DSPographer

Lives in United States United States
Works as a digital signal processing engineer
Joined on Jan 10, 2005
About me:

Canon 5D mark II camera.

Canon 28-135/3.5-5.6_IS_USM, 24/2.8, 50/1.8II, 100/2.8_USM_macro, 200/2.8L lenses.

Sigma EF-500_DG_Super flash.

Comments

Total: 53, showing: 1 – 20
« First‹ Previous123Next ›Last »
In reply to:

jnd: Man, they really should have invested into PDAF.

The be fair, this is more like phase detect manual focus, since the corrections are calculated and sent up from ground control.

Link | Posted on Mar 18, 2022 at 16:17 UTC
In reply to:

Greg Ohio: When I was a kid, the largest optical telescope in the world was the 5.1 meter Hale Telescope. This is 6.5 meters. It will still be in a tie for the 14th largest telescope ever made. And, it's a million miles out in space.

Yes, Webb has about the same resolution in the infrared band as the Hubble space telescope has in the visible band. The future "Extremely Large Telescope" should have much greater resolving power, but only over the limited field of view over which its adaptive optics can correct atmospheric distortions. Of course ground based telescopes suffer not just from the blurring effects of the atmosphere, but also from the absorption and emission of light [by the atmosphere], especially outside of visible wavelengths.

Link | Posted on Mar 18, 2022 at 16:01 UTC

Nice article, but could measurements use a unit a bit more standardized than the size of a human hair?
I looked it up to convert, and as I expected the size varies: from about 17 to 181 microns across.
https://hypertextbook.com/facts/1999/BrianLey.shtml

Link | Posted on Sep 9, 2020 at 21:49 UTC as 32nd comment | 4 replies
On article Canon EOS R5 vs. R6: What's the difference? (257 comments in total)
In reply to:

Flashback: Both have anti-aliasing filters, that's odd. Thought we'd since the last of these. Maybe to mask some of those dual-pixel anomalies?

The R6 seems cracking value with it's full feature set, but that 'low res' 3.68Mp viewfinder is a real bummer. I'm currently shooting the 5.76Mp S1R and it's a joy to use.

Excellent article BTW

Canon's dual pixels do not generate any anomalies.
Anti-aliasing filters are necessary when shooting video to avoid ugly moving moire patterns.

Link | Posted on Jul 14, 2020 at 22:35 UTC

Does the camera have autofocus or is it fixed focus?
If autofocus, does it have tap-to-focus in the app?

Link | Posted on Nov 11, 2019 at 00:49 UTC as 31st comment | 1 reply
On article Hands-on with the DJI Mavic Mini (162 comments in total)
In reply to:

Merida: Does DJI offer a map of its flight restrictions?
I want to see where I can fly this drone before I buy it.

After seeing Lee Jay's comments I checked out my area of Connecticut. My property is just within 8 miles of two uncontrolled airfields with grass runways but I have never seen a low-level GA aircraft in my area in over thirty years. I think the problem is mostly with the FAA not DJI: the controlled airfields have an automated process to get an LAANC, but these unused fields don't. So, the only legal way to fly a drone here is with a paper authorization, even though the risk of collision from an aircraft from these fields is essentially zero. Looking at a map, this seems to be a very common situation.

Link | Posted on Nov 4, 2019 at 18:26 UTC
On article Nikon announces Z-mount lens roadmap (284 comments in total)
In reply to:

tkbslc: Looks like Z mount APS-C is going to make even EOS M look good.

But EF-M and RF are incompatible: at least Nikon's mirror-less aps-c camera can mount its full frame mirror-less lenses.

Link | Posted on Oct 10, 2019 at 17:38 UTC

In this case I expect that the "Tetracell" technology allows the marketing department to crow about 43 megapixels in a camera best used in 11 megapixel mode.

Link | Posted on Sep 24, 2019 at 16:32 UTC as 18th comment | 1 reply
On article [UPDATE] Fujifilm GFX 100 added to studio test scene (301 comments in total)
In reply to:

DSPographer: These high res cameras seem to break the comparison mode of the studio scene image comparison tool. When I try to compare the Fuji GFX 100, the P1 XF 100MP, and the Lumix DC-S1R in high res mode it selects a resolution lower than any of them when the comp tab is selected.

Basing the comparison resolution on the smallest image, but allowing that image to be either up-scaled or down-scaled to the nearest megapixel value in the above sequence, would make it less critical to have small steps between resolutions. So, if you had a 46MP camera, it would allow comparison with higher res cameras at 51MP given the second sequence above.

Link | Posted on Jun 4, 2019 at 22:11 UTC
On article [UPDATE] Fujifilm GFX 100 added to studio test scene (301 comments in total)
In reply to:

DSPographer: These high res cameras seem to break the comparison mode of the studio scene image comparison tool. When I try to compare the Fuji GFX 100, the P1 XF 100MP, and the Lumix DC-S1R in high res mode it selects a resolution lower than any of them when the comp tab is selected.

With a ratio of 1.414 the sequence would be: 36MP, 51MP, 72MP...
This might be better if the cost of each additional resolution is large.

Link | Posted on Jun 4, 2019 at 22:01 UTC
On article [UPDATE] Fujifilm GFX 100 added to studio test scene (301 comments in total)
In reply to:

DSPographer: These high res cameras seem to break the comparison mode of the studio scene image comparison tool. When I try to compare the Fuji GFX 100, the P1 XF 100MP, and the Lumix DC-S1R in high res mode it selects a resolution lower than any of them when the comp tab is selected.

I would suggest a fixed ratio scaling for the tool such as 1.26
Then the sequence would be 36MP, 45MP, 57MP, 72MP ...

Link | Posted on Jun 4, 2019 at 21:55 UTC
On article [UPDATE] Fujifilm GFX 100 added to studio test scene (301 comments in total)

These high res cameras seem to break the comparison mode of the studio scene image comparison tool. When I try to compare the Fuji GFX 100, the P1 XF 100MP, and the Lumix DC-S1R in high res mode it selects a resolution lower than any of them when the comp tab is selected.

Link | Posted on Jun 4, 2019 at 20:36 UTC as 78th comment | 5 replies
In reply to:

(unknown member): Post a photo of the most beautiful place on earth with a false geotag leading to the edge of a bottomless pit of quicksand.
Don’t worry too much about it the oil rigs will soon spoil the view.

Using a phrase like "The Climate Cult" is a convenient way to signal that we should ignore everything you say. Thanks for identifying that your statements lack touch with reality.

Link | Posted on Dec 12, 2018 at 18:53 UTC
In reply to:

Jon555: So presumably the question is whether it consumes more or less than $1080 in electricity in two years ($3400-$2320)? I guess less.

BTW I doo wonder if people buying the machines and running them in their own houses are at risk of getting raided by the Drug Enforcement people, as usually the house in the street with the big heat signature if the one growing drugs...

USA electricity rates range from 9.7 to 29.3 cents per kW-h.
https://www.chooseenergy.com/electricity-rates-by-state/

Link | Posted on Jan 11, 2018 at 21:42 UTC

I'm not sure we will understand the significance of this announcement until an iced-tea company has a chance to evaluate the details of the block-chain algorithm.

Link | Posted on Jan 9, 2018 at 20:00 UTC as 75th comment
In reply to:

mosswings: Very cool, but two comments:

1. Numerical aperture of 0.2: this is roughly equivalent to an f-number of 2.5.
2. Efficiency of 20% - if I understand this right, only 20% of the incident light is transmitted, a light loss of 2.1 stops. Makes sense - this is effectively a thin metal disk with slots at various spacings and widths. It is far less transparent than a glass element - even many glass elements.

