Starship Just Photographed Itself From Space. Here's Everything That Actually Happened.

Added: 2026-06-03

[00:00:00]Last Friday, two satellites took a photograph of a rocket from space while that rocket was still flying. Not from the ground, not from a chase plane, from orbit. And here's the part that most people glossed over. Those two satellites, they were deployed by that same rocket. Minutes earlier, they had been sitting in its belly and now they were in space turning around, pointing their cameras back at the vehicle that had just released them and beaming the footage home in real time. That happened on May 22nd, 2026 when SpaceX launched Starship for the 12th time. But that's not actually the most important thing that happened that day. The most important thing is something quieter.

[00:00:37]Something that happened at 7:36 p.m.

[00:00:39]over the Indian Ocean that almost nobody is talking about. And by the end of this video, you'll understand exactly why it matters more than anything else SpaceX has ever done. To understand what happened on May 22nd, you need to understand what Starship actually is.

[00:00:54]because most coverage gets this wrong.

[00:00:56]People describe Starship as the world's biggest rocket. That's true, but it's almost beside the point. Starship is 121 meters tall. That's taller than the Statue of Liberty with her pedestal. It produces more thrust at liftoff than any rocket in history, more than the Saturn V that took humans to the moon. And it's designed to be fully and rapidly reusable, meaning SpaceX wants to fly it, land it, and fly it again the same day if possible. But here's the thing that actually makes Starship different.

[00:01:25]The economics. Every other rocket in existence is essentially a pipe bomb.

[00:01:29]You use it once and throw it away. That means the cost of getting 1 kilogram into space has a floor. You cannot get below a certain price because you're always building a new rocket. Starship breaks that floor. If SpaceX can reuse this vehicle the way they reuse the Falcon 9 booster dozens of times, the cost of reaching orbit drops by an order of magnitude. Not 20% cheaper, 10 times cheaper, maybe more. That changes what's possible in space. It changes what satellites cost. It changes whether a moon base is a science project or a viable infrastructure investment. It changes whether Mars is a fantasy or a plan. Everything depends on whether this rocket works. The last time Starship flew was October 2025. 7 months passed before flight 12. In the space industry, 7 months is a long time, and the reason for that gap tells you everything about what SpaceX was attempting. This wasn't an update. This was a near total redesign. SpaceX calls it version 3, Starship V3. And when they say thousands of changes, that's not marketing language. They mean it. Start with the engines. The new Raptor 3 engine, the engine that powers both the booster and the upper stage is fundamentally simpler than its predecessor. Previous Raptor variants had complex plumbing, external heat shields, exposed components. Raptor 3 removes all of that. The insulation is built in. The pipes are rerouted. The whole engine is cleaner, lighter, and designed to be manufactured faster. Each Raptor 3 saves approximately 1 ton of mass at the vehicle level compared to the previous version. There are 33 of them on the booster alone. The booster itself changed, too. Previous versions had four grid fins. The wing-like structures that steer the rocket during its return fall to Earth. The V3 booster has three, but each one is 50% larger and structurally stronger. Fewer parts, less complexity, more robustness. The upper stage called ship got its heat shield overhauled. This is the layer of tiles on the bottom of the vehicle that protects it from the 1,600° plasma of re-entry. Every time Starship comes back to Earth at orbital speed, that heat shield is the only thing standing between a successful landing and a fireball. Getting it right is not optional. And then there was the launchpad. SpaceX built an entirely new one, orbital launchpad 2 or OLP2. The old pad used hydraulic actuators to control the giant robotic catch arms they call Mechazilla. The new one uses electromechanical actuators instead. No hydraulic fluid, theoretically more precise, more reliable, but also completely untested at full operational scale. They were launching a brand new rocket from a brand new pad with brand new engines for the first time. The original launch date was May 20th. It didn't happen. During the countdown, a water diverter system beneath the pad triggered an automatic hold. The water diverter is essentially a flood system.

[00:04:18]It releases enormous quantities of water onto the pad at the moment of ignition to absorb the acoustic energy from the engines. Without it functioning correctly, the shock waves from 33 Raptor engines firing simultaneously could damage the rocket before it even leaves the ground. The hold gave engineers time to look at the data. What they found was a hydraulic pin on one of Mechaz-Zilla's catch arms that failed to retract on the new electromechanical pad. This was exactly the kind of first time integration issue they'd been prepared for. But prepared for doesn't mean insignificant. They scrubbed reset for May 22nd. The 48 hours between the scrub and the rescheduled launch were not quiet. The rocket had to be detanked. All the cryogenic propellants, liquid oxygen and liquid methane, had to be safely offloaded and then reloaded.

