noirdesir

noirdesir

Lives in Switzerland Switzerland
Works as a Engineer
Joined on Nov 4, 2006

Comments

Total: 414, showing: 1 – 20
« First‹ Previous12345Next ›Last »
In reply to:

Wonder Kid: 4k is a 2.4x crop?? Forget any indoor or wide angle shots.
I see nothing worth the extra money

4K and APS-C have a noticeably different aspect ratio which makes crop ratios based on the diagonal somewhat misleading. What the camera does in 4K is simply to crop the top and bottom to get from 3:2 to 16:9 (6000 x 4000 to 6000 x 3376). Thus the horizontal AOV is maintained, only the vertical AOV is unavoidably reduced. The camera is reading a 6K (6000 x 3376) portion from the sensor, demosaics it and scales it down to 4K (3840 x 2160) before writing it to the card.

It's the best possible solution unless you oversize the sensor.

Direct link | Posted on Feb 3, 2016 at 18:43 UTC
In reply to:

G1Houston: The Olympus guy made a big deal about a key design feature in the PEN: not hitting the nose with the rear of the camera. Do you need to make a special body just to do that? Why don't they change EM1, EM5, EM10, all of them, by moving the EVF to the left? Problem solved???

I for one is tired of this retro design thing. Olympus thinks they can tap into a niche crowd to make a few quick bucks (thus the unrealistically high initial asking price). The PEN to Olympus is the Df to Nikon. Let's wait and see.

"Why don't they change EM1, EM5, EM10, all of them, by moving the EVF to the left?"

Isn't that what they essentially did?

Direct link | Posted on Jan 27, 2016 at 18:04 UTC
In reply to:

Xentinus: I can't find a reason to buy a such camera,while you can put a pancake lens in front of sony a6000.I know a little bit bigger but you can change your lens any time you want and you also have cool AF system and viewfinder.
Or you can go with Pana/Oly and cool lenses!

If thickness didn't matter, there wouldn't be any justification for the RX100 either. The GR is even thinner than the original RX100 and 6 mm thinner than the current RX100 (35 vs 41 mm).

Direct link | Posted on Jan 21, 2016 at 22:18 UTC
On article Hands-on with the Fujifilm X-Pro2 (72 comments in total)
In reply to:

NicoPPC: Fujifilm is doing good stuff but I can not image shooting a whole day with a poor digital viewfinder. I would prefer a small dslr.

Hey guy, can't you build a DSLR in the form of an old Olympus OM ?

In regard to DSLR vs OM size, the PD AF module in any DSLR adds height (below the lens mount) to the DSLR.

Direct link | Posted on Jan 16, 2016 at 03:14 UTC
On article Hands-on with the Nikon D5 (404 comments in total)
In reply to:

Joe Ogiba: Nikon D5 @ 102400 iso :
https://c2.staticflickr.com/2/1560/24253425102_ef4e73a1db_o.jpg

Sony A7s II @ 102400 iso :
https://c2.staticflickr.com/2/1646/23734817103_49c8e9f961_k.jpg

One is a video frame, the other a still frame. I would not be surprised if can get a cleaner image from a video frame if the video compression itself uses temporal compression on static area in the image and thus can reduce noise significantly.

Direct link | Posted on Jan 14, 2016 at 21:54 UTC
On article Hands-on with the Nikon D5 (404 comments in total)
In reply to:

steelhead3: Where are Allison's hands when you need them? I hope DP changes practice and gives these pro cameras a review. I really want to see the phase1 put through the paces.

I always cringe when I see anybody taking a photo while holding the camera with only one hand, in particular for cameras like the RX10.

I also have never seen anybody holding a camera with a vertical grip putting a hand on both grips. I've owned a D3 and I never shot it like this, the second hand always went to the lens.

Direct link | Posted on Jan 13, 2016 at 01:26 UTC
On article Hands-on with the Nikon D5 (404 comments in total)
In reply to:

Nick932: Finally something more expensive than a Leica and everybody accepts the price as normal £5199. For about £250 you get a lot more than than a Leica SL. you get R(SLR), Size, Weight.

Leica SL: $7500
Nikon D5: $6500

Direct link | Posted on Jan 13, 2016 at 01:17 UTC
In reply to:

callaesthetics: If i put a full frame 300 f4 lens with an adapter on the Olympus i get the same image and exposure correct? And it only uses a small image circle from the full frame lens, the Olympus is engineered so that it uses only a image circle made for m43 sensor, so why are the lens the same size? Shouldn't this lens be smaller?

Interesting comparison is the Olympus 150 mm f/2 + 2x TC combo. With the 4/3 to m43 adaptor it has a combined length of 210.5 mm and diameter of 100 mm. For comparison the 300 mm f/4 has a length 227 mm and a diameter of 92.5 mm.

While the TC combo is a bit shorter, it is expected to have a noticeably worse IQ as TC combos differ significantly from optimal designs.

