bclaff

Lives in United States Metro-West Boston, MA, United States
Joined on Nov 4, 2006

Comments

Total: 24, showing: 1 – 20
« First‹ Previous12Next ›Last »
On article Sony a9: more speed, less dynamic range (578 comments in total)
In reply to:

bclaff: For those who may be curious I have revised the estimated Photographic Dynamic Range (PDR) available at http://www.photonstophotos.net/Charts/PDR.htm#Nikon%20D5,Sony%20ILCE-7RM2,Sony%20ILCE-9(estimate)
Still looks like the A9 is on track for having about the same PDR as the A7RM2.

Well, the article does say "We've taken an initial look" and my PDR values are estimates so it's hard to say for sure and we'll know better when more data is available.
FWIW, I'd say that even the existing numbers I have put the cameras pretty close together down to ISO 200 or 250.

Link | Posted on May 19, 2017 at 22:41 UTC
On article Sony a9: more speed, less dynamic range (578 comments in total)
In reply to:

bclaff: For those who may be curious I have revised the estimated Photographic Dynamic Range (PDR) available at http://www.photonstophotos.net/Charts/PDR.htm#Nikon%20D5,Sony%20ILCE-7RM2,Sony%20ILCE-9(estimate)
Still looks like the A9 is on track for having about the same PDR as the A7RM2.

Rishi,
Regarding the semantics of "about the same" and "won't reach A7RM2 levels". "About the same" could mean "less than but not by much".
I see no inconsistency.

Note that PDR is not a function of EDR (unlike DxOMark); so that reasoning does not apply.

Note that based on the estimated value the A9 is no more than 0.8 EV below the A7RM2 and that gap is likely to shrink with final numbers.

Regards,

Link | Posted on May 19, 2017 at 19:38 UTC
On article Sony a9: more speed, less dynamic range (578 comments in total)
In reply to:

bclaff: For those who may be curious I have revised the estimated Photographic Dynamic Range (PDR) available at http://www.photonstophotos.net/Charts/PDR.htm#Nikon%20D5,Sony%20ILCE-7RM2,Sony%20ILCE-9(estimate)
Still looks like the A9 is on track for having about the same PDR as the A7RM2.

That doesn't relate directly to dynamic range. It's a visual test for ISO Invariance that is largely affected by whether the noise sources are upstream or downstream of the amplifier.
On my site the best tool for determining ISO Invariance is:
Shadow Improvement of Photographic Dynamic Range versus ISO Setting
http://www.photonstophotos.net/Charts/PDR_Shadow.htm
(The A9 curve is a bit lumpy right now with the estimated values.)
and PDR is best for dynamic range.

Link | Posted on May 19, 2017 at 19:03 UTC
On article Sony a9: more speed, less dynamic range (578 comments in total)
In reply to:

bclaff: For those who may be curious I have revised the estimated Photographic Dynamic Range (PDR) available at http://www.photonstophotos.net/Charts/PDR.htm#Nikon%20D5,Sony%20ILCE-7RM2,Sony%20ILCE-9(estimate)
Still looks like the A9 is on track for having about the same PDR as the A7RM2.

Not sure what you're trying to accomplish. My PDR estimates don't relate directly to ISO Invariance.

Link | Posted on May 19, 2017 at 18:17 UTC
On article Sony a9: more speed, less dynamic range (578 comments in total)
In reply to:

bclaff: For those who may be curious I have revised the estimated Photographic Dynamic Range (PDR) available at http://www.photonstophotos.net/Charts/PDR.htm#Nikon%20D5,Sony%20ILCE-7RM2,Sony%20ILCE-9(estimate)
Still looks like the A9 is on track for having about the same PDR as the A7RM2.

Rishi, not exactly sure what you're asking. The A9 is pretty close to the A7RM2 at all ISO settings estimated except at ISO 100 where it's currently 0.80 lower. My experience is that my estimates tend to be low, especially at low ISO settings; so I expect that gap to narrow. But my initial impression is that below ISO 640 the A9 won't reach A7RM2 levels.

Link | Posted on May 19, 2017 at 17:51 UTC
On article Sony a9: more speed, less dynamic range (578 comments in total)

For those who may be curious I have revised the estimated Photographic Dynamic Range (PDR) available at http://www.photonstophotos.net/Charts/PDR.htm#Nikon%20D5,Sony%20ILCE-7RM2,Sony%20ILCE-9(estimate)
Still looks like the A9 is on track for having about the same PDR as the A7RM2.

Link | Posted on May 19, 2017 at 12:25 UTC as 34th comment | 11 replies
On article Canon EOS-D30 Review (9 comments in total)

Would be fun to do a sensor test on this for historical reasons
(I have Nikon tests back to the D1).
If you still have a D30 and are willing to collaborate, get in touch.