[00:05:04]The PAD team worked through the night.

[00:05:05]Engineers reviewed data. The weather window held at 5:30 p.m. Central time on May 22nd, 2026. They tried again. When Starship lifts off, it does not feel like a normal rocket launch. Most rockets climb fast and clean, a needle punching through the sky. Starship is different. It's so massive, so heavy, so laden with propellant that in the first seconds it seems to barely move. It hangs above the pad for a moment. fire and steam and 74 megs of force before slowly, unmistakably beginning to rise.

[00:05:38]All 33 Raptor 3 engines on booster 19 ignited at liftoff, clean, no shutdowns, no anomalies. The vehicle climbed through the atmosphere. The booster burned and then at approximately t plus 2 minutes and 45 seconds, something called hot staging happened. Hot staging is exactly what it sounds like. Before the booster's engines shut down completely, the upper stages engines, the ship's six Raptor 3 engines, ignite while the two vehicles are still physically connected. The upper stage essentially pushes itself off the booster while both sets of engines are firing simultaneously. It's violent.

[00:06:12]It's aggressive. It's faster and more efficient than the traditional approach of waiting for the booster to cut out before separating. And it worked. Stage separation confirmed. Ship 39 continued upward. Booster 19 began its return. And that's where the two stories diverge.

[00:06:27]After separation, ship 39 kept climbing until it reached SECO, Space Engine Cutoff, the point where the engines stop and the vehicle is coasting in space on a suborbital trajectory. On this flight, ship 39 was carrying 22 payloads. 20 of them were Starlink satellite simulators, dummy versions of SpaceX's next generation V3 Starlink satellites, roughly the same size and mass as the real thing. Flying them in this configuration lets engineers validate that the deployment mechanism, the release sequencing, and the trajectory all work as planned before you risk actual hardware. The other two were different. SpaceX calls them dodger dogs, two actual modified Starlink satellites equipped with imaging cameras. Their job was to fly ahead of ship 39 after deployment, turn around and photograph the vehicle from space.

[00:07:13]This had never been done before. Think about what that means operationally.

[00:07:18]SpaceX released these satellites from the ship, tracked their position relative to the vehicle, received their camera data in real time, and use that imagery to inspect Starship's heat shield from orbit. Why does heat shield inspection from orbit matter? Because the longerterm plan is to fly Starship not to the ocean, but back to the launch site to catch it with the Mechazilla arms, turn it around, refuel it, and fly it again within hours. To do that safely, engineers need to be able to assess heat shield condition while the vehicle is still in space. If tiles are damaged, you need to know before re-entry, not after. The Dodger dogs were a proof of concept for that inspection capability, and they worked.

[00:07:57]For the first time in the history of the Starship program, engineers watched footage of their rocket in space, captured by satellites that rocket had just released. While ship 39 was deploying payloads, booster 19 was already falling back toward Earth. The plan was a controlled splashdown in the Gulf of Mexico. Not a Mechaz-illa catch, SpaceX had explicitly decided not to attempt catching the booster on this flight. This was the first flight of a completely redesigned vehicle, and adding a precision catch to that list of objectives would have been unnecessary risk. What they needed the booster to do was separate cleanly, execute a boost back burn to reverse its trajectory, and descend in a controlled manner to a planned splashdown zone off the coast of Starbase. The separation was clean. The boost back burn started, then it stopped early. SpaceX engineers watching the telemetry saw the burn terminate ahead of schedule. The booster wasn't generating enough deceleration.

[00:08:52]Observers tracking the vehicle noted it wasn't following the expected descent arc. Aerospace journalists covering the liveream noted the same thing in real time. SpaceX later confirmed that multiple Raptor 3 engines failed during the boostback burn. The exact number hasn't been publicly disclosed, but enough engines failed that the booster lost the controlled descent it needed.

[00:09:12]Booster 19 came apart over the Gulf of Mexico. SpaceX streamed the entire event. They didn't cut the feed. They showed everything, including the loss.

