ProfHankD

ProfHankD

Lives in United States Lexington, United States
Works as a Professor
Has a website at http://aggregate.org/hankd/
Joined on Mar 27, 2008
About me:

Plan: to change the way people think about and use cameras by taking advantage of cameras as computing systems; engineering camera systems to provide new abilities and improved quality.

Comments

Total: 941, showing: 61 – 80
« First‹ Previous23456Next ›Last »
In reply to:

PKDanny: CRAZY PRICE!!!

L is for Luxury. Don't think that's right. This is a workhorse for poster printing... but, hey, whatever they think will make it sell better.... ;-)

Link | Posted on Mar 9, 2016 at 04:11 UTC
In reply to:

PKDanny: CRAZY PRICE!!!

Actually, pretty normal for such things. My E-format (44") HP was something like $4500 back around 1999/2000. Nice for printing posters, and at Kinko poster-printing prices, paid for itself in just a few conferences worth of research exhibits. The catch is that they're not so good if you only use them every so often. I used mine for maybe 25-50 posters a year, nearly all printed in the one month before the primary conference my group exhibits at, which meant that the inks would dry out and it would take printing maybe ten feet of test pattern to get 'em going again.

That said, not sure why the red line is there... is that now Canon's generic way of saying "expensive stuff you should be impressed by?" ;-)

Link | Posted on Mar 9, 2016 at 03:16 UTC
On article Sony introduces Cyber-shot DSC-HX80 30x travel zoom (79 comments in total)
In reply to:

falconeyes: That's F/36 (equivalent) at the long end. The Raleigh-limit for max. possible resolution (i.e., at vanishing contrast with a perfect lens) at F/36 is 6 MP.

Sony advertizing a 6 MP camera as 18 MP should be reported to market watch agencies.

Otherwise, this is a tiny camera which certainly can provide a lot of fun and usable images up to HD quality.

Rayleigh limit is an angular resolution limit based on just-distinguishable PSF overlap, and it is 1.22 * wavelength * aperture, again as actual f/number -- which is precisely different from Fraunhofer by a linear factor of 2, which means 2x2 for area. That still doesn't come to 6MP by any numbers I plug in and, in fact, I already had a 2x2 factor in the 2MP number to allow for a Nyquist sampling of the Fraunhofer value (which is basically why the linear factor of 2 difference was there). BTW, in the worst case, it can be argued that another factor of 2x2 is needed due to the Bayer matrix, which would bring us to about 8MP at the long end of the zoom and to about 18MP at the short end.

Sampling above Nyquist is a good thing, and there is very little reason to believe that a 12MP sensor would have any better IQ than an 18MP one correctly scaled to the same resolution (e.g., using SINC resampling) -- at least not using a sensor with a high fill factor, which Sony BSI sensors have.

Link | Posted on Mar 8, 2016 at 11:35 UTC
On article Sony introduces Cyber-shot DSC-HX80 30x travel zoom (79 comments in total)
In reply to:

falconeyes: That's F/36 (equivalent) at the long end. The Raleigh-limit for max. possible resolution (i.e., at vanishing contrast with a perfect lens) at F/36 is 6 MP.

Sony advertizing a 6 MP camera as 18 MP should be reported to market watch agencies.

Otherwise, this is a tiny camera which certainly can provide a lot of fun and usable images up to HD quality.

I don't know how you're doing that math. The Fraunhofer diffraction central spot size is 2.44*wavelength*aperture (the actual aperture). When you plug-in the numbers, at f/6.4 the size sensor used ought to be sampling more like 2MP (roughly 1080 video) for green light.

Is that false advertising? Nope. Sampling beyond Nyquist is good, and Sony does use all 18MP for sampling. False advertising would be claiming the lens delivers 18MP resolution, which I'm sure they don't claim... and some less-careful competitors sometimes have made claims like that. ;-)

As for me, I'm just shocked that they can do all this in a pocketable camera with a pop-up OLED EVF for under $400 MSRP! That's truly shocking.

Link | Posted on Mar 8, 2016 at 01:24 UTC
In reply to:

ProfHankD: I don't really see the appeal of such strong ND filters -- for the technical reason that they are very likely to reveal minor light leaks, etc. I have other methods for simulating long exposures in bright light, namely my TDCI (time domain continuous imaging) research and simple stacking of shorter exposures, both of which can produce higher-quality results than use of an ND filter.

TDCI is still NSF-funded university research... but we're getting close to having public domain software out for it. Probably first consumer-useful stuff out sometime this summer; it's particularly nice with things like the 960FPS video mode on the Sony RX100 IV. The basic idea in TDCI is that the image data is converted into a set of continuous waveforms specifying how light level varies for each pixel over time and the exposure interval to be represented by an image is determined after the fact by integrating the area under each pixel's waveform for the given time interval.

Link | Posted on Mar 5, 2016 at 23:37 UTC
On Connect post What a view: Aukey Super Wide Angle lens quick review (64 comments in total)
In reply to:

Jahled: Those samples are absolutely dreadful

The ones from the $15 (in 2006) Sakar camera don't look great because the camera is terrible (just a 184-pixel diameter image circle), but on a 7MP camera from 2006, you get a 4MP circle that's already comparable; look at this image from the paper: http://aggregate.org/dit/peepfish/IMG_8850.png

On a modern cell phone, you do much better with a door peephole. The primary defect is field curvature, and the tiny sensors in cell phones have enough extra depth-of-focus to repair that. In fact, the $1 cell phone fisheye converters I've gotten via eBay (the aluminum ones, not the jelly ones which are truly abysmal) also do at least as well as the images shown in this article. I didn't have any modern images handy, but can dig some up/shoot new if people really have any doubt.

BTW, the door peepholes generally overstate view angle too. A typical 200-degree one delivers about 170 usable view angle.

Link | Posted on Mar 5, 2016 at 14:38 UTC
On article What a view: Aukey Super Wide Angle lens quick review (64 comments in total)
In reply to:

Jahled: Those samples are absolutely dreadful

The ones from the $15 (in 2006) Sakar camera don't look great because the camera is terrible (just a 184-pixel diameter image circle), but on a 7MP camera from 2006, you get a 4MP circle that's already comparable; look at this image from the paper: http://aggregate.org/dit/peepfish/IMG_8850.png

On a modern cell phone, you do much better with a door peephole. The primary defect is field curvature, and the tiny sensors in cell phones have enough extra depth-of-focus to repair that. In fact, the $1 cell phone fisheye converters I've gotten via eBay (the aluminum ones, not the jelly ones which are truly abysmal) also do at least as well as the images shown in this article. I didn't have any modern images handy, but can dig some up/shoot new if people really have any doubt.

BTW, the door peepholes generally overstate view angle too. A typical 200-degree one delivers about 170 usable view angle.

Link | Posted on Mar 5, 2016 at 14:38 UTC
On article What a view: Aukey Super Wide Angle lens quick review (64 comments in total)
In reply to:

Jahled: Those samples are absolutely dreadful

There are $4 door peepholes that produce better IQ... and I'm not kidding: http://aggregate.org/dit/peepfish/

Also, they say 238 degrees, but that isn't what these images seem to show. In fact, it looks like less than 180 diagonal. The shot with the buildings behind the guy doesn't come close to showing all the way up/down the street, which 180+ should (238 diagonal should be >180 horizontal).

DPReview: if you got hands on one of these, don't just quote the manufacturer, please measure the actual view angle.

Link | Posted on Mar 5, 2016 at 13:25 UTC
On Connect post What a view: Aukey Super Wide Angle lens quick review (64 comments in total)
In reply to:

Jahled: Those samples are absolutely dreadful

There are $4 door peepholes that produce better IQ... and I'm not kidding: http://aggregate.org/dit/peepfish/

Also, they say 238 degrees, but that isn't what these images seem to show. In fact, it looks like less than 180 diagonal. The shot with the buildings behind the guy doesn't come close to showing all the way up/down the street, which 180+ should (238 diagonal should be >180 horizontal).

DPReview: if you got hands on one of these, don't just quote the manufacturer, please measure the actual view angle.

Link | Posted on Mar 5, 2016 at 13:25 UTC
In reply to:

ProfHankD: I don't really see the appeal of such strong ND filters -- for the technical reason that they are very likely to reveal minor light leaks, etc. I have other methods for simulating long exposures in bright light, namely my TDCI (time domain continuous imaging) research and simple stacking of shorter exposures, both of which can produce higher-quality results than use of an ND filter.

