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: 829, showing: 141 – 160
« First‹ Previous678910Next ›Last »
In reply to:

J A C S: Gear used in this story: Sony AS and Sony 24-70.

Amazing shot!

Overlapping post timing... I started my post before you posted yours. So, I'm not sure which of us agrees with the other, but we do agree. :-)

Direct link | Posted on Sep 4, 2015 at 13:38 UTC
In reply to:

J A C S: Gear used in this story: Sony AS and Sony 24-70.

Amazing shot!

Excuse me, but the A7S is what he won -- not what he used to take the shot! Sony doesn't even make DSLRs, let alone weatherproof ones, so it's not at all clear he shot it with a Sony. In fact, although he's certainly a Sony shooter now, the only lens he describes on his WWW site that matches the description here is a Nikon 70-200mm f/2.8G VR, and his site says he used to use a pile of Nikons.

In sum, congrats to him -- and to Sony for being open-minded enough to give the nod to a photo taken with a different brand camera (it's probably got a Sony sensor in it anyway ;-) ).

Direct link | Posted on Sep 4, 2015 at 13:26 UTC
In reply to:

ProfHankD: As I've announced here before, I've been working on a free software tool that does credible repair of Sony arw2 files using computational texture synthesis. I posted a little example at http://www.dpreview.com/forums/post/56389545 and I've actually been able to do even better than that; I hope to post a test version this weekend.

In studying images and going through dozens of repair algorithms, I've gotten a much better feel for what's going on (and hope to have a research paper on it at IS&T Electronic Imaging 2016). Sony used to support a raw format that didn't use this compression, and then phased it out after several cameras supporting both... but it does seem that the issue is worse with newer bodies and certain postprocessing (e.g., AMAZE Bayer interpolation, which has become quite common). As a practical issue, I think the computational repair is good enough to make artifacting issues extremely rare, but Sony probably should go back to supporting both raw formats....

Esstee: Not looking for funding on this; it's a relatively small project that I view largely as a community service.

That said, I do expect to get a good research publication or two out of it and I would hope that this would add to my track record and help convince companies like Sony and Canon that it might be worthwhile working with me and supporting research on some of the grander things I'm trying to do. I have a strong track record in parallel supercomputing and have worked with many companies in that domain, but I've only been publishing in computational photography for about four years.

Direct link | Posted on Sep 3, 2015 at 12:13 UTC
In reply to:

ProfHankD: As I've announced here before, I've been working on a free software tool that does credible repair of Sony arw2 files using computational texture synthesis. I posted a little example at http://www.dpreview.com/forums/post/56389545 and I've actually been able to do even better than that; I hope to post a test version this weekend.

In studying images and going through dozens of repair algorithms, I've gotten a much better feel for what's going on (and hope to have a research paper on it at IS&T Electronic Imaging 2016). Sony used to support a raw format that didn't use this compression, and then phased it out after several cameras supporting both... but it does seem that the issue is worse with newer bodies and certain postprocessing (e.g., AMAZE Bayer interpolation, which has become quite common). As a practical issue, I think the computational repair is good enough to make artifacting issues extremely rare, but Sony probably should go back to supporting both raw formats....

Texture synthesis does have the potential to slightly increase the effective tonal resolution (reduce posterization).

Rather than each pixel having a value, my current best algorithm models each pixel as having an unknown value in a range defined by the combination of the bit precision, gap between mapped tone curve values, and a noise model. This is done for all pixels, not just the ones represented by fewer bits, because the tone curve affects all pixels. The processing synthesizes a 16-bit value for each pixel by sampling texture around 1089 same-color pixels. Thus, if the noise model is accurate, one can expect roughly the same tonal benefit as stacking images -- tonal resolution theoretically could be improved by as much as log2(1089) bits per pixel! Of course, the noise model isn't that good, so the real improvement is probably more like 1-2 bits at most, but the synthesized raw does not "skip" tonal values as the regular arw2 processing of the tone curve does.