Note that comparisons of size with (FF) DSLR 300 mm f/4 lenses suffer from the fact m43 has a flange distance that is about 25 mm shorter.

Direct link | Posted on Jan 6, 2016 at 16:41 UTC
In reply to:

thx1138: LOL at the price. Seriously delusional. The mirrorless fanbois would be in hysterics if Canikon released a 300 f/4 that cost $2500. Hell the Nikon with DO technology is only $2K and has to be corrected for a 4x larger sensor.

Of course you compare it against a 600 mm lens (if you want to compare it against a FF system). But you need to compare it against a 600 mm f/8 FF lens which doesn't exist (there was a Leica Telyt 560 mm f/6.8 which comes close but is a 45-year-old design). What does exist for FF is the 300 mm f/4 + 2x TC combo (and a 400 mm f/5.6 + 1.4x TC which yield 560 mm f/8).

Size and weight-wise, the 300 mm f/4 + 2x TC can compete against this 300 mm f/4 m43 lens, but IQ-wise it cannot. Even the very best lenses suffer noticeably from a 2x TC (http://www.photozone.de/nikon_ff/499-nikkorafs200f2vrff?start=1 and that lens can be stopped down by two stop with the TC before reaching its peak IQ, an f/8 lens on FF rarely and if then only barely improves from further stopping down).

Direct link | Posted on Jan 6, 2016 at 16:28 UTC
In reply to:

sunjester: Seems really huge, expensive, and slow for a pro lens equivalent 210mm f4. When used on a 1" sensor.

Sunjester's point is that if somebody is saying this lens is equivalent to a 600 mm f/4 lens on FF, then a 210 (220) mm f/4 on 1" is also equivalent to 600 mm f/4. And thus applying this 'logic', this lens is equivalent to 220 mm f/4 on 1".

Direct link | Posted on Jan 6, 2016 at 15:55 UTC
On article Nikon fills in the blanks on professional grade D5 DSLR (554 comments in total)
In reply to:

wetsleet: ISO (ASA) 3000000?
Is it time for a return of the DIN logarithmic scale instead?

(obviously, I understand the marketing dept prefer impressively big numbers)

Note that Canon announced a camera with 19 μm pixels (2.3 MP), the ME20F-SH, which had an ISO rating of up to 4'000'000 (that camera was supposed to ship just last month for $30'000). This led Wikipedia to extend their table for film speeds up to 4'000'000/67º (3'200'000 would be 66º).

Direct link | Posted on Jan 6, 2016 at 01:20 UTC
In reply to:

PatMann: Not quite 645 full frame, but close (about 93.4%). My Pentax 645 slides are about 41.5 x 56mm in image area. Certainly still a long way from 6 x 6 or 6 x 7 medium format film sizes.

And the 'trend' to call the largest imaging sensors available (outside of scanning backs and maybe military projects) 'medium' format isn't ridiculous? Names and labels very often outlive the reasons their coinage was based on, that doesn't make them necessarily ridiculous.

Direct link | Posted on Jan 3, 2016 at 22:01 UTC
In reply to:

Henry McA: Why is everything so noisy? Is it the "advanced" technique to expose for the highlights and just pull the shadows up? Well, you get noise - even at iso 320 - what a shame.

The noise appears because of the limited saturation capacity of the sensor (as any sensor has them). Take 8 pictures and stack them and you get rid of the noise.

Direct link | Posted on Dec 22, 2015 at 16:36 UTC
In reply to:

noirdesir: How do they stitch the two sets of Bayer data from the two exposures together to a combined raw file? I would expect some artefacts after such a stitched raw is run through another raw converter when viewed at 100%.

Maybe I am simply too demanding when I think stitching must be done pixel-perfect. Take a flagpole which might easily cross the stitching boundary, you tilt the camera all so slightly between the two exposures and now you have a bent flagpole. And a slight tilt along the vertical axis and the relative position change between objects in different distances.

In a standard bitmap image, the two stitched images can be moved in one-pixel increments to make them match. Stitching Bayer data requires them to be moved in two-pixel increments. Doesn't this reduce the possible relative positioning accuracy between the two exposures?

Direct link | Posted on Dec 19, 2015 at 00:40 UTC

How do they stitch the two sets of Bayer data from the two exposures together to a combined raw file? I would expect some artefacts after such a stitched raw is run through another raw converter when viewed at 100%.

Direct link | Posted on Dec 18, 2015 at 23:19 UTC as 40th comment | 3 replies
On article Sony finalizes buyout of Toshiba's sensor business (91 comments in total)
In reply to:

Horshack: Not great for competition but at least we wont be having debates over whether the Nikon D7300 has a Sony or Toshiba sensor in it :)

Imaging sensor fabs operate at much less demanding process size scales than what Intel is doing. Canon, which is quite behind, has a 500 nm and a 180 nm fab. The former is used for FF sensors where small process size matters less.

Numbers I have seen for the process size used by other imaging sensor manufacturers range from 120 to 300 nm. With 180 nm being good enough for most but the smallest pixels.

