ clear.
• 4′ spike line upstage confirmed visible to the cast.
• Cue 1 loaded and standing by. PANIC hotkey tested.
• Focus mode on; no notifications; UPS on and not in battery.
• Backup machine powered, workspace current, HDMI switch reachable.

Redundancy

The stack gives you something a single projector cannot: if one Panasonic dies mid‑performance, the show continues. The portal and legs drop to roughly half brightness and every shadow goes hard, but the picture is complete and correctly mapped, and nobody in the house will name what changed. Tell the operator this in advance so the instinct is to note it and keep going, not to stop and troubleshoot. Fix it at interval.

A single MacBook is a single point of failure for every image in the show. At minimum, keep a cloned media drive and an exported workspace so a spare machine can be running in ten minutes. If a second Mac is available, keep it powered, loaded and parked behind a manual HDMI A/B switch on all three outputs — that turns a show‑stop into a fifteen‑second gap. Export the workspace at the end of every rehearsal and every performance; date the file.

The three surface groups have very different resolutions, brightnesses and shadow risks. Content that ignores that will look like three unrelated shows.

Working with the lighting designer

• Shutter the front light off the canvas. A 9.5 fL image dies under 5 fL of ambient front light. Keep spill on the portal and cyc below roughly 1 fL for a 10:1 contrast ratio.
• Cyc lights and projection on the same cyc are mutually exclusive. Pick one per look and write it into the cue sheet.
• Grade leg pair 3 about 40% brighter than pair 1 to compensate for the beam spread (§03).
• Agree the haze policy per cue — the Epson beam is visible in haze (§05).
• Warm front light on a cool blue projection surface reads as muddy grey. Ask for cooler front light in the ice‑world scenes, or accept desaturation.

Everything above describes the map. These two plates show it carrying a scene: how a look distributes across the portal frame, the six leg panels and the cyc, at true geometry and true proportion. Both are original compositions built for this venue — your licence covers the script and score, not another production’s scenic design, so every plate here is drawn from scratch.

Layer plan for this look

Notes for whoever builds these

• The portal is your sharpest surface at 66 px/ft, and nothing crosses it above 6′. Put the fine work there. The cyc runs 52 px/ft and goes soft at the extreme top and bottom — atmosphere, not detail.
• Compose for the lit height of each pair — 18′‑8″, 19′‑9″ and 21′‑5″ of a 22′ panel; the border in front cuts it there, so anything above is bleed.
• Grade the upstage legs brighter. Pair 3 receives about 9.9 fc stacked against pair 1’s 13.9, so identical files will not read as one world. Lift pair 3 roughly 40% (§03).
• Slow motion on the legs. Performers cross in front of them constantly, and fast movement behind a moving body reads as a glitch. Drift, breathe, shimmer — do not travel.
• Keep the bottom 6′ of the portal legs dark — anyone on the apron throws a shadow up to about that height (§05).
• Feather the outer 12″ of every leg file and both cyc edges so the shadow and coverage margins land as a falloff rather than a line (§06).
• Keep the top of every leg file in the mid‑tones. Above 17′‑5″ only one of the two front‑of‑house projectors reaches the panel, so that band carries a 1.8× gain ramp baked into the file (§06). Anything near white up there will clip once the ramp is applied. It is also the least‑seen part of the panel — behind a border on pair 1, above eye height everywhere — so this costs the design nothing.
• You author one set of files, not two. Both Panasonics read the same tiles; only their warps differ (§07). Nothing in this section changes because there are two of them.

Every number in this document that came from a drawing rather than a tape measure. Confirm these before load‑in and the plot holds; skip them and you will be re‑warping in tech.