I just posted a DPReview challenge: ND, or not too ND?

It takes a while to process challenges, but it should be announced March 10 and open the following week. The idea is to get people posting shots done with various strengths of ND filters and stacking. For example, 1@ND15 would be 1 shot taken with an ND15 filter and 5@ND3 would be 5 shots stacked (averaged) using an ND3 filter. Maybe we'll get an actual answer as to which works best?

In theory, stacking should produce higher IQ, with a much improved SNR. However, if the temporal gaps between stacked exposures are significant, or if the shutter firing multiple times causes camera movement (for which your stacking software doesn't compensate), it is possible that a longer sequence of shots might produce artifacts that wouldn't be seen using a stronger ND filter. Hopefully, we'll get this question answered over the next few weeks.... ;-)

Link | Posted on Mar 5, 2016 at 13:04 UTC

I don't really see the appeal of such strong ND filters -- for the technical reason that they are very likely to reveal minor light leaks, etc. I have other methods for simulating long exposures in bright light, namely my TDCI (time domain continuous imaging) research and simple stacking of shorter exposures, both of which can produce higher-quality results than use of an ND filter.

Link | Posted on Mar 5, 2016 at 05:13 UTC as 28th comment | 5 replies
In reply to:

ProfHankD: Did Canon say something? Nope.

The "read between the lines" is that Canon can't figure-out how to make really new products while keeping their very cost-effective standard guts in them. Sadly for them, they no longer have the infrastructure to compete with engineering-intensive companies like Sony.

My suggestion to Canon is to embrace the main sources of their innovation over the last decade: CHDK and ML. If Canon can't afford to do the research and development, especially in software/firmware, there are people in the open source world who can and are willing to, including academics like me. All Canon has to do is decide to support rather than ignore/frustrate these efforts....

To Chris Yates:

Isn't every academic looking for funding? ;-) Although not necessarily funding from Canon; I've had a variety of funded research efforts that have tried to use Canons and been thwarted by Canon's lack of support.

Link | Posted on Mar 2, 2016 at 15:34 UTC

Did Canon say something? Nope.

The "read between the lines" is that Canon can't figure-out how to make really new products while keeping their very cost-effective standard guts in them. Sadly for them, they no longer have the infrastructure to compete with engineering-intensive companies like Sony.

My suggestion to Canon is to embrace the main sources of their innovation over the last decade: CHDK and ML. If Canon can't afford to do the research and development, especially in software/firmware, there are people in the open source world who can and are willing to, including academics like me. All Canon has to do is decide to support rather than ignore/frustrate these efforts....

Link | Posted on Mar 2, 2016 at 05:02 UTC as 48th comment | 14 replies
On article Opinion: Enthusiast compacts have finally come of age (494 comments in total)

Re: the Purse Test

My wife doesn't like to carry a camera (beyond her cell phone). However, my camera bag can fit inside her "purse" -- many women's handbags are the size of small backpacks. ;-)

The real test is pockets, although they vary a lot too....

Actually, the really important test is invulnerability to accidentally being turned-on resulting in the lens attempting to extend and stripping gears when it hits other stuff in whatever purse/pocket/bag contains it.

Link | Posted on Feb 28, 2016 at 17:06 UTC as 106th comment

I can't wait to see the camera in image 5 on eBay:

"New Other: Minty fresh Fujifilm X-Pro2, never touched by human hands. From a non-smoking no pets environment."

Link | Posted on Feb 28, 2016 at 16:34 UTC as 16th comment
On article Opinion: Enthusiast compacts have finally come of age (494 comments in total)

In the early days of digital cameras, the strange truth was that the IQ of the best fixed-lens cameras often beat that of DSLRs with kit zooms under friendly circumstances. For example, in optimal lighting, the Canon PowerShot S70 easily and disturbingly out-resolved and out-IQ'd the Canon EOS 300D with the kit zoom (we tested both extensively for a 3D capture project I was working on at the time).

Last year, I bought a Sony RX100 IV to use in another research project that needed the 960FPS video. I expected the IQ of that advanced 1" sensor under good lighting to blow away my old, lower resolution, APS-C NEX-5. It doesn't.

