Pirate Photog: How much bigger would the camera have to be if Fuji put a FF sensor inside with an f1.8 or 1.4 lens? And how much more would it cost to produce?
You're absolutely right - I clearly wasn't having one of my more brilliant moments.
You're right, the *area* needs to double, so the diameter only needs to increase by 1.4x (Root 2).
falconeyes: I don't know if the m4 sensor is visibly noisier than the m3 sensor. I have trouble to see it and prefer to wait for a lab testing.
However, IIRC what Sony said in a recent interview, Sony moved the ADCs from the sensor chip off to the stacked logic layer, using copper vias. This way, Sony can incorporate more ADCs and increase readout speed by a significant factor. Which is used by a much faster electronic shutter and video.
However, it made me wonder immediately if readout noise wouldn't increase marginally. I am especially looking forward to DxO testing on DR at base ISO to have a closer look at that. The DRAM chip though may have nothing to do with this.
If Sony managed to incorporate the ADC array onto a separate logic layer for their larger chips w/o any negative impact on image quality then it will allow for a whole new generation of image sensors. E.g., they could implement arbitrarily low ISO and arbitrarily high dynamic range in the digital domain of the stacked sensor.
The problem is it's entirely up to the manufacturer *or user* to decide which value in a Raw file corresponds to a given output brightness.
DxO uses a part of the ISO standard so that its results are consistent, in terms of measuring the engineering characteristics of the sensors.
We use the level that the camera tries to give you (because that's what most Raw processors will give).
A crude analogy would be whether you measure the performance of an engine in isolation or whether you measure the output from the gearbox that comes with it.;
Both are 'accurate' but both are telling you slightly different things.
To stretch the analogy still further, ISO as it's usually used refers to, say, wheel revolutions per minute, while DxO looks at the number of revolutions the mainshaft of the engine makes. They're measuring similar things but there's no reason to expect them to line up, and comparing them doesn't tell you whether the wheels are rotating as much as the manufacturers suggests.
Mike FL: DPR;
In "Studio test scene comparison" tools, should "Illumination reduced to 8 EV" be " Illumination reduced to 8 LV"?
I refer the honourable gentleman to the answer I gave some moments ago.
El Guapo: Why is the fuji consistently 1/3 of stop brighter (or slower in shutter speed) as compared to all other cameras in the low light studio comparison? That's a third of a stop advantage in image quality. The exposure triangle should be matching for all cameras unless the test for the fuji had different lighting from the rest. What gives?
To expand on what Rishi said:
Our JPEGs show the result you're likely to get if you use the camera's meter or manually assess the histogram to choose exposure (and we report the exposure parameters so you can identify the 1/3EV disparity that you spotted).
Our Raw files are shot with standardised exposure parameters so that each camera is given the same amount of light per unit area, so that you can visually assess noise on a common basis.
Technically, yes, but more people are likely to understand what 'EV' is referring to.
Equally, we're guilty of using F3.5 rather than f/3.5 on occasion.
Dimit: Something's going wrong with DPR's studio tool : RX100iii better than RX100iv ?????Should we worry about the uniformity of the procedure carried out by your photographers?
All testing has inherent error. You can do everything possible to minimise these sources of error but you can't eliminate them and the net result of all of their contributions.
We tried shooting this camera several times to ensure it wasn't an alignment or focus error on our part and we simply can't get better results. The differences you're seeing are the differences that are likely to exist between picking up any two copies of these cameras (and neither shows any sign of being 'broken' or looks likely to be out-of-spec).
There are slight differences but there's every chance that if we asked for another III and IV we'd get the opposite results with those copies.
Arguably the problem is that our test scene tool lets you analyse differences smaller than the combined error of the test.
Mike FL - please refer to my answer to your other post.
DxO does not measure the part of the ISO standard that the manufacturers (and most people) are using. They are not showing a difference between truth and claims, they are showing the difference between the measure of sensitivity that they use and the measure of sensitivity that manufacturers use.
Our test scene is not based on manufacturer claims and all results are fair and consistent with almost all real-world usage.
Mike FL: Rishi Sanyal;
Regarding to your statement of "Actually, our results are entirely comparable between cameras despite us not reporting measured vs. manufacturer ISO because for any given ISO, we always give cameras the same exact exposure and amount of light.".
