duartix

duartix

Lives in Portugal Lisboa, Portugal
Works as a IT Engineer
Joined on Mar 6, 2007

Comments

Total: 248, showing: 1 – 20
« First‹ Previous12345Next ›Last »
In reply to:

duartix: Unless that black circle near the grip is another camera/sensor pair, I fail to even imagine how can they generate a focus overlap picture from a single combination of sensor/lens.

Even very advanced software solutions by Google (the google camera app on Android is such an example) need a minimum amount of parallax to reconstruct/generate a depth map and thus create a that focus overlap, but doing this temporally (à la Google) isn't practical on a rangefinder where you need immediate feedback.

Plus, if it is indeed a secondary camera/sensor pair, you will have to account for parallax errors on the focus overlap reconstruction. And what is much worse, you will have to be very careful on where you place your fingers when you shoot.

P.S. Before anyone else does it, I'm always ready to make a fool of myself. :)
http://konost.com/?page_id=7182
Yes there is. Now I'm just questioning the placement.

@Barry Pearson
I didn't understood your explanation to it's full extent, but I don't think that's how it works. This http://konost.com/?page_id=7182 shows a complete sub-area of the main sensor being superimposed. It's how they compute the offset that is the mistery. Maybe it's just a concept and they haven't worket it out yet.
But one thing seems sure: the two superimpositions look clearly parallaxed, thus probably beeing sourced from different points. Since the one on the display shows final DOF that can't be it. Probably one comes from the optical viewfinder and the other from the secondary sensor between the main lens and grip.

Direct link | Posted on Feb 26, 2015 at 17:46 UTC

Unless that black circle near the grip is another camera/sensor pair, I fail to even imagine how can they generate a focus overlap picture from a single combination of sensor/lens.

Even very advanced software solutions by Google (the google camera app on Android is such an example) need a minimum amount of parallax to reconstruct/generate a depth map and thus create a that focus overlap, but doing this temporally (à la Google) isn't practical on a rangefinder where you need immediate feedback.

Plus, if it is indeed a secondary camera/sensor pair, you will have to account for parallax errors on the focus overlap reconstruction. And what is much worse, you will have to be very careful on where you place your fingers when you shoot.

P.S. Before anyone else does it, I'm always ready to make a fool of myself. :)
http://konost.com/?page_id=7182
Yes there is. Now I'm just questioning the placement.

Direct link | Posted on Feb 25, 2015 at 15:16 UTC as 45th comment | 5 replies
On Hands-on with Canon's 'not-coming-to-USA' EOS M3 article (517 comments in total)
In reply to:

Atsyn: I dont see the point of this EOS M compared to µ43, fuji or sony mirroless...

@brudy
In your dreams perhaps. I suggest you check with reality here:
http://www.dpreview.com/reviews/olympus-om-d-e-m10/13
On a blind test you will never tell the difference.

Direct link | Posted on Feb 6, 2015 at 18:11 UTC
On Olympus OM-D E-M5 II First Impressions Review preview (1321 comments in total)
In reply to:

olyham: The high res mode may do interesting things with a waterfall...be good to see the result

Perhaps you should upgrade your studio poster to include a waterfall. ;)
Looks popular.

Direct link | Posted on Feb 5, 2015 at 11:56 UTC
On Olympus OM-D E-M5 II First Impressions Review preview (1321 comments in total)
In reply to:

tkbslc: Seems like a waste to put effort into making the video mode nicer and then leaving off the headline 2015 video spec - 4K.

It's biggest problem with video is not about not featuring 4K.
It's because it is still heavily line skipping on the FullHD readout. Yet, 2/3 (bitrates/frame rates/readout) is not completely bad at all.

Direct link | Posted on Feb 5, 2015 at 11:53 UTC
On Olympus OM-D E-M5 II First Impressions Review preview (1321 comments in total)
In reply to:

UnderDriven: Olympus may want to consider implementing a 4-image mode which does not increase resolution, but does provide full color sampling at every pixel...

It's there already @UnderDriven.

It's called the 40 MP multi-exposure mode. It just needs you to dowscale the result in the end and it takes a little longer than what you wanted. ;)

You are welcome.

Direct link | Posted on Feb 5, 2015 at 11:50 UTC
On 3,200MP LSST camera gets construction approval article (257 comments in total)
In reply to:

duartix: So, where is the Equivalence Police right now?

Not for an instant. But perhaps we should. Sellers gotta sell.

Direct link | Posted on Feb 4, 2015 at 17:37 UTC
On 3,200MP LSST camera gets construction approval article (257 comments in total)

So, where is the Equivalence Police right now?

Direct link | Posted on Feb 3, 2015 at 18:05 UTC as 38th comment | 2 replies
On Canon EOS-1D C sees $4000 price drop in North America article (249 comments in total)
In reply to:

photogeek: I'd buy one, but the lens is not quite large enough. Can they make it twice as large (and heavy) as the body? I'll pay extra.

Yes, I'll have one too if they add an extra battery pack.

Direct link | Posted on Feb 2, 2015 at 10:15 UTC
On 2014 Readers' Polls: The results are in! article (375 comments in total)
In reply to:

newworld666: No real game Changer in the list like the 7DII ? or even QX1 ?

No game changer?
What do you call the LX100 then?

Direct link | Posted on Jan 23, 2015 at 14:39 UTC
On Opinion: The myth of the upgrade path article (1410 comments in total)
In reply to:

razadaz: In recent years photography seems to have become more like Hi-Fi, where many people seem to have become more interested in the fidelity of the system they have than the music they listen to.

Because you can't quantify photography that's where the majority turns to. To something they can measure...