Compared to that 14 nm plants like Intel's can easily cost an order of magnitude and more than those imaging sensor plants.

Direct link | Posted on Dec 5, 2015 at 02:23 UTC
On article Sony finalizes buyout of Toshiba's sensor business (91 comments in total)
In reply to:

Impulses: " Except for Canon and Samsung, both of which manufacture their own APS-C and larger sensors, every other camera manufacturer, including Nikon, will now pretty much have to rely on Sony. "

Uhh, Panasonic has used their own sensors in all M4/3 bodies over the last couple years but one IIRC (GH3?). AFAIK they still hold a 49% stake in the sensor manufacturing business they built and probably still design them from the ground up.

Also, rumors keep percolating about where Samsung's camera assets may end up... I'd be surprised if they sold the actual sensor manufacturing division tho since that's probably more lucrative than camera's themselves but at the same time not something a smaller company might be able to take on.

Keeping fabs running thru demand is the only way to make the huge initial investment in them profitable (as well as offsetting the costs to keep manufacturing processes up to date).

Bob Newman (bobn2) has said that the sensor in the Leica SL and Q is a CMOSIS design, which Wikipedia agrees with (like the M 240 for which there is no doubt about CMOSIS). I remember that Leica was quite open about CMOSIS having designed the Leica M 240 sensor but being very cagey who designed the Leica Q sensor. Make of that what you want. The Leica Q and SL sensor is rumoured to be fabbed by a joint-venture of Panasonic and TowerJazz whereas the Leica M 240 sensor is definitely fabbed by STMicroelectronics.

Regardless, image processing, AF algorithms, video codecs, main processor are all areas where Panasonic likely has made some and possibly big contributions.

Direct link | Posted on Dec 5, 2015 at 01:20 UTC
On article Sony finalizes buyout of Toshiba's sensor business (91 comments in total)
In reply to:

Horshack: Not great for competition but at least we wont be having debates over whether the Nikon D7300 has a Sony or Toshiba sensor in it :)

I've put together what seems to be a pretty uncontested list of which company provided the sensor 'design' for Nikon's DSLRs going back to the D2h:
http://www.dpreview.com/forums/post/56873778

But as you said, from now the difference between Sony and Toshiba becomes moot.

Direct link | Posted on Dec 5, 2015 at 00:47 UTC
In reply to:

fenceSitter: DPReview about the Pentax 645Z: "the larger sensor area does mean more light is gathered than comparable full-frame 35mm DSLRs."

I'm afraid this is patently false. Pentax's fastest medium format lenses are so much slower than those available for full-frame DSLRs that it completely obliterates any light-gathering advantage that the larger sensor might otherwise have.

To wit, a full-frame DSLR sensor behind a lens @ f1.4 gathers more than twice as much light as the 645Z sensor behind a lens @ f2.8, and f2.8 is the fastest there is in Pentax's current 645 portfolio.

And I am discussing the fact that for all we know, the 645Z very likely has a higher saturation capacity than (almost) all FF cameras. All the evidence we have (from PDR over sensor age and make to qualitative reports by serious people) is consistent with that assumption and there is no evidence that would indicate the opposite.

And it is worth discussing because that 'higher saturation capacity' was the original point of contention. Settle that and all other discussions become moot.

Direct link | Posted on Dec 3, 2015 at 23:01 UTC
In reply to:

fenceSitter: DPReview about the Pentax 645Z: "the larger sensor area does mean more light is gathered than comparable full-frame 35mm DSLRs."

I'm afraid this is patently false. Pentax's fastest medium format lenses are so much slower than those available for full-frame DSLRs that it completely obliterates any light-gathering advantage that the larger sensor might otherwise have.

To wit, a full-frame DSLR sensor behind a lens @ f1.4 gathers more than twice as much light as the 645Z sensor behind a lens @ f2.8, and f2.8 is the fastest there is in Pentax's current 645 portfolio.

The 645Z uses a modern Sony CMOS sensor (released in 2014), with a pixel pitch of 5.3 μm. It has the same base ISO (100) as almost all semi-modern FF sensors and among the Sony FF sensors only the D810 deviates from this (the Nikon sensors in the D3/D3s/D700 started at ISO 200, the Leica M9 with a Kodak sensor at ISO 80 and Leica SL with a CMOSIS (?) at ISO 50). While ISO ratings have a lot of leeway, past experience shows us that they do indicate roughly the saturation capacity.

But going beyond that, we have many reports from serious testers that the 645Z, ISO for ISO, fares better than (almost) all of the current FF sensors (the D810 potentially coming close).

We also have measurements by Bill Claff that put the 645Z ahead of all FF cameras at base ISO (again with the D810 coming close):
http://www.photonstophotos.net/Charts/PDR.htm

Direct link | Posted on Dec 3, 2015 at 13:51 UTC
Total: 414, showing: 1 – 20
« First‹ Previous12345Next ›Last »