Really? Lets selecting RAW and ISO=200 in DPR's "studio test scene comparison tools":
- Nikon D4: F5.6, 1/100s
- Fuji X-A1: F5.6, 1/60s
Will you still say " for any given ISO, we always give cameras the same exact exposure and amount of light."?
Mike FL - The JPEGs will still be the ones that give the correct brightness, so you can infer something about ISO accuracy from that (ISO as you encounter it being an ISO-based standard).
So we give you the information necessary to judge how the sensitivity compares to your expectations (it's more complex than just saying accuracy - ISO is essentially arbitrary, at this point), and, in Raw, we give you results shot on a consistent basis, so that you can compare noise results.
I repeat, though: DxO does **not** test 'ISO' as you recognise it. They use a different part of the standard in order to normalise their results. It's a perfectly sensible thing to do but a camera can be entirely ISO accurate (and consistent with reasonable expectations of what that might mean) and still not line up in DxO's numbers. Because that's not what they're testing.
Otto K - some systems don't have the lenses we'd ideally choose. We tried a series of lenses with the Samsung NX1 and the zoom was the best performing one that we had access to at the time.
We hope not to have to use that zoom long-term.
wombat661: At least you try to be less biased, but still got some ways to go...
For sports nothing beats DSLR for focus tracking especially in low lights AND you have a fast lens that needs accurate focus. All those tests for mirrorless use small aperture and bright light, so everything are in focus anyway. Is a lie just like Olympus M4/3 mirrorless claim to be as good as APSC when they just "mis-labled" their ISO setting. For that reason, you can't recommend mirrorless for the enthusiast.
Mirrorless takes time to turn on (even if that time is short), so you can't capture split second moment i.e. kids and babies, unless you set it to be on all the time, and that eats thru batteries. That time for EVF to turn on will irritate the hell out of some users. For that reason, you can't recommend mirrorless for the enthusiast unless you tell them what they are getting into.
Lastly, lens for Mirrorless and DSLR weights about the same. Been discussed before with data.
DxO uses a part of the 12232 standard, but it isn't the one everybody else is using when they say ISO. It's the one that makes most sense for them to use but it isn't what manufacturers (or most people) mean when they say ISO and there's no reason that the two would overlap.
As DxO says: ["As tests show, the ISO settings reported by camera manufacturers can differ significantly from measured ISO in RAW. This difference stems from design choices, in particular the choice to keep some “headroom” to avoid saturation in the higher exposures to make it possible to recover from blown highlights."](http://www.dxomark.com/About/In-depth-measurements/Measurements/ISO-sensitivity)
The fact that people keep incorrectly interpreting this data as 'cheating' or inaccuracy shows that bringing this other standard into play has done the opposite of clear things up.
In fact this risks hiding cameras that genuinely do give unexpectedly dark images, for a given exposure and lighting level.
DxO does *not* measure ISO accuracy.
Their graphs do not show 'claimed' ISO against 'actual' ISO, they show ISO (a JPEG standard) against a saturation-based assessment of sensitivity. DxO notes that you should not expect the two to line up.
Each manufacturer chooses which value in its Raw file it wishes to use to represent middle grey. This is a trade-off between greater highlight capture and increased shot noise.
Almost all Raw processing software (include DxO's, so far as I know), recognise this chosen value and use it as the starting point for processing.
Canon's choice of tone happens to be closer to the one that matches the standard DxO uses - it doesn't mean it's more accurate or truthful, it just means they've chosen a particular noise/highlight trade-off. The D7200 there's around a 0.5EV difference, with the Olympus it's near 1EV.
Indeed by that logic, the E-M5 II's ISO 100 is more true (?) than the 70D or D7200's are. I'm afraid you've misunderstood the data.
Mike FL - that is essentially the way we work: as close as possible to 85mm equiv on cameras with built-in lenses. Prime lenses with roughly equivalent FL for ILCs.
The reason we don't use the Av/Tv that gives the correct brightness for Raws is that 'correct brightness' is a moving target, when it comes to Raw. We used to use the Raw that corresponded to the correct-brightness JPEG (again, based on the assumption of using the camera's meter and JPEG output as a guide for exposure).