The RX100 IV is an awesome camera capable of producing museum-quality images, but even the very best 1" sensors seem to be a step behind the IQ of good APS-C. BTW, I don't see as clear a step between APS-C and FF sensors. This is a system effect, lens + sensor, and it's subtle, but I don't think 1" is long-term sweet spot in system price or performance.

Link | Posted on Feb 28, 2016 at 16:21 UTC as 119th comment
On article Worth the wait? A look inside the Pentax K-1 (649 comments in total)
In reply to:

ProfHankD: Looks like a very solid DSLR for $1800. I wish Sony would do a few of the things Pentax does... like the pixel shift/AA sim and, more importantly, the trap focus support (which I'm assuming this has since Pentax has had that for a long time). Just have to see how excited people get about a FF DSLR in a mount that doesn't have tons of AF glass... because a big, bright, OVF basically means it would be really hard to use with all that excellent old Pentax manual focus glass.... I hope they at least can do peaking on the live view....

Bright OVFs use focus screens that do not create a fully-formed focus plane in the finder -- that's why old manual-focus SLR OVFs are darker: it costs light loss (by scattering from the focus plane) to form a focusable image. Beyond that, no focus aids in there. Replacing the screen might work (I've done that on a DSLR before), but tends to mess-up the metering, which usually works by reading inside the OVF path. (The PDAF sensor is usually fed from a different path through the main mirror and hence is not affected.)

Trap focus can be used, where shutter firing is delayed until the PDAF detects focus for a manual lens, but that takes some getting used to because you'd be turning the lens focus to trigger the exposure. Trap is really intended for pre-focus, such as bride & groom walking down the aisle into the spot you pre-focused on.

Link | Posted on Feb 18, 2016 at 15:19 UTC
On article Worth the wait? A look inside the Pentax K-1 (649 comments in total)

Looks like a very solid DSLR for $1800. I wish Sony would do a few of the things Pentax does... like the pixel shift/AA sim and, more importantly, the trap focus support (which I'm assuming this has since Pentax has had that for a long time). Just have to see how excited people get about a FF DSLR in a mount that doesn't have tons of AF glass... because a big, bright, OVF basically means it would be really hard to use with all that excellent old Pentax manual focus glass.... I hope they at least can do peaking on the live view....

Link | Posted on Feb 18, 2016 at 14:41 UTC as 123rd comment | 4 replies
On article Picasa will be phased out in favor of Google Photos (155 comments in total)

I use Linux and have 20TB of photos on my disk array. How does Google Photo help with that?

I also want to note that I've been a foster dad, and have many photos of our foster kids -- which by law cannot be online. As foster parents, we're required to prepare a "lifebook" as personal history for each kid, but there are potentially serious, and fairly obvious, issues with the wrong people getting access to such photos. Similar constraints now generally apply for photos taken of your own kids in school plays, etc. -- because there may be other kids in the same activities who are foster kids, etc.

In sum, running Picasa locally has been great for picking-out the photos of our foster kids to make lifebooks, but I have not seen any way that Google Photo legally can be used for any such thing. I would encourage Google to think about doing what many companies have done with abandoned software: make the last version an open source release so that others could continue to support it.

Link | Posted on Feb 16, 2016 at 14:32 UTC as 25th comment
In reply to:

ProfHankD: Well, I guess now I can stop trying to build this sort of adapter.... :-(

Congrats to Techart! Leica M mount has a 27.80mm flange distance, so nearly all old SLR (and even many rangefinder) lenses should be able to be adapted to this and anything that uses unit focusing should deliver native IQ. Lenses with floating elements or internal focus generally will not produce as good IQ, but I assume that, like longer telephotos, they'll really just use the AF to tune the manual focus within the 4.5mm flange motion that the AF can drive... so even they will probably be ok in many cases because the focus range isn't that large.

They also say it can drive lenses up to 700g -- and that means it should be able to drive ANY lens when held by the lens, because the A7SII is the heaviest of the Sony E bodies and it only weighs 627g, so it would really just be moving the body back and forth.

I guess now I need to start designing 3D-printable adapters for everything to M mount.... ;-)

There is also the touchless shutter app using the EVF sensor -- just wave your hand in front of the EVF to fire the shutter.

Link | Posted on Feb 16, 2016 at 06:09 UTC
Total: 941, showing: 61 – 80
« First‹ Previous23456Next ›Last »