Usee: Sorry, but within the comparison tool it wipes away the pattern in the green area around the Nike logo on the shirt of the left player - even at the "Large" setting...
OK, it reduces the file size noticably, but I would have liked a comparison with JPEG 2000 and also different jpeg compressions...
...I see only examples of "noise reduction" with a heavy loss on detail.
They should at least improve the comparison tool, or maybe also, the format......b.t.w. which is the reason why JPEG 2000 isn't used that much?
Jpeg 2000 is there; it's listed in the dropdown as JP2K-Kakadu.
I took some test shots of this lens down at my local photo store. I was disappointed at the sharpness. It would be fine for web resolution, but not for any kind of print at 8x10 or larger. It also has a plastic bayonet mount. Yes it's cheap, but I was hoping for better quality and would have paid more for good optics and a metal mount. I didn't buy it, hoping for a version 2.
While the DR test does look nice for the Panasonic, look at the other parts of the photo and the G7X really shines (as does the RX100 III.) I'm not sure I'd want to pay the extra for the Panasonic and sacrifice a longer lens and superior resolution of the Canon or the quality of the Sony. The Sony lens looks deliciously sharp compared to the others.
Lensjoy: I'd like to see some sample images emphasizing green foliage. I've found the D800 to be weak in hue separation of greens, and it would be a help to see how the D750 compares.
I see it in Raw conversions using Capture NX 2. There's a D800 ad shown in print magazines of a woman standing in a forest with a dress that has a forest print on it. Interestingly, this image has exactly the same issue. All the greens look the same, when there ought to be a lot more hue variation.
I'd like to see some sample images emphasizing green foliage. I've found the D800 to be weak in hue separation of greens, and it would be a help to see how the D750 compares.
goblover: Is it just me, or the Canon G7x sample was motion blurred? And the LX100? If you see the top left and bottom right, the red water colour paint, and colour circle on the bottom left, the letters are blurred. The circle also not a full circle, a bit distorted. I'm not the expert, maybe DP Reviewers can help clarify why is that?
It may be the G7X wasn't aligned parallel to the test scene. Look at the color wheel on the lower left vs. the upper right for the G7X. The upper right is much crisper, and actually a bit better than the RX100 in that part of the image. On the lower left, the two are reversed with the RX100 looking better. Both cameras ought to be re-shot with the lens wide open to check alignment before doing the final test image.
Read the public's comments at this link:
My own comment is below: http://www.lensjoy.com/Blog/Blog_home.htm
Please make sure to add your voice.
Lensjoy: The human eye is most sensitive to green. When calculating luminance (Y) from the respective RGB values for gamma 2.2, the formula is: Y = 0.2126*R^2.2 + 0.7152*G^2.2 + 0.0724*B^2.2Blue is only 7 percent of perceived luminance! So if I were to design a sensor doing what Sigma is purporting to do, I would put the high resolution four pixels in the center green layer, not the top blue layer.
Perhaps Sigma is doing something more complex than the schematic in the article above implies, but from the above description I expect this sensor design to have faults that we'll see remedied in a future version. I wouldn't buy the camera yet.
Thanks, this explanation really helps. My next question would be, what is the exposure latitude like for that top layer since it receives all wavelengths together rather than just red, green, or blue? It will be interesting to see how the sensor stacks up in that department. Since there's no blue color filter, I also wonder how it deals with yellow since a bright yellow will register high intensity on the top layer as well as the red and green layers, making it difficult to distinguish from white light. I've seen some Foveon tests on color separation, but never seen a white/yellow pattern. Maybe there's a reason for that, because it shows the problem. Admittedly, white/yellow distinction is a fairly rare situation photographically.
The human eye is most sensitive to green. When calculating luminance (Y) from the respective RGB values for gamma 2.2, the formula is: Y = 0.2126*R^2.2 + 0.7152*G^2.2 + 0.0724*B^2.2Blue is only 7 percent of perceived luminance! So if I were to design a sensor doing what Sigma is purporting to do, I would put the high resolution four pixels in the center green layer, not the top blue layer.
The schematic of the sensor layout looks wrong to me. The eye is most sensitive to green, and least sensitive to blue. If wanting to capture maximum luminance resolution, the green layer in the middle of the stack should have the highest resolution, not the blue layer.
The idea is interesting, but the execution needs more thought. Without knowing anything about the lens' optical characteristics, I can go into Photoshop and apply these two operations to the original version above: 1. Filter/Sharpen/Smart Sharpen Lens Blur 79%, 5.7 pixels (more accurate checked)
2. Filter/Sharpen/Unsharp Mask 47%, 5.7 pixels
I haven't corrected it for chromatic aberration, but examining the green channel alone it's clear that the corrections above produce a better result with much better separation of tonality in the details, and more fine detail. If I had a chromatic aberration tool I am confident I could outdo their effort. Perhaps someone here can apply that with these steps and post the final result.
What steams me the most about Adobe's plan is that if I have existing files and don't keep my subscription current, I can't edit those files as I could with a perpetual license. Also, if I don't use Photