JeremiahL: Is it just me, or are the full size images incredibly noisy? I have a several year old Fujifilm X-Pro, and get less noisy pics. And the images in the D810 sample gallery blow this out of the water at comparable ISOs.
I know they're huge images and can be resized, but what's the point if they *have* to be resized due to noise?
I'm a little disappointed because I've been waiting for this camera to come out so I can decide between it and the D810. At this point it's looking like not much of a decision :-/
I'm not arguing against studio comparisons; I'm pointing out that, contrary to the apparently prevalent notion that casual photos are worthless for evaluations or comparisons, these photos contain a wealth of information that gives great insight into the performance of the camera and that can be used for comparisons if sufficient care is taken.
Moreover, the regrettable tendency of recent dpreview photographers to post heavily worked sample images, degrades the value of the samples while offering little in exchange.
@photominion: If the difference between two systems is small enough that we're down to differences that could reasonably be attributed to the transmission of the lenses, I'm happy to call it a draw.
@Eric: No. For all of the reasons I already explained, the information to do such comparisons are present in the files (with caveats for ISO calibration and reasonable exposure durations).
First, we absolutely do want to see the effects of photon shot noise. Photon shot noise is the visible manifestation of the effects of the pixel capacity (well depth) and other important sensor properties. Higher capacity means more shot noise in total, but higher SNR in the final image. Second, even if we wanted to have the same photon shot noise between the two cameras, the steps you suggest would be far from adequate. The steps you suggest would only control the photon irradiance. Properties of the sensor, such as quantum efficiency, pixel size, and well capacity, would influence the total photon shot noise. These are some of the most important properties determining the performance of the sensor and are things we'd definitely want to see the effects of in any comparison between cameras.
There's nothing particularly difficult about using shots taken in less than controlled conditions. Of the criteria I gave, the only one that is tricky is confirming that the ISO ratings are comparable. Everything else can be done by looking at the files and absolutely does not require controlled conditions.
The ISO issue requires access to the cameras for careful tests. Until such tests are done, it's best to draw provisional conclusions and wait to confirm them until the ISO ratings are confirmed to be comparable.
The notion that the total amount of light in the scene has an influence on noise performance has scant basis in reality. A simple experiments demonstrates this: Take 2 shots. For the second shot, cut the amount of light in the scene in half, but double the exposure. So long as you are dealing with exposures comfortably under 1 second, there will be no difference in the noise levels between the two shots. The reason is that the sensor sees the same amount of light in both cases and every noise source, except dark current noise, will be identical between the two shots.
The reason why I said the keep the exposure below 1 second is that for modern sensors dark current is very small and should not be a significant noise source for short exposures.
Comparing across scenes and cameras works fine if you compare regions with the same brightness level in the files, use short exposures, use the same white balance, and make sure the ISO ratings are comparable.
I can see cases where exposing for the purpose of exploiting raw malleability would make sense, but can you justify shooting all of your shots that way for a sample gallery? Why not bracket?
@Revenant: I don't think there's anything that can be concluded from these galleries. The current approach hides a large number of choices made by the photographer in raw conversion. There's no way of knowing whether what we're seeing reflects those choices or gives us insight into how the camera performs.
A more transparent way to approach things would be to shoot raw+jpeg, then post both jpegs and raws. This this would show what the camera's jpeg engine can do and then also provide raw files for people to explore to see how much things can be pushed, etc.
Rishi - then what is the value of these galleries? Back in the day, dpreview photographers posted unmolested jpegs and they were useful for exactly what they were - examples of how the camera's jpeg engine worked. Recently, dpreview has started posting images with unspecified tweaks applied in raw conversion. Good images show what is possible, but anything short of a perfect image leaves questions about what went wrong. How much was the image pushed? Is there noise because there are issues with the camera, or because the shot was underexposed and pushed? A poorly exposed jpeg is obvious as a poorly exposed jpeg. There's nothing misleading about it. jpegs answered a clear question. It wasn't the question that everybody was asking, but they did provide a clear answer. raw conversions with unspecified edits don't answer any question. I clicked through your photos for entertainment purposes, but wasn't able to infer anything about the limitations of the camera based upon what I saw.
ThatCamFan: Judging by both the ACR and JPEG samples most of the shots seem out of focus or is the camera that bad? I have seen one photo so far out of 15 I checked that was actually not out of focus.
Agreed - it's focused on the right hand of the lady on the right. I'd also like to point out what was done to this shot. The notes say, " Exposure +0.70 | Highlights -84 | Shadows +84 | Clarity +10 | Vibrance +35 | Curve Adjustments: Lights +10, Darks +15" So, we have a shot that's not in focus and that was seriously underexposed due to backlighting. The final result isn't sharp and shows artifacts of noise reduction and reduced dynamic range from pushing the shadows so hard. At best one might imagine using such a photo as part of discussion about shadow detail. I don't see a reasonable role for so many shots like this in a gallery of this type unless they are trying to make a point about a badly malfunctioning camera or poorly performing photographer.
Frank_BR: "The E-10 looked like an SLR, but was actually a fixed-lens camera."----------------------------------------------------------------------------------I think the E-10 is a SLR. The mirror is the beam splitter.
Thank you! The E-10 was indeed an SLR and the fact that it was a fixed-lens camera has nothing to do with this. I'm disappointed that even dpreview is getting this kind of thing wrong these days.
thielges: I think that the title should read "CMOS *Image Sensor* Inventor..." CMOS itself (as a platform for digital logic) was invented back in the 1960s by Frank Wanlass.
Eric - I have elsewhere referred to you as the inventor of the "modern CMOS sensor" (with image implicit). I think this is not overly long or technical, yet still fair both to you and to those upon whose shoulders you have stood.
Eric, I pointed out above that you cited Gene Weckler. You have been clear about Weckler's contribution in your writing. You give Weckler credit for the basic passive pixel which begat the early passive pixel CMOS sensors. You have also been clear about the history of active pixel designs before this idea was applied to CMOS. I have no problem with YOUR scholarship and credit attribution on this.
I don't think it's correct for dpreview to say that you invented the CMOS sensor, or the CMOS image sensor, and I doubt you think that's correct either.
Words have meaning and the truth matters - perhaps not to everybody, but to people who understand what the words mean, the truth often does matter. I suspect it matters to Gene Weckler. Perhaps some day you will invent something - or perhaps you already have. In any case, I hope that you will get proper credit for it.
He also didn't invent the CMOS sensor. He invented the active-pixel CMOS sensor. CMOS sensors are attributed to Weckler, 1967. Fossum himself cites Weckler for this in his papers on the topic.