Internal 32-bit float stereo paired with lossless 120mm reach from a 20mm lens — all inside a genuinely pocket-sized 10-bit 1-inch sensor gimbal camera?

[Emanuel]

Quote

•  

Originally posted in another thread, but given what it is, I think it deserves a place of its own.

There’s something very real happening here right now. This is not just a minor upgrade.

: )

Insta360 sample for focal length range.

source 

(from Leica HQ BTW)

 

And that detachable screen is basically an on-set field monitor. WOW What a killer combo : X

[eatstoomuchjam]

Starting at $770!! 

🤣🤣🤣🤣🤣🤣🤣

Guess they saw GoPro's pricing on the Mission 1 series and said "hold my beer."

 

[Emanuel]

Well, for the price of a good but still affordable zoom lens, I still find it quite a deal! ; )

And AFAIK, it’s the first gimbal camera you can simply lock:

No idea whether the Osmo Pocket 4P will bring that too, but the non-Pro version doesn’t seem to.

EDIT -- I had heard that the Pocket 4 didn’t have it, but as far as I’ve been able to confirm now, they’ve introduced it across both new Pocket 4 models. Looks like the battle between the two brands has brought some benefits for their customers... ;- )

[QuickHitRecord]

I've never been tempted by one of these smaller pocket gimbals, but the image quality actually looks really nice. It doesn't look over-sharpened or GoPro-like at all. Of course, it's Brandon Li, who can make anything look good. The detachable remote is also really clever.

I want my next short to be captured guerilla-style in the city. I was originally considering the hacked EOS-M, but this has me re-thinking my strategy. I just really wish it had a mic input so that I could capture timecode.

[kye]

To me the zoom is the most significant function here.  This is meant to be a walk-around film-the-surroundings camera and my experience is that everyone who rigs a camera for this purpose uses a zoom with pretty significant zoom lens.

The biggest con of this is the fact it's a gimbal, and therefore stabilises rotation but not position, leading to the dreaded bobbing movement and foreground parallax errors.  These might be my candidate for the least cinematic image attribute of all time (linked to timestamp):

This is why gimbals need the fourth axis for walking, and why people don't shoot gimbal shots with any foreground in them.  By applying less stabilisation you end up with a more stable looking shot because the stabilisation doesn't call so much attention to itself.

[Emanuel]

Yes, indeed, but a few third-party brands offer that simple, inexpensive accessory, and it can easily be added to your setup.

 

[Emanuel]

Impressive if nothing else ; )

And here’s another feature that is, undeniably, not only pure fun:

...but just useful enough to say the least : X

[Emanuel]

Breaking news. The product is so obvious no-brainer, DJI has decided to declare a legal war against their most serious threat:

https://petapixel.com/2026/06/11/dji-is-suing-insta360-for-violating-multiple-osmo-pocket-patents/

The apocalyptic precedent is Kodak v. Polaroid, where Kodak was effectively pushed out of instant photography, but that was a much deeper, ecosystem-level disaster, not just another Tuesday in consumer electronics litigation. 

DJI has itself been through patent warfare with Autel over drones. And Insta360 recently faced GoPro in a camera-related dispute that did not exactly erase its current lineup from existence. 

Apple and Samsung spent years throwing patent grenades at each other before settling. 

[Emanuel]

This is where computational photography becomes impossible to ignore and where it starts to look more like computational cinematography.

These Luna Ultra low-light comparisons are not simply about sensor size, ISO or aperture. They are about the entire imaging pipeline. The Luna is not just “seeing better” in the traditional optical sense; it is processing harder. It is lifting shadows, stabilising exposure, reducing noise, rebuilding colour, brightening faces and subjects, and delivering an immediately usable image in situations where more traditional camera pipelines can look extremely dark straight out of camera.

The Sony FX3 comparison is especially revealing. The FX3 is a serious full-frame low-light video/cinema camera, here shown in 4K S-Log 3 at ISO 12,800, f/2.8 and a 180° shutter. That is not a weak setup. The Canon R6 V comparison is also striking, with the camera shown in 4K C-Log 2 at ISO 25,600, f/2.8 and 1/50. Yet, in these posted frames, both traditional camera images look dramatically darker than the Luna Ultra low-light mode.

Of course, this does not mean the Luna is “better” than an FX3 or a Canon cinema-style camera in any broad professional sense. It is not. Log footage is not supposed to be judged as a finished image before grading, and larger-sensor cameras still have major advantages in colour depth, dynamic range control, lens choice, codec robustness, monitoring, rigging and production workflow. Properly exposed, lit and graded, they remain much more flexible tools.

But that is not really the point. The disruptive point is that the Luna is delivering a visible, usable image immediately, in situations where a traditional cinema pipeline still expects the operator to expose carefully, light when necessary, grade later and finish the image in post.

The iPhone comparison makes the argument even more interesting, because a phone is already a computational camera. Smartphones built their reputation by using software to overcome small sensors: multi-frame processing, HDR, tone mapping, noise reduction, subject detection and AI-assisted reconstruction. So when the Luna Ultra in 4K low light Mode produces a dramatically more visible image than the iPhone 17 Pro Max 1x lens in this kind of scene, the story is no longer just “computational camera beats traditional camera”. It is “a dedicated computational video camera can beat a flagship smartphone at its own game”.

That distinction matters. Smartphone night photography and smartphone low-light video are not the same thing. Phones can do extraordinary things with stills because they can merge frames, hold exposure, stabilise aggressively and tolerate a more processed photographic look. Video is harder. It has to work continuously, frame after frame, without destroying motion, faces, texture and temporal consistency. In this example, the iPhone image remains very dark, while the Luna Ultra clearly prioritises subject visibility, colour recovery and immediate usability.

So the real lesson is this: low-light performance is no longer just a hardware contest. It is sensor plus lens plus stabilisation plus temporal denoising plus tone mapping plus AI reconstruction plus exposure strategy. The most useful image may come not from the device with the largest sensor, but from the device with the smartest pipeline.

The Luna does not replace cinema cameras. It does not replace phones either. But it attacks the space between them: fast, handheld, available-light video where the most important question is not “which camera has the biggest sensor?”, but “which camera gives me the shot I can actually use?”

And in these comparisons, the answer is surprisingly uncomfortable for devices that are much bigger, much more expensive, and much more established.

source