Prefabbing bullseye dormers vs. framing on the roof?

I’ve tried it both ways and honestly, stick-framing a bullseye while standing on a 10/12 pitch is a recipe for a headache. Last summer we did a set of four using the old-school method of cutting the headers and trimmers in place, and we spent half our time just chasing the curve with a jigsaw and a belt sander. Trying to get those tight radii to line up with the common rafters while balancing on a walk-board is never as precise as you want it to be. Your ground-built jig approach is the way to go, especially with the foam inserts. We used a similar setup with some 3/4-inch marine plywood templates to keep the circular frame from racking during the lift. While the crane cost hurts, you probly saved forty man-hours of fiddling with shims and bracing at height. I’ll take a few hundred bucks in rigging fees over three days of precarious chisel work any day. The precision you get on the floor means your flashing and trim go on much cleaner later, wich is where the real money is usually lost on these custom details.

I have to disagree on the pre-build being the cleaner route. Every time I've seen a shop-built bullseye show up on-site, we spend half a day shimming the base becuase the roof plane isn't as flat as the subfloor where it was mocked up. That crane rental alone kills the margin on smaller residential jobs.

Instead of building a heavy unit, we've had success stick-framing them using a series of CNC-cut 3/4-inch birch plywood ribs. We set our trimmers and headers like usual, then sister the main circular frame directly to the rafters. By using thin strips of Luan or bendable plywood for the barrel vault, you can adjust the radius on the fly. It's alot easier to pull a string line to ensure the dormer face is perfectly plumb with the exterior wall when you aren't fighting a two-hundred-pound pre-assembled box. Going the 'skeleton' route means two guys can handle the whole install with nothing more than a few ladders and some GRK structural screws. You lose a bit of the controlled enviornment of the shop, but you gain the ability to make sure the transition into the shingles is acutally tight. For me, avoiding the logistical nightmare of a crane is always the win.

I skipped the heavy framing and used layers of 1/4-inch luan to build up the barrel on-site. I found that if I cut my circular headers out of triple-layered 3/4-inch ply on the ground, I coud just haul those up and drop them onto the rafters. It gave me the stability of a jig without the weight of a whole unit. The real trick was using a heat gun on the PVC trim later; it saved me from fighting the wood grain on those tight curves. If you go the pre-built route again, maybe try skeletonizing the frame to keep the weight down so you can skip the crane and use a material lift instead.

Back in the 80s, we didn't have CNC or fancy foam jigs. We just used a trammel point to scribe our radii on a double stack of 2x12s and cut them with a sharp ship auger and alot of sweat. I’m with the ground-build crowd, but the trick to avoiding that crane fee is to build the skeletons light and skin 'em once they're pinned. If you’re worried about the roof plane being out of whack, just leave your tail cuts long and scribe them to the rafters in situ. It beats hauling a thousand-pound box up a ladder any day.

You're overthinking the weight. I always throw togehter a simple cradle with some LVL scraps and use my Festool Carvex to get those radii spot on before they ever leave the deck. The real bottleneck isn't the framing, it's the radius window frame itself. If you aren't using Flex Trim or a custom Marvin unit, you're going to fight that curve for days. I’ve found that using 3M All Weather Flashing Tape 8067 is the only way to get a tight seal on those circular sills without it bunching up like a cheap tarp. Don't worry about the crane—it's cheaper than a leak.

Quick update — that idea about using triple-layered 3/4-inch ply for headers relly stuck with me. I tried mocking up a lighter version on a scrap section to see if we coud ditch the heavy crane and just manhandle the frames into place. It seems way more stable than the foam jig. Do you guys find that the luan layers hold up well against moisture during the build?