Direct link | Posted on Sep 3, 2015 at 10:17 UTC

As I've announced here before, I've been working on a free software tool that does credible repair of Sony arw2 files using computational texture synthesis. I posted a little example at http://www.dpreview.com/forums/post/56389545 and I've actually been able to do even better than that; I hope to post a test version this weekend.

In studying images and going through dozens of repair algorithms, I've gotten a much better feel for what's going on (and hope to have a research paper on it at IS&T Electronic Imaging 2016). Sony used to support a raw format that didn't use this compression, and then phased it out after several cameras supporting both... but it does seem that the issue is worse with newer bodies and certain postprocessing (e.g., AMAZE Bayer interpolation, which has become quite common). As a practical issue, I think the computational repair is good enough to make artifacting issues extremely rare, but Sony probably should go back to supporting both raw formats....

Direct link | Posted on Sep 3, 2015 at 03:45 UTC as 181st comment | 4 replies
On article Holga Digital camera project launched on Kickstarter (150 comments in total)
In reply to:

nerd2: LCD-less camera 8MP 1/3.2" CMOS sensor? My smartphone has 16MP 1/2.6" BSI CMOS sensor WITH WQHD OLED display. Why on the earth do we need this kind of monstrosity?

Sometimes less is less, and the only thing you have more of is marketing hype.

Direct link | Posted on Aug 31, 2015 at 21:06 UTC

"What's hard to deny, though, is that no camera maker has gone back to a blank piece of paper to work out how a digital camera could work, rather than how to make a digital camera that works like their film era cameras did."

Wrong! Companies like Casio (e.g., QV10), Minolta (e.g., DImage V), and Kodak (e.g., DC260 Digita) did that in the early days of digital cameras, which explains why these companies are the market leaders today. Oh wait... they're not. ;-)

I think Sony has been doing the most exploration of the design space lately, but it actually has worked for them. Probably because they have a tradition of exploring design spaces for consumer electronics, which cameras now are. Now if only we could get Sony to allow other people's code to run in their cameras... the Sony culture of controlling everything about the device doesn't mesh well with the internet-connected intelligent behavior we expect from devices.

I explore using CHDK to reprogram Canon PowerShots. ;-)

Direct link | Posted on Aug 28, 2015 at 18:58 UTC as 158th comment
On article Sony Alpha 7R II: Real-world ISO invariance study (368 comments in total)
In reply to:

ProfHankD: As a service to the community and a research project at the University of Kentucky, I've been working on a free program to credibly repair the artifacting in ARW2 files. I asked for aligned training images differing only in exposure at http://www.dpreview.com/forums/post/56345899 , and have gotten some from various cameras. However, I also can use the above type of ISO sequence for training. I'd like DPReview's permission to use the above images for this purpose and, with appropriate citation, in any scholarly publications on the topic.

PS: The repair algorithm isn't quite good enough to be posting yet... hopefully within the next week or so I'll post a link here and/or in the FF Sony E forum.

The tool still isn't ready for posting, but here are two crops from the above ISO 100 raw data that show the repair quality is fairly good:
http://www.dpreview.com/forums/post/56389545

Direct link | Posted on Aug 28, 2015 at 17:03 UTC

The day ML (Magic Lantern) supports this is the day it becomes interesting.

Right now, it looks like a weak competitor to a cheaper two-year-old Sony. Adding ML and dropping price (as happened with the EOS M) could make me buy a small fleet of these for research use (which didn't happen for the original EOS M -- I stopped at one). At least the 11-22mm f/4-5.6 pricing is reasonable... although 11mm on a 1.6X crop isn't really competitive with Sony's 10-18mm f/4 on 1.5X crop.

Direct link | Posted on Aug 27, 2015 at 09:43 UTC as 138th comment | 2 replies
On article Sony Alpha 7R II: Real-world ISO invariance study (368 comments in total)

As a service to the community and a research project at the University of Kentucky, I've been working on a free program to credibly repair the artifacting in ARW2 files. I asked for aligned training images differing only in exposure at http://www.dpreview.com/forums/post/56345899 , and have gotten some from various cameras. However, I also can use the above type of ISO sequence for training. I'd like DPReview's permission to use the above images for this purpose and, with appropriate citation, in any scholarly publications on the topic.

