Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple and Canon Announce Mirrorless Camera Lens Designed for Spatial Video Capture
- Thread starter MacRumors
- Start date
- Sort by reaction score
EOS VR Utility
Software that converts images captured by the EOS VR SYSTEM into VR 180° images in equirectangular projection that can be viewed and edited in a VR environment.
Wait, why does the paid subscription app required to output the spatial video require 2x the RAM for MacOS as it does for Windows? I thought Apple said the opposite, that 8 GB of memory on a Mac is just as good as 16 GB on a PC. 🤪
Neat. Stereoscopic lenses have been around since the dawn of photography, when they thought there'd be room in the market for 3d images, so it's interesting to see Apple Vision give the concept a future. I'd love to see a converter that'd process existing stereo images into the format, so we can finally give 3d images dating back to the 1800s a path forward. ...for the few thousand people who bought an Apple Viz i guess, but, assuming they come down to earth in the future...
In all likely hood, it won't be released. It's just a development announcement.
Also, it's "designed to capture spatial video for the Vision Pro headset." The Vision Pro effectively has 0% of the VR headset market when rounded to a whole number, and it's not available in every country. Canon might design the lens on paper, but they won't invest money into producing something that will only sell hundreds of units globally. Apple might pay for units to be produced, just as a marketing gimmick to make the VP seem like it's a real product.
With the iPhone I assume they were doing some computational analysis of the images to generate the 3D image pairs. (after all they were using two different focal length lenses using two of it's three cameras (wide and ultra-wide - I believe) to get a stereo pair). It's likely that with the Canon lens that there is also some post processing of the images involved.
Not an ideal way to get good quality realistic 3D images.
BTW - I was not really pleased with the quality of the 3D on their Parquor immersive video - the close-ups had a really un-natural look to them.
The lack of resolution on the immersive videos - wether because of the capture or display - was really a let down for me.
I'm assuming the capture end, because the computer generated stuff looks much better.
They released another VR lens with a price of around $1100, so… it wont be cheap but it wont be like the older 5.2 version which is around $2000 on sale.
Judging by the lens in the picture, it’s less glass than the 3.9, which often translates into lower cost, but have to assume this is going to be a low volume line, so higher price. Maybe it will all come out in the wash to be around the $1100?
This reviews a similar lens from canon that explains a number of technical details…
but essentially as others have said its a split sensor approach.
And on a related note… AVP 2.0 beta includes the computational spatial photo tech discussed in the keynote today, and i was pleasantly surprised at how good it was at converting my existing 2D collection of pics, especially portraits. This really could be a game changer, both hardware and software.
That's right. You take some pictures so the AI companies can scrape your date.
Don't worry, they will claim it's to stop CSAM or some other **** that puts you in an awkward position so if you complain you look like a nonce.
Never mind the fact that it's THEM who are. Taking everyone's pictures without permission. Taking everyones knowledge.
You'd be a fool to trust any of these Big American companies at this point.
Shower of **** the lot of them
Don't worry, they will claim it's to stop CSAM or some other **** that puts you in an awkward position so if you complain you look like a nonce.
Never mind the fact that it's THEM who are. Taking everyone's pictures without permission. Taking everyones knowledge.
You'd be a fool to trust any of these Big American companies at this point.
Shower of **** the lot of them
Cause the single sensor records a split image across two lenses. Quite simple.
It is easy to understand. The lens projects two images onto the sensor. If you looked at the unprocessed image it would look like two nearly identical images
So you have a fear of dual-lens cameras? How it is different from the same Canon camera but with a conventional lens?
It depends on the camera to subject distance. It would be OK for up to maybe 8 to 10 feet. (we don't know the exact specs yet)Didn’t think cameras so close would give a good 3D recording.
But with such a short focal length the camera to subject distance will be quite short. This lens would be good for a kind of "in your face style" videography. So it might be a good match for VR goggles.
Well, the guy certainly lives up to his username. There’s nothing AI about this lens, of course, but someone upthread did mention VisionOS 2 adding a feature to add depth to pictures via AI.
Of course, that kind of AI is probably going to be done on device, but that user probably doesn’t care about that one bit. I’m not particularly AI bullish, but I’m not an AI bear either (I just think it’s overhyped for what it is, and part of me wonders if we’re not due for another AI winter). But the response to AI has really given me insight into the mentality of the Victorian Luddites. Remember, the Luddites were opposed to industrial weaving. But industrial weaving gave better jobs to many people who would have been stuck in sustenance agriculture otherwise and led to so many other jobs developing, and Ludd’s followers just couldn’t see how it would play out, so they responded with fear (and aggression).
It doesn’t. They say it will work as minimum with 8GB Ram on both and have recommended as well. The recommendations are based on the models and general capability - as recommendationsEOS VR Utility
Software that converts images captured by the EOS VR SYSTEM into VR 180° images in equirectangular projection that can be viewed and edited in a VR environment.app.ssw.imaging-saas.canon
Wait, why does the paid subscription app required to output the spatial video require 2x the RAM for MacOS as it does for Windows? I thought Apple said the opposite, that 8 GB of memory on a Mac is just as good as 16 GB on a PC. 🤪
I assume it will be two separate images hitting each side of the sensor and there will be a specialized software to properly crop and use it to render 3D
Here is the 1930s version of the Canon lens. It does the same thing the new Canon lens does but uses film. It was unaffordable for most normal people. But, really the idea is about the same. It records two images in one frame.
Notice that this being a Leica, is also a "mirrorless' camera.
Notice that this being a Leica, is also a "mirrorless' camera.