Link | Posted on May 5, 2017 at 15:24 UTC as 1st comment
On article Hasselblad X1D-50c First Impressions Review (316 comments in total)
In reply to:

bobn2: If the GFX50S and X1D have the same sensors then they have the same ADCs, because it's a Sony Exmor design with the ADCs on the sensor. I'm wondering whether the innards of the two might not be very similar, and possibly also the lenses - the H cameras have a lot of Fulifilm in them and Hasselblad's lenses are Fujifilm.

I has been demonstrated by mathematical modeling as well as by read noise measurements on many other cameras that quantization is not a measurement issue until read noise in Digital Numbers (DNs) drops well below 1 DN. (This includes the Nikon D810)

At 14-bits the X1D the read noise would still be about 1.6 DN, below the 1.9 DN figure for the GFX 50S but still well above 1 DN. 14-bits would be enough for the X1D and it's unclear to me why they went with 16-bits.
Your conjecture that the improved read noise for the X1D is better than the GFX 50S due to ADC bit depth is unfounded.

I'm not an EE but I know there are many other factors in the external circuitry and firmware that could affect read noise.
My conjecture would be slower read-out on the X1D versus the GFX 50S; we have measured this effect before on some cameras that support both 12-bit and 14-bit read-out.

Link | Posted on Apr 5, 2017 at 14:13 UTC
On article Hasselblad X1D-50c First Impressions Review (316 comments in total)
In reply to:

dbateman: A 0.08EV advantage sounds to me like you need to look at 10 cameras from Fuji and 10 cameras from Hasselblad. I bet with in the brand you would see a 0.08 EV difference. Ie your within your error for looking at the exact same thing!

Yes, 0.08 is barely within the accuracy of the Photographic Dynamic Range (PDR) test. A slight variation on data collection or sample variation could explain such a small difference. Certainly there is no photographic consequence.

Link | Posted on Apr 4, 2017 at 15:46 UTC
On article Hasselblad X1D-50c First Impressions Review (316 comments in total)
In reply to:

bobn2: If the GFX50S and X1D have the same sensors then they have the same ADCs, because it's a Sony Exmor design with the ADCs on the sensor. I'm wondering whether the innards of the two might not be very similar, and possibly also the lenses - the H cameras have a lot of Fulifilm in them and Hasselblad's lenses are Fujifilm.

Bob, you're very knowledgeable about these things.
Are we sure that all the ADC is on chip and none is external?
Could bit-depth be select-able via firmware?
In any case, 14-bits would be sufficient to capture read noise at even the lowest ISO setting so I wonder if the 16-bit choice was just convenient.

Link | Posted on Apr 4, 2017 at 05:18 UTC
On article Hasselblad X1D-50c First Impressions Review (316 comments in total)
In reply to:

Class A: Your comment about not losing dynamic range when increasing ISO does not make sense at all.

Please look up the definition of ISO. Please look at dynamic range charts published by DxOMark.

The only way to not lose dynamic range would be to not push in post-processing, meaning that you don't effectively increase the ISO value.

The same effect can be achieved by just using any other camera with an "ISO-less" sensor and shooting it at base ISO or at least below the intended final ISO. This gives one highlight protection if needed and all the effective ISO required (up to blowing out highlights).

You are creating a difference between a "traditional camera" and this Hassy, which simply does not exist. I'm sure not even Hasselblad would support your very strange statement.

Regarding dynamic range; here we are tripped up by semantics.
The ISO setting controls two things, exposure and the downstream processing of the sensor.
Above the ISO 1600 setting the downstream processing remains at ISO 1600 while exposure respects the ISO setting.
So no dynamic range is lost, you simple underexpose more and more.
This is what many people do manually when they have an ISO Invariant sensor; so it's nice to see the camera taking care of it for us.

Link | Posted on Apr 4, 2017 at 05:15 UTC
On article Sony SLT a99 II Review (1573 comments in total)
In reply to:

NJOceanView: Did I miss the part where you shoot something six ISO stops too low and push it back to normal to show how much better this is than a Canon?

I imagine you're referring to ISO Invariance. One way to judge that without pixel peeping is to look at the Input-referred Read Noise ( ww.photonstophotos.net/Charts/RN_e.htm#Canon%20EOS%201DX%20Mark%20II_14,Sony%20ILCA-99M2_14)
Here you'll see the 1DX Mark II isn't ISO Invariant until you reach ISO 5000 while the A99 Mark II is ISO Invariant as soon as the high conversion gain kicks in at ISO 400.
FWIW, IMO, this isn't a characteristic that necessarily makes one camera "better" than another.

Link | Posted on Feb 14, 2017 at 12:50 UTC
On article Video: a look at the Sony Cyber-shot RX1R II (125 comments in total)
In reply to:

l_d_allan: Do I need my eyes checked? I'm not "seeing a real difference" between the magnified "glowing 30" numbers at about 1:40 into the video ... comparing uRAW to cRAW.