PS: The repair algorithm isn't quite good enough to be posting yet... hopefully within the next week or so I'll post a link here and/or in the FF Sony E forum.

Direct link | Posted on Aug 25, 2015 at 16:44 UTC as 26th comment | 2 replies
On article MIT proposes new approach to HDR with 'Modulo' camera (115 comments in total)
In reply to:

forpetessake: "MIT proposes new approach" -- LOL, as many people already mentioned here, this idea has been discussed for years, there is nothing new about it.
The other ideas that were discussed are: global shutter -- multiple shorter exposures instead of a single longer one. Then, multiple switching load capacitors -- I think some company already implemented it. Then, pixels with different sensitivity (at the expense of noise). And probably more.
What the DPR should have mentioned though, is that all those approaches extend dynamic range by proportionally lowering the sensor base ISO, they can't improve the dynamic range at the same ISO -- the only way to do the latter (at the same CMOS) technology is to increase the sensor size.
There is a limit how low ISO can go before it becomes impractical. If say, ISO 25 is the lowest people can be interested in, compared to base ISO 100 now, then it means any of those methods can increase dynamic range by maximum 2 stops.

Most academic institutions only issue press releases on research when there has been some externally recognized accomplishment, but MIT regularly puts out press releases on work that is no more significant than the average technical conference paper. Fair enough; they are a private institution that profits greatly by this type of branding. However, DPReview shouldn't be so lazy and/or desperate for content that they blindly echo these releases as though they were important advances.

If you want to give a glimpse of what's next, look at things like papers that won awards in technical publication venues, etc.

Direct link | Posted on Aug 21, 2015 at 00:06 UTC
On article iFixit tears down Sony's new a7R II to find its secrets (290 comments in total)
In reply to:

marshwader: Looks a tad fragile inside the "alpha dog". Not much of a chassis. Better not drop it.

Don't drop ay such device. ALL tech with micron-level alignments is fragile. However, this is classic Sony: every part engineered to fit together as one coherent system rather than random pieces glued/screwed/cabled inside a camera-shaped box. The button structure is particularly impressive -- all the buttons implemented as external covers over board-mounted switches without cables should make it much more robust. Just beautiful.

Direct link | Posted on Aug 19, 2015 at 21:26 UTC
In reply to:

Leandros S: I don't understand how it's a 2015-2016 award when only the first half of 2015 could be accounted for.

From the EISA WWW site: 50 experts voting, decision made end of June, products in some reviewer hands start of year, widely for sale by Oct. 1. That explains why it's A7II and not A7RII for Sony, but I'm shocked the Canon 5DS qualified on the in-reviewer-hands timing. I wonder if qualifying for this award has anything to do with the big gap between announcement and availability for the 5DS?

Direct link | Posted on Aug 18, 2015 at 01:26 UTC
On challenge Stacking Exposures (4 comments in total)
In reply to:

luirod: Dose it include Insects

As subjects? Sure.

Direct link | Posted on Aug 15, 2015 at 03:23 UTC
On challenge Stacking Exposures (4 comments in total)
In reply to:

Xezi: Congrats on this challenge, it got some of the the most interesting pics of the year imo.

I'm pleased too. Not even anything to disqualify! I guess folks willing to deal with stacking are actually also smart enough to read and follow challenge rules.... ;-)

(Oh oh. Did I just curse things by saying that long before the challenge is over?)

Direct link | Posted on Aug 15, 2015 at 03:22 UTC
In reply to:

everybodyisone: "interchangeable mount" Dreamland. ... I wish more manufacture would follow that...

For manual still lenses, M42 would have been THE standard if only Nikon hadn't made the F flange distance too long. For cinema lenses, blame Arri. Their PL mount has a huge 52mm flange distance, which is even longer than old Adaptalls.

