Reader John W writes:
Sony just made it plain that they’re not just investing in sensor dominance. See this press release on the founding of “Sony AI”:
https://www.sony.net/SonyInfo/News/Press/201911/19-118E/
It’s not only hard / expensive to build world-class imaging sensors. Cutting-edge software capabilities can also be very hard to bootstrap. Sony’s moat appears to be growing.
DIGLLOYD: Panasonic just quit the sensor business, which Sony dominates. There are not mamy players.
A company that does not invest in AI is a goner, whatever the industry. Though I’m still waiting for RI (rudimentary intelligence) in so many products.
Sony Announces the Establishment of Sony AI with the mission to unleash human creativity
... Sony AI will combine world class fundamental research and development with Sony's unique technical assets, especially in Imaging & Sensing Solutions, Robotics and Entertainment (Games, Music and Movies), driving transformation across all existing business domains and contributing to the creation of new business domains.
In addition, one of Sony AI's long-term goals is to contribute to the resolution of shared global issues extending beyond Sony's business domains. Sony AI will drive the research and development of AI in both physical and virtual space through multiple world-class flagship projects as well as other explorative research projects, including AI ethics.
Initially, Sony AI will launch three flagship projects in the areas of gaming, imaging & sensing, and gastronomy. The adoption of new AI technologies developed through these flagship projects will be critical to further enhancing the value of Sony's gaming and sensor businesses in coming years. This research will be pursued in close collaboration with the relevant Sony Group business units.
...
2019-12-04 - SEND FEEDBACK
Related:
computational photography, dynamic range, editorial, electronic shutter, Expose To The Right, exposure, focus stacking, focusing, frame averaging, golden age of photography, MTF and Micro Contrast, Robin D, shutterWhy should it even be a consideration when choosing a camera as to whether it has Eye AF or frame averaging support or focus stacking support or AF-ON separate from the shutter or... <insert_your_own_favorite>. Most such features should be taken for granted now, just like autofocus and exposure.
Further below, I list the things that could make cameras better for everyone, at no downside in other aspects of operation.
I hope that Nikon, Canon, Leica, Panasonic, Sony, Phase One, Fujifilm all read this page and formulate a plan to improve/add.
In many if not most of these cases, there is no credible rationale for not getting it done. The key shortcomings can be lumped into several categories:
- Features that improves sharpness.
- Features that improve image quality, including dynamic range.
- Design that simplifies the user experience e.g., eliminating all JPEG-centric stuff for those of us that shoot raw. Real customization, not just a My Menu.
- Features that raise the hit rate (e.g., Eye AF that really does focus on the iris of the eye)
- Addressing anything that reduces the chance of missing the shot or getting a sub-optimal picture in any technical way, e.g., auto-ETTR, deadly accurate autofocus.
- Features that automate challenges that need not be challenges (exposure, focus stacking, etc).
- Features that make it easy for the photographer determine that optimal image quality has been captured (exposure, dynamic range, sharpness, depth of field, etc), turning that into a guarantee in some cases (no verification needed).
- Computational photography of various kinds, which will surely kill off any camera maker that does not embrace it (Sony has publicly stated that it Sony is investing in AI for photography).
I have a deep well of ideas in my noggin, as many other photographers do. Why do camera makers apparently design without serious input? Not-invented-here syndrome? No one at any of the camera companies has ever contacted me for input. Ditto for other photographers I know—this is not about me but about listening to skilled photographers out there, and there are many. Why camera companies ignore such a rich resource is puzzling, because asking and listening are one inexpensive way to prioritize R&D—all it would take is tasking one person at the company to make it their mission.
Addressing these issues would make cameras more competitive and useful—much more in some cases yet without adding any complexity for those that don’t use the features (at least if designed right). Still, I (sadly) expect this list to be ignored by most or all camera makers.
Going forward, any camera make that makes no effort to address these issues in new cameras and firmware is going to get called out by me here on a regular basis with some kind of Clown Award, at least if the easy stuff is left unaddressed. There is no defensible reason for not getting much of this stuff right.
After all these years, anyone who understands programming and computers* or just usability alone can only throw up hands in frustration at the scattershot approach camera makers take to their feature set. It makes no sense either technically or for marketing, given how trivial some of these things are to add or fix.
Camera Functionality Deficits
You don’t have to be an engineer to scratch your head at the lack of some of this stuff in many or most cameras. Most of it is not hard.
| Feature / Behavior |
Difficulty |
Sony |
Nikon |
Fujifilm |
Canon |
Leica |
Self timer:
– option to remain enabled for power on/off and between shots
- at least supports 1/2/3/5/10/15/20/30/60 seconds
- supports user selectable time
- supports multiple frames at specified interval |
trivial |
5
- too few choices
- no flexibility |
5
- too few choices
- no flexibility |
5
- too few choices
- no flexibility |
5
- too few choices
- no flexibility |
1
- 2/12 seconds only |
| |
|
|
|
|
|
|
Long exposure:
- must not require remote release
- must support one press to start, stop
- supports dialing in any time in minutes/seconds
- must support arbitrary timed exposure min:sec
- counts time remaining for exposure, dark frame too |
trivial |
0
- cannot shoot longer than 30 seconds
- does not display time remaining
- cannot enter duration |
6
- does not display time remaining
- cannot enter long duration |
TBD |
TBD |
3
- - does not display time remaining
- cannot enter long duration |
| |
|
|
|
|
|
|
Lossless compressed raw format
- well established technology |
trivial |
0 |
10 |
10 |
10 |
0 |
| |
|
|
|
|
|
|
Save/restore settings
- save/restore settings to camera card
- simple text file for easy editing
- must also save My Menu |
trivial |
5 |
5 |
TBD |
5 |
0 |
| |
|
|
|
|
|
|
AF-ON support:
- supports release without refocusing
- support AF-ON independent of shutter |
trivial |
10 |
10 |
10 |
10 |
LeicaQ2: 0
Leica SL: TBD |
| |
|
|
|
|
|
|
True raw histogram
- histogram of actual raw data for each color channel (no RGB processing) |
easy |
0 |
0 |
0 |
0 |
0 |
| |
|
|
|
|
|
|
Frame averaging
- option to keep all
- option to store 2/4/8/N-frame averages
- option to
- full electronic shutter
- FAST |
easy |
|
|
|
|
|
| |
|
|
|
|
|
|
pixel shift
- FAST so checkerboarding declines as an issue |
moderate |
2
- generally unusable in the field |
0 |
0 |
0 |
0 |
| |
|
|
|
|
|
|
Aperture and focus, Live View including magnified:
- stop down option
- wide open option
- choosable aperture option
- toggle aperture while focusing
- avoids unpredictable aperture choice
- must support AF across frame
- allow focus anywhere in frame
- must have deadly-accurate mode |
easy |
5 |
7 |
5 |
TBD
- no control |
3
- no control |
| |
|
|
|
|
|
|
Focus stacking support:
- AF single-focus to INF
- AF focus near point to far point
- control over step size
- accounts field curvature for INF
_ must support focusing to frame edge
- must operate FAST
- analyze image and don’t do stupid stuff like 10 badly blurred frames past INF
- option to preconfigure near-to-far setups eg 1m to INF, 3.5m to INF, etc
- auto detect near/far blur boundaries and narrow step size locally |
moderate (basic)
challenging (advanced) |
0 |
6
- does not understand INF
- does not offer far point, keeps taking images way past INF
- relatively slow |
4
- serious bugs with tele lenses
- difficulty focusing in outer zones
|
0 |
0 |
| |
|
|
|
|
|
|
Auto detect camera movement in capture
- audible warning
- blur rating
- option to take up to N images keeping least motion M of N
|
moderate |
0 |
0 |
0 |
0 |
0 |
| |
|
|
|
|
|
|
Auto detect subject motion in capture
- audible warning
- blur rating with mask
- option to take up to N images keeping least motion M of N |
moderate |
0 |
0 |
0 |
0 |
0 |
| |
|
|
|
|
|
|
Auto-ETTR exposure mode
- best possible ETTR exposure
- adds EXIF flag for pull adjustment relative to 18% gray
- specular highlight support; allows 0.DD% blowout
- bias control for shadows or highlights when both cannot work
- electronic shutter
- retakes automagically if calculation wrong
- gets it right to 1/10 stop
- just works!
|
moderate |
0 |
0 |
0 |
0 |
0 |
| |
|
|
|
|
|
|
16-bit dynamic range
- high speed electronic shutter melds exposure brackets
- produce single raw file |
0 |
0 |
0 |
0 |
0 |
0 |
| |
|
|
|
|
|
|
AF accounts for focus shift:
- option to focus wide open but compensate
- option to focus stopped down
- option to focus at fixed aperture |
moderate |
5
cannot focus wide open and compensate for shooting aperture, no behavioral control |
5
cannot focus wide open and compensate for shooting aperture, no behavioral control |
5
cannot focus wide open and compensate for shooting aperture, no behavioral control |
TBD
no behavioral control |
TBD
no behavioral control |
| |
|
|
|
|
|
|
Auto select best sharpness of N exposures
- option to keep all
- mark/flag best exposures with ranking
- option to keep best M of N
- mode that show mask of what has high micro contrast in taken image |
challenging due to CPU speed |
0 |
0 |
0 |
0 |
0 |
|
|
|
|
|
|
|
Electronic panorama mode
- electronic shutter for quiet/fast
- works like iPhone |
challenging due to CPU speed |
1
no ES |
0 |
0 |
0 |
0 |
| |
|
|
|
|
|
|
multi-shot high-res mode a la Panasonic
- intelligent motion detection, mitigation |
challenging |
0 |
0 |
0 |
0 |
0 |
| |
|
|
|
|
|
|
Your editorial... just rammed it home to me how Sony, now with an utterly dominant position in the market, is now leading the industry backwards. You may not know of it but believe it or not Sony had a rough (and I mean it was a toy, a joke) focus bracket app in "PlayMemories" . The least said the better on that score, but it proved to me that Sony is brain dead when it comes to software.
They have great engineers I am sure, but they could do with some photographers and have the two groups talk to each on occasion, because as your list makes clear they are the worst of all your listed camera makers at including features that make the Photographers job easier, more professional, and more fun. Good grief it took them until a year ago to add an intervelometer on the A&something I had in my D300 in 2010 I love the output from the A7R3 but I can't and don't love the camera.
DIGLLOYD: let’s not forget ergonomics and haptics—while the Sony A7R IV is another improvement, it just doesn’t rise to the feel of the Nikon Z7.
I’m not sure Sony is leading things the wrong way so much as omitting things and what they do implement not done well (e.g. pixel shift). I wonder what Sony’s investment in AI will yield but hopefully not PlayBadMemories.
Mark A writes:
This is a great and well thought-out list and I hope this sparks mainstream manufacturers to improve engagement. I can't imagine how being a hardware or software engineer at these places is deeply pleasing if some of the most obvious things are either not occurring to them (pathetic) or being suppressed by management (sad).
In any case, quick comment is this: Hasselblad tried to do this with Ming Thein and I had expected more, as I'm sure he did, too. It may be worth at least recognizing that attempts have been made and maybe Ming Thein or other people can help figure out why such seem like "tissue rejections." I suppose https://www.red.com/ could be an example of a company that does it right? Corporate America (and Japan?) is pretty much the same with suppressed entrepreneurial spirit, no real meritocracy, HR-driven organizations and cultures, management with economically perverse incentives, etc.
One other thing just occurred to me as I'm writing this. If a camera's hardware is sufficient, perhaps community-driven software is an answer ala https://magiclantern.fm/ Great leadership is required to make that work really well but perhaps medium well would suffice to get people what they need?
DIGLLOYD: agreed, it’s quite possible that engineers are not allowed to follow sensible priorities. OTOH, it’s also possible that many otherwise talented engineers have nil talent or interest in rethinking approaches to long established assumptions. I observed that firsthand for 25 years in the professional world.
As for Ming Thein and Hasselblad... brilliant minds are a poor fit for just about any corporate culture unless it comes from the top (think Steve Jobs, Elon Musk, Jeff Bezos). Most of the time, corporations create a stultifying quagmire for the those capable of original thinking.
Things like MagicLantern are not a solution IMO, any more than Linux is a substitute for macOS. And in a world of hackers, cameras are already getting hacked.
The “20” is for the year 2020 (prior C1Pro is v12), Phase One apparently moving to an Adobe style system, ironic but makes sense too.
One readers suggests that C1 Pro 20 solves some bugs I’ve been seeing, so I’ve downloaded it and will be using v20 going forward.
See what’s new in Capture On 20.
Thomas Fitzgerald video has a good change summary with a video.
Some surprises I ran into:
- Shortcuts for Fit to Window and Actual pixels are no longer cmd-0 and cmd-option-0—kind of a crazy discrepancy if you use Photoshop heavily as I do. Fortunately keyboard shortcuts can be fixed/reassigned.
- I wasted 10 minutes getting nowwhere until I realized that C1 disables everything if the files are on a camera card—and doesn’t indicate what's wrong. It’s also very annoying if I want to do a quick check—files must be copied over first.
- My floating palette was wiped out and my palette setup was mangled. Not exactly a seamless transition.
- Export performance is exceptionally inefficient (as algorithmically sorry as it gets)—C1 serially processes files, so I see about one CPU core being used. This is a major performance loss on my 8 core iMac 5K.
- GPU usage is negligible for most operations. But that is also true of Photoshop.
Capture One 20