However, while this can be argued to be a good choice in terms of relevance, it means that you can't compare Raw noise performance on a common basis, since some cameras will have received more or less light than others. This is why we now use fixed shutter speeds and aperture values.
Yes, in PP.
I'm not sure I understand the second part of your question.
So our current protocol is: use the same lighting level for all but the highest ISO settings (sometimes the lack of fast enough shutter requires the lights to be reduced by 1 or 2EV - but it's an edge case).
JPEG: We display the image that most gives the correct image brightness (because the camera meter and human behaviour will tend to make this the result most people will get). Then report the shutter speed and aperture value.
Raw: We display the image shot at a chosen shutter speed and aperture combination, brightened or darkened to a standard brightness. This should show any real-world impact of any behind-the-scenes decisions the manf has made about ISO mapping.
From what I can remember of the X100, yes I'd expect it to require more exposure or greater ISO to achieve the same image brightness as most other cameras (though not a whole stop).
While what you ask for is entirely reasonable, a test that is fair, accurate and relevant is a lot harder than it sounds.
We changed our test protocol a while ago, from one that was fair and accurate to one that tries to be more relevant for Raw (it was slanted towards JPEG shooting and assumptions about the use of the camera's meter before). We'll be trying to mark the shots that pre-date this change as soon as we can.
Here's a question to ponder (at it relates to the thing DxO *is* showing): how do you assess the 'brightness' of a Raw file?
Mike FL - My point is that DxO doesn't show ISO accuracy. It's simply not something they test. The graphs you're looking at essentially show how the manufacturer chooses to expose its Raw files. ISO as it's stated on the camera is only concerns with JPEG brightness, so cameras can appear almost anywhere on those graphs and yet still give the correct JPEG brightness.
The interesting situation with Fujifilm is that their cameras appear to give JPEGs that are darker than you'd expect, when exposed to a given amount of light. *This* is what raises questions about ISO accuracy (though there are sub-sets of the ISO standard that essentially even allow this), but it's not what DxO is showing you.
CarolineL: I am REALLY upset about the loss of Aperture. I write a blog and bought Aperture last year and LOVE it! IPhotos has lost so many of the features I use. What is the alternative, Adobe Lightroom or is there something else? It seems a big leap to have to pay a monthly sub as compared to buying $79 Aperture. Also does any one know whether you can still buy a photo book from Apple using Aperture?
You only have to pay a monthly subscription for Lightroom if you want the CC version and the rolling updates that come with it.
There is also a standalone version (Lightroom 6) that costs $149 but whose feature set won't change until the next paid-for upgrade, whenever that comes. It also lacks the cloud-based features of the CC version.
Some of the older cameras were not shot using the newer protocol for Raw shooting. We're trying to identify these so that we can mark them as non-standard. That should clarify matter.
It does appear that Fujifilm cameras are mis-reporting their ISO settings (though the ISO standard allows essentially endless wiggle-room). We used to test and report this in our reviews.
However, this is not what DxO shows. A more accurate description of the axes on their graphs might be: Saturation-based sensitivity vs ISO, since the 'Manufacturer ISO' is the thing almost everybody is referring to, when they say 'ISO' and the standard DxO is using is not (though it makes sense for their testing).
The Sony RX1 should give you a pretty good idea, since that's a 35mm F2. An F1.4 would be considerably larger again, since its entrance pupil would need to be twice the diameter.
Eyechap: It takes a while to work out how to get the best out of this camera. I'm glad I waited through 100, 100S for the 100T
Use: Set-up #2 >> selector button setting >> FnThen : Shooting menu >> Function (Fn) setting >> Fn 3 >> Focus area
Result? You can still have 7 programmable Fn buttons AND quick access to the Focus Area and not just the 3 programmable buttons that were suggested in the review.
Having one of the Fn buttons taking you into AF point mode is the default camera behaviour, I believe. However, this doesn't quite give the immediacy of direct AF point selection - which is what's new for the T, and something I really appreciated.
I think I wrote that page of the review too much from the perspective of someone who's used the X100 and X100S extensively. I'll try to find time to go back and make sure I've made clear that you can still set one button to enter AF point mode, if you don't need direct AF point selection.