If anything, the slightly different magnification of the uRAW and cRAW may impact the visual impression created.

@Rishi,
Emil got those values from me and his photon noise explanation explanation is not unlike the CIE Lightness explanation I give in my [NEF Compression](http://www.photonstophotos.net/NikonInfo/NEF_Compression.htm) article.
Also, that excellent Open Forums post is only a portion of his larger article [Noise, Dynamic Range and Bit Depth in Digital SLRs](http://www.photonstophotos.net/Emil%20Martinec/noise.html) (see page 3 The Consequences of Noise for An aside on "lossy" NEF compression)
Regards
P.S. - What is the correct URL syntax in these posts?

Link | Posted on Dec 19, 2015 at 19:25 UTC
In reply to:

bclaff: @Rishi,
As far as I know, every place the article says 14-bit should read 13-bit.
I'm not aware that Sony uses anything higher than a 13-bit ADC in any of their cameras.
The raw data is stored at 14-bit for encode/decode convenience.
Regards,

@Rishi,
FWIW, a 13-bit ADC doesn't limit EDR to 13EV.

Link | Posted on Sep 7, 2015 at 02:42 UTC
In reply to:

cr2shooter: Raw should be just that - raw and unmolested. I wouldn't consider buying any Sony camera when they implement goofy stuff like this. You can rely on companies like Canon to provide a thorough, reliable implementation.

@VirtualMirage,
I'm obviously not entirely conversant regarding raw software for Sony files :-)
But it's still, as I see it, that the solution is not just a firmware change.

Link | Posted on Sep 4, 2015 at 20:30 UTC
In reply to:

cr2shooter: Raw should be just that - raw and unmolested. I wouldn't consider buying any Sony camera when they implement goofy stuff like this. You can rely on companies like Canon to provide a thorough, reliable implementation.

@Rishi,
My main point is that it is not simple a firmware fix from Sony's point of view.
I think they could do the firmware quite easily but unless they modify their software (Image Data Converter? Sony RAW Driver, etc.) it's only a partial solution.
To do one without the other might be viewed as a bad business decision; driving Sony camera owners to other software (even though Sony's software is free).

Link | Posted on Sep 4, 2015 at 18:33 UTC
In reply to:

cr2shooter: Raw should be just that - raw and unmolested. I wouldn't consider buying any Sony camera when they implement goofy stuff like this. You can rely on companies like Canon to provide a thorough, reliable implementation.

@Rishi,
Aren't there multiple moving parts here?
1) 13-bit ADC might not be enough bit depth for the newer better sensors. But probably not a big deal (and this is hardware)
2) Tone curve might not be optimal. Nikon's lossy curve with more and smoother steps is better. But again, maybe not a big deal.
3) 11+7 compression. This has bad worst case behavior and is probably the biggest issue.
Bottom line: Sony doesn't appear to have the will to make the firmware and software changes needed to address these concerns.
It's simple, but not enough, to adjust the camera firmware.
The hard part is the Sony software that would then have to read the new and improved raw files.

Link | Posted on Sep 4, 2015 at 16:21 UTC
In reply to:

bernardf12: The A99 produces 14 bit RAW images from the Bionz X so Sony has no excuse. If the focus points were spread evenly across the frame the A99 would have been an interesting option to me.

Sony is rumored to release the a7000 and the a99ii soon and it will be very interesting to see if they have the same limited RAW capability built in.

I've never seen an A99 file with true 14-bit data; only 14-bit raw with 13-bit data.
If you think you have one then contact me by email or Private Message (PM); I'd like to get that file.

Link | Posted on Sep 4, 2015 at 16:08 UTC
In reply to:

bclaff: @Rishi,
As far as I know, every place the article says 14-bit should read 13-bit.
I'm not aware that Sony uses anything higher than a 13-bit ADC in any of their cameras.
The raw data is stored at 14-bit for encode/decode convenience.
Regards,

True, but why do you think that slope of 2 is there. Surely the Sony guys aren't complete idiots. Note that even Figure 12 of the article you cite (http://www.rawdigger.com/howtouse/sony-craw-arw2-posterization-detection) has a y-axis labeled "Linear 13-bit data in 14-bit space"

Link | Posted on Sep 4, 2015 at 14:13 UTC
In reply to:

bclaff: @Rishi,
As far as I know, every place the article says 14-bit should read 13-bit.
I'm not aware that Sony uses anything higher than a 13-bit ADC in any of their cameras.
The raw data is stored at 14-bit for encode/decode convenience.
Regards,

It's quite obvious from the gaps in the histograms.
This is also how we (Jim Kasson, myself, and others) detect when the camera drop into even lower bit depths.
Try a tool like RawDigger on some files.

Link | Posted on Sep 3, 2015 at 13:50 UTC
Total: 24, showing: 1 – 20
« First‹ Previous12Next ›Last »