There are applications for this technology, but it does not appear to be in low-light capable imaging systems - that is, consumer cameras - at this time.

You can see the supplemental figures for the article without paying. Figure S8 shows the efficiency, 20% is the peak- which is for blue light, for red light the efficiency is around 5%. The efficiency is the net transmission as you guessed, the efficiency is lowered by the crossed polarizers that they needed to remove the background light to maintain decent contrast. The diffraction-only lens shown would therefore not be of much use for ordinary photography.
A different configuration was modeled with the diffractive component correcting the aberrations of an ordinary simple refractive lens. This hybrid approach does apply to photographic lenses, but it isn't a new idea: Nikon and Canon etc already have lenses with diffractive optical components.

Link | Posted on Jan 5, 2018 at 18:16 UTC
On article UPDATED: Sony a7R III is still a star eater (460 comments in total)
In reply to:

Henrik Herranen: Umm, if the graphs are correct, the attenuation at fs/2 (Nyquist) is about 1.5 dB when compared to DC. This should hardly be visible in images. Then again, the image/video community uses the decibel scale incorrectly in so many imaginative ways (starting from the very incorrect definition of PSNR), so for all I know these dB numbers could just as well be potatoes.

(Short rant: if an audio signal's amplitude is halved, the power of the signal is lowered by a factor of 4, or approximately by 6 dB. But if an image RAW sample value is halved, power of the signal is lowered only by a factor of 2, or approximately by 3 dB. But, everywhere in the video industry people are using the dB scale incorrectly, using the incorrect assumption that a ratio of 1:2 is -6 dB, and not -3 dB as it should be. End of rant.)

The processing that Jim is performing shows the existence of filtering, but it doesn't characterize how local that filtering is. Instead of showing week filtering of all red channel pixels, the low measured effect probably just reflects that only a fraction of the pixels are affected. Further investigation is needed to show how strong the "eating" effect is and what the effect looks like.

Link | Posted on Nov 21, 2017 at 19:30 UTC
On article Sony a7R III sample gallery (252 comments in total)
In reply to:

Amnon G: Did you notice the weird texture in the left eye of the model in the 3rd image? Looks very processed.
Also look at the right eye in photo #6.
I have never seen humans with patterns like this.

So this is over processing or we found the way to identify replicants ;-P

Image 55 clearly shows the blue-tinted contact lens over this model's light brown iris. I don't know why colored contact lenses normally use such a crude dot pattern, but they do.

Link | Posted on Oct 26, 2017 at 19:43 UTC
On article Sony FE 100mm F2.8 STF gallery and first impressions (319 comments in total)
In reply to:

sh10453: I like this lens. I'm not a fan of bokeh circles. Nice subject isolation.
Thanks Rishi.

Since all the control over the amount of apodization occurs in the first stop, I think a valid criticism is that this lens should have allowed finer aperture increments than 1/3 stop. [Edit] Wait, can you actually do this by simply manually adjusting the aperture in "clickless" mode?

Link | Posted on May 11, 2017 at 13:51 UTC
On article Sony FE 100mm F2.8 STF gallery and first impressions (319 comments in total)
In reply to:

sh10453: I like this lens. I'm not a fan of bokeh circles. Nice subject isolation.
Thanks Rishi.

I agree that stopping down more than a stop is not what this lens is intended for. However, I see stopping down by 1/3 stop (as shown in the article yielding a less strong apodization effect) as a valid option: still different from a traditional lens but not as buttery-smooth as wide open. This seems more flexible to me than a lens which only has the equivalent effect of this lens stopped down by 1/3 stop. If Rishi instead always wants fully hard edged bokeh balls of a lens with no apodization at all, then he doesn't represent the photographers this less is targeted to: those that want the option in some situations to have smoother bokeh.

Link | Posted on May 11, 2017 at 13:37 UTC
Total: 53, showing: 1 – 20
« First‹ Previous123Next ›Last »