[00:09:21]>> It looks like we just had an early boost back shutdown. Again, a reminder, the booster was planned to essentially head into the Gulf uh for a splashdown, but it has shut its engines down early on into boostback.

[00:09:35]>> And that choice matters because it signals something important about how SpaceX treats test data. A failure that you can observe and measure is infinitely more valuable than one you can't. The critical question now is whether the engine failures were a design issue with Raptor 3, something systemic that will appear on every future flight or an isolated anomaly. in this specific hardware. That distinction will determine how quickly SpaceX can close the gap and catch the next booster. Here is where most coverage of flight 12 lost the thread. The booster failure was the dramatic moment. It made headlines. It was the visible failure, the piece of hardware you could watch break apart on a live stream. So that's what the news cycle focused on. But the more important event happened 2 hours later over the Indian Ocean in near silence. Ship 39 executed a banking maneuver during re-entry. A deliberate sideways sweep through the upper atmosphere designed to mimic the approach profile that a future vehicle would need to fly to return to the launch site. This was rehearsal for the catch, not the catch itself, the pattern that would eventually lead to it. Then it descended through 1,600° plasma through the region where the heat shield does its job or doesn't. Where past vehicles either held together or didn't.

[00:10:47]It held together. At approximately 7:36 p.m. Central time, ship 39 touched down in the Indian Ocean upright under what SpaceX described as controlled descent.

[00:10:58]It then toppled as all ocean landings do because the vehicle is topheavy once the engines stop and burned, which was expected, which was planned. But before the topple and the burn, for a few seconds, the most powerful upper stage ever built was sitting upright on the surface of the Indian Ocean. The heat shield had worked. The engines had worked. The re-entry had worked. On the first flight of a completely new vehicle with a completely new engine after seven months of complete redesign, Elon Musk called it epic. SpaceX president Gwyn Shotwell said, "Our collective next among the stars has become so much closer." Those aren't typical post-launch remarks. Those are people who know exactly what just got proven.

[00:11:37]Here's the context that puts all of this in scale. NASA has selected Starship as the human landing system for Artemis 4th, the mission that is supposed to land astronauts on the moon. Not a backup plan, not a parallel program. The plan, the current target for that crude lunar landing is 2028. That's 2 years from now. And for that to happen, SpaceX still needs to demonstrate inorbit propellant transfer, the process of refueling Starship in space so it has enough fuel to land on the moon and come back. They need to demonstrate a full orbital flight. They need to demonstrate docking with NASA's Orion capsule. None of those things have happened yet, but none of them can happen until the vehicle itself works, until it can reach space, deploy payloads, survive re-entry, and land safely. Those are the prerequisites. And as of May 22nd, 2026, all four of those boxes are checked.

[00:12:30]Flight 12 didn't complete the mission, but it proved the foundation. It proved that Starship V3 can fly under engine out conditions. Ship 39 reached SECO on five of its six engines, not six. One engine failed in flight and the ship still made it, still deployed its payloads, still survived re-entry, still landed. That is what robustness looks like. And it matters enormously because every subsequent milestone, the orbital flight, the refueling, the moon landing is built on top of that foundation.

[00:13:03]There's one more piece of context. 2 days before flight 12, Elon Musk announced that SpaceX is going public.

[00:13:10]The IPO is targeting June 2026. The timing isn't a coincidence. A company whose flagship rocket just demonstrated survivable re-entry, firstofind payload deployment, and orbital photography on its first flight of a new design is a very different company to investors than one whose last two test flights ended in mid-air explosions. Flight 12 wasn't just a test flight. It was also a proof of concept for a business. The most powerful rocket ever built photographed itself from space. Its heat shield survived. Its payloads deployed. Its ship landed. The booster is gone. Lost to the Gulf of Mexico. Taken down by engines that failed at the worst possible moment. But here's the thing about SpaceX's approach to testing. They don't need every flight to succeed. They need every flight to teach them something. And flight 12 taught them a great deal about Raptor 3's failure modes, about the new pad, about heat shield performance, about what a V3 vehicle can survive. The next booster is already being built, the next ship is already in production, and NASA's clock is running. This is the channel for the science that doesn't get the coverage it deserves. If that's what you're looking for, subscribe. There's a lot more coming.