The "Intelligent Zoom" feature could make this a useful tool.
However, I don't buy their claim that their algorithm is AI. The threshold for any algorithm to classify AI is rather high and I doubt they come even close. IMHO, it is a classical image processing app (like optical character recognition which once was considered artificial intelligence -- but not anymore).
falconeyes: Is this the level of innovation Adobe thinks is acceptable to CC users?
It saves to click 4 corners with the mask tool and then use content-aware fill which was inroduced with CS5, I believe. Actually, it just makes a script a built-in feature.
Where is the real innovation? A new tool like content-aware fill as such?
The last innovative feature brought to PS was content-aware fill introduced in CS5. It is based on the PatchMatch algorithm (2009), with contributions from Princeton University, University of Washington and Adobe. 3 of 4 authors are now with Adobe.
I am sure there is enough imaging research going on. It is just Adobe isn't investing anymore in the core strengths of its product. Rather, they add bells and whistles like upload to the cloud etc.
I wouldn't. But I don't see noteworthy innovations brought to PS since a couple of releases now. Instead, I see "new features" which would normally go as actions.
Grig: I don't see destruction compared to FZ1000... Sorry... May be good idea to change it to "outperforms" or something like it if you really think so, but not destroys for sure... But to me FZ Leica lens is sharper and less CA... Especially on the sides at F8 at 400mm
I agree the "destroys" headline should remain reserved for findings which deserve such wording.
In this particular case, I would have appreciated to read "outperforms the competition".
Joe Ogiba: I have not purchased a HDD in years and all of my PCs and Macs have SSDs.
Most photographers do still rely on RAID5 HDD arrays for their >10TB archives.SSD arrays not only are more expensive, they do also fail in a less predictive way.
Is this the level of innovation Adobe thinks is acceptable to CC users?
Not a bad idea for a plugin or filter program.
However, I'll look forward to more deep testing of its mask tool vs. e.g., Topaz ReMask which would be my reference for the tree vs. sky example.
The example video with the couple vs. water created a terrible mask though.
Sean65: The use of a tripod in bright light and narrow streets is more than a little odd.
I know the places photographed in this film. Using a tripod (plus long coat in hotr weather btw) sets you apart from the crowd which is necessary to take good pictures. If you look closely, he tends to put down his tripod to then shot freehand with his Sony ;)
Which all makes sense if you know how tourists (when they stop walking) in, e.g., Fez, are approached all the time otherwise ...
saeba77: image the same quality sensor and similar AF in a mirrorless body as the A6xxx.Damn if only Nikon making that not cannibalize his own market:(
@Zerixos,agree with your arguments. Except that an EFCS (electronic first curtain shutter) removes the requirement to close shutter in LV first, without the mentioned negative impacts. The only constraint is that EFCS is hard to make for 1/2000s or shorter.
MrTaikitso: When Lytro launched their first (innovative but impractical) camera, I said Canon or Adobe should buy them. Neither did, and being they had so much venture finding ($200m if I recall) I knew they would survive to eventually get their business model right, and they have, as their remarkable machine announced at NAB shows. Canon make very good (high end) movie cameras, but have Sony and Panasonic on their tail, but most importantly of all, Canon's greatest technology has always been their AF. So it would have made sense for them to invest in a technology that makes focusing obsolete in specific use cases - namely Lytros.
Instead, they launched the unversatile unimaginative M series so Panasonic and Oly ate them for lunch.
I have never owned a Canon camera (love their scanners and printers!), but know how good their gear can be from other users and results, but they cannot rest on their past reputation.
Buy Lytro Canon & launch a revolutionary multi-focus APSC mirrorless.
@MrTaikitso: if you look closely, what Lytro does is not disruptive.
More precisely:Both Lytro (M=6) and Canon (M=1.4) use a Multipixel sensor with M^2 sensels per microlens. They are closer than you think and just use a technology originally introduced by Fuji (phase sensels). Multipixel sensors patents cannot be defended (obvious technology) and soon, all sensors will be of Multipixel type (cf. Sony and Samsung).
Personally, I think M=2 (quad pixel) is the sweet spot.
This news is welcome.Esp. in BIF this camera seems to be made for, backlit subjects are very frequent and the ability to push shadows is an important asset for the D500.
Donald B: I think you need to compare it to the k3ii at iso 6400 raw .
Made my day.Logic errors in arguments this large are rare :)
falconeyes: Thanks for the additional info.As a matter of fact, it means the Lytro Cinema is an example of the "Multi-Pixel AF Sensor" with 21 MP and Multi=6x6.The Canon Dual-Pixel AF Sensor is another example with Multi=2x1.For some time now, I advocate a broader adoption of Multi-Pixel AF Sensors, specifically Quad-Pixel with Multi=2x2.OTOH, it is a significant step-back from the original idea of a real Light-Field or plenoptical camera which requires a much higher sub-ray resolution being captured. But this stepped-back specification yields a much more interesting camera overall, still increasing the available depth of field by a factor six.
Just adding to it...We do not know yet the 35mm-equivalent properties of this camera. But as it looks massive, even F-Stop numbers of 6*N could imply shallow-enough depth-of-field to make this camera operate more like a conventional camera which requires constant focus operation.
Thanks for the additional info.As a matter of fact, it means the Lytro Cinema is an example of the "Multi-Pixel AF Sensor" with 21 MP and Multi=6x6.The Canon Dual-Pixel AF Sensor is another example with Multi=2x1.For some time now, I advocate a broader adoption of Multi-Pixel AF Sensors, specifically Quad-Pixel with Multi=2x2.OTOH, it is a significant step-back from the original idea of a real Light-Field or plenoptical camera which requires a much higher sub-ray resolution being captured. But this stepped-back specification yields a much more interesting camera overall, still increasing the available depth of field by a factor six.
I think the D5 is all about the kind of images you get, not the image quality of each. Don't forget, at daylight an RX100 easily matches the image quality of an DSLR. But the DSLR may catch images the RX100 will fail at.
In my German Nikon forum, the following image from the Bundesliga Bayern Munich game was posted (Player: Vidal) -> http://www.nikon-fotografie.de/vbulletin/picture.php?albumid=30124&pictureid=498687Credit Michael Weber, IMAGEPOWERD5 JPEG ooc, AF-S VR FL 600/4 + TC-14E III + strong crop at ISO 8000 f/5,6 1/1250sMr. Weber: "Image taken from across the playing ground, the opposite side of the goal and previously, I struggled with mild blur in similiar situations."That's the kind of images which do actually sell!Anything else, this DPR article included, can only give a weak impression of how the camera performs, or doesn't. In my German Nikon forum, Michael writes: "I am more than content (with the D5), the more the longer I work with it".
Robert Kempen: Leica reports record revenues for the past financial year with 'double digit' profits - unlisted company, of course, so does not need to provide exact profit numbers, but the revenue number is a new record for Leica's photographic division
Bottom line: M, Q & SL are selling like hot cakes and so is the new S
B&H must be smiling...
@Robert Kempen> unlisted company, maybe, but in Germany business does still follow the rules ;)-> https://www.bundesanzeiger.de/ebanzwww/wexsservlet?session.sessionid=f96ae031a172c833e3a010224fdbaecb&page.navid=detailsearchlisttodetailsearchdetail&fts_search_list.selected=eccf989a8a489f92&fts_search_list.destHistoryId=61614
They make profit around 38 and 14 million € (2014 and 2015 resp.) but don't shift any of the profit into the next year.
Most expensive tripod ever: microgravitation :)
falconeyes: This is a tiny project for a company the scale of FB. Cool but tiny.
Nevertheless, I am glad to see that undercover projects are still making it to the surface in large corporations :)
Tiny it is.Miniscule team and prototype-level hardware solution.Roughly equivalent to what a decent engineering university's thesis work would achieve.
RPJG: Does anyone know why 14 horizontal cameras are required, given the quite wide 19mm-equivalent field of view - e.g., for better stitching results?
@Mark BanasAssuming the standard 100° FoV (diagonal) of current VR glasses, 2x8k translates to about full HD per eye.
ProfHankD: In the past few years it seems everybody has discovered that you can stick an array of cameras together for 360 capture. The Surround 360 does not seem to be a particularly clever nor cheap way to construct the multi-camera, and I wouldn't call overnight stitching fast. As far back as 1999, my research group showed two different multi-camera 360-degree rigs on autonomous platforms, with real-time stitching and video wall pan/zoom on a Linux cluster (fed captures via 802.11 wireless), in our research exhibit at IEEE/ACM SC99: http://aggregate.org/EXHIBITS/sc99_360.jpg
That said, we didn't do stereo, and the stereo video processing and encoding is far from trivial. I applaud Facebook freely releasing the design and software for others to use and help improve. I look forward to the actual open source release, which they say will happen "on GitHub this summer."
I agree. I am loking forward to learn which algorithm exactly they adopted to compute the stereo disparities. That seems to be at the heart of the news ;)