It looks like Samyang didn't really redesign the optics for this, nor accommodate different aperture ring directions (e.g., EF vs. F), but simply devised a universal mount that could fit INSIDE PL enough to work. Cool. We'll have to see if Samyang's universal mount ends up seeing broader use or not. It would be nice if it did, but I shoot with 18mm flange distance Sony E-mount bodies now, so almost all lenses work with cheap adapters anyway. Actually, ALL of the non-E mounts they support -- PL, EF, F, and MFT -- are adaptable to E-mount (and to most mirrorless mounts).

Direct link | Posted on Aug 12, 2015 at 15:10 UTC
On article Rough and ready: Olympus Tough TG-4 review (275 comments in total)

For rugged use, I have an Olympus Stylus 1030 SW and an Olympus Stylus Tough TG-860 (as well as a Panasonic Lumix DMC-TS2). The Olympus cameras are really robust and excellent in features; for example, the TG-860 has a flip LCD and 21mm equivalent wide. However, IQ has always been weak compared to regular compacts (I use a lot of Canon PowerShots with CHDK giving me raw access).

Raw gives a bit of a second chance to rescue images, and I think that's great, but IS THE RAW HERE REALLY RAW? Most compacts have lenses with a wider view angle than advertised and bad distortion, but I don't see that in the supposedly uncorrected raw images here. In the studio comparison, you don't seem to have corrected CA for the raws; did you correct cropping and distortion or does the camera deliver a "partially cooked" raw?

Direct link | Posted on Aug 10, 2015 at 12:57 UTC as 84th comment | 1 reply
In reply to:

Marcelobtp: The thing that most standout here was the lenses!
This lenses have very nice qualities!
Not so surprised about resolution nothing that i care too much seeing on a monitor(4k).
Colors i cant really judge but looks like we are seeing better skin tones out of the camera.
Good to see.

Rishi: well, I can see why you like it... although I'm not a fan of starbursts. I suppose such short focal lengths can be extra-picky about adapter tolerances, and a slightly short adapter might have thrown some floating-element compensations? The one on the Metabones adapter looks much more competitive in IQ with the Sigma 8-16mm on APS-C, although transverse CA looks pretty heavy.

I haven't tried fixing the CA on your shot, but I suspect after that, this Canon lens would give a little more usable resolution on FF than I get from the Sigma on 24MP APS-C. If only the price were not 4X the Sigma.... ;-)

Direct link | Posted on Aug 9, 2015 at 00:59 UTC
In reply to:

Marcelobtp: The thing that most standout here was the lenses!
This lenses have very nice qualities!
Not so surprised about resolution nothing that i care too much seeing on a monitor(4k).
Colors i cant really judge but looks like we are seeing better skin tones out of the camera.
Good to see.

Vanitas Photo: read that; not using a mac, downloading originals and viewing under Linux (yeah, I'm a geek). BTW, the "to taste" raw processing used generally looks a little coarsely shaded and oversharpened to me.

No matter; that Canon 11-24mm just isn't competing well on FF with my Sigma 8-16mm on APS-C, which makes me sad because I really want a FF ultrawide zoom. I wish somebody would make a cheap little 12-16mm f/8 that delivers great IQ on FF....

Direct link | Posted on Aug 8, 2015 at 13:04 UTC
In reply to:

Marcelobtp: The thing that most standout here was the lenses!
This lenses have very nice qualities!
Not so surprised about resolution nothing that i care too much seeing on a monitor(4k).
Colors i cant really judge but looks like we are seeing better skin tones out of the camera.
Good to see.

I think the Sony/Zeiss lenses did very well, with IQ of the Batis 85mm f/1.8 in the second shot just about perfect. Unfortunately, that awesomely wide Canon EF 11-24mm F4L USM seems to have IQ fall off pretty quickly toward the corners.

I really would love to have a <=12mm FF ultrawide zoom for my A7II (and future A7RII), but so far I haven't seen a clear IQ win over my Sigma 8-16mm at 8mm on my NEX-7 (8mm APS-C is 12mm FF equiv). Worse still, the price is a little higher on the SIgma 12-24mm and is MUCH higher on this Canon....

Direct link | Posted on Aug 8, 2015 at 00:11 UTC
Total: 829, showing: 141 – 160
« First‹ Previous678910Next ›Last »