Direct link | Posted on Jan 13, 2015 at 10:25 UTC
On Canon announces five PowerShot compacts article (150 comments in total)

Is Zoom the new Megapixel?
Even if so, there is still a lot to be explained here...

Direct link | Posted on Jan 12, 2015 at 16:30 UTC as 3rd comment

Oh, the convergence!

Direct link | Posted on Jan 6, 2015 at 17:54 UTC as 11th comment
On BPG image format aims to replace JPEGs article (208 comments in total)
In reply to:

TangoMan: Oh no! More skin smoothing!
Examination of all the available images on the image comparison tool show that it works by smoothing away texture. Hair on arms disapear, fruits appear to be made of plastic, and skin... well, it's airbrushed to anormalcy. And white or black dots on green just disappear.
If it catches up, I'm afraid texture will become a thing of the past on most images on the web.

Yes, but...
http://www.dpreview.com/articles/0350654437/bpg-image-format-aims-to-replace-jpegs?comment=2722892690

Direct link | Posted on Dec 16, 2014 at 18:39 UTC
On BPG image format aims to replace JPEGs article (208 comments in total)
In reply to:

Shiranai: JPG2000 wasn't able to replace JPG.
WebP wasn't able to replace JPG, altough its heavily forced by Google.
Have your try, BPG...

Looking at the comparison, I notice JPG2000 is the only format that doesnt completely blur the points on the green band in medium&large quality. If it didn't have the block artifacts it would almost look like the original. BPG looses lots of small details like these points and the texture on his legs. I guess it will behave badly with noisy photos.

Yes, those points are blurred beyond existence in BPG.

But JPEG2000 doesn't do them justice either and makes them look like mosaic artifacts, which leads me to suspect this is highly related to color subsampling.
Anyway, I can live better with those than with severe edge ringing, blocking and aliasing that is present in all but BPG.

It sure looks promising but so did JPEG2000 10 years ago... :(

Direct link | Posted on Dec 16, 2014 at 16:06 UTC
In reply to:

Summi Luchs: My guess is that they will use the sensor-shift to de-Bayer the image. They can move the red, green and blue pixels between the exposures so, that they get a full RGB sample for every pixel (like a Foveon sensor). This would not give us 'true' 40MP resolution, it is simply the same calculus of 'equivalence' Sigma uses for its cameras. (Sigma uses a factor of three, Oly gets a lower factor of equivalence, as the Bayer sensor is not RGB but RGGB).
The gain would be better color information, that what makes Foveon images special.

I don't think they do the same as Hasselblad and other earlier attempts to increase true resolution by sensor movements. Earlier sensors had more blind gaps between the photosites making such techniques meaningful. Newer sensors have a dense array of microlenses (to gather more light per photosite), so there is not much left to sample in between.

A bit of sampling theory...

If they can manage subpixel shifts as it's all pointing to, then it does provide extra spatial resolution, because in reality what you are doing is over sampling.
This means that in theory, the Nyquist frequency limit could be doubled. :)

This is much more significant to final IQ than fully sampling every color channel. Humans have much more perception of spatial resolution than color resolution.

A proof of this is JPEG, in which color subsampling plays an almost as important role as compression.

Direct link | Posted on Dec 5, 2014 at 11:46 UTC
On High-end pocketable compacts 2013 roundup article (264 comments in total)
In reply to:

duartix: What?
Is this really the High-end pocketable compacts meeting?
Why can't I see the Panasonic LX100, then?

Thank you @jdob123.
Seems like I had a different definition of pocket.

Direct link | Posted on Nov 27, 2014 at 15:20 UTC
On High-end pocketable compacts 2013 roundup article (264 comments in total)

What?
Is this really the High-end pocketable compacts meeting?
Why can't I see the Panasonic LX100, then?

Direct link | Posted on Nov 27, 2014 at 14:57 UTC as 39th comment | 5 replies
On Canon announces EOS C100 Mark II article (279 comments in total)
In reply to:

duartix: Ctrl + "F": 4K - 0 of 0
Ctrl + "F": 4:2:2 - 0 of 0
Ctrl + "F": RAW - 0 of 0
Ctrl + "F": neutral - 0 of 0
Ctrl + "F": zebra - 0 of 0
Ctrl + "F": peak - 0 of 0
Ctrl + "F": xlr - 0 of 0
Ctrl + "F": prores - 0 of 0
Ctrl + "F": XAVC - 0 of 0
Ctrl + "F": 10 bit - 0 of 0
Ctrl + "F": 120fps - 0 of 0
Ctrl + "F": price - $5,499

WTF? Is this supposed to be a professional camera??? Because only the price says so!

Thanks for the corrections. I was about to write that either the camera or the press release were nothing near "professional", so it looks like it's a bit of both.
"Neutral" was me trying to check for the existence of an integrated ND filter.

Direct link | Posted on Oct 24, 2014 at 09:50 UTC
On Canon announces EOS C100 Mark II article (279 comments in total)

Ctrl + "F": 4K - 0 of 0
Ctrl + "F": 4:2:2 - 0 of 0
Ctrl + "F": RAW - 0 of 0
Ctrl + "F": neutral - 0 of 0
Ctrl + "F": zebra - 0 of 0
Ctrl + "F": peak - 0 of 0
Ctrl + "F": xlr - 0 of 0
Ctrl + "F": prores - 0 of 0
Ctrl + "F": XAVC - 0 of 0
Ctrl + "F": 10 bit - 0 of 0
Ctrl + "F": 120fps - 0 of 0
Ctrl + "F": price - $5,499

WTF? Is this supposed to be a professional camera??? Because only the price says so!

Direct link | Posted on Oct 23, 2014 at 15:31 UTC as 37th comment | 5 replies
Total: 248, showing: 1 – 20
« First‹ Previous12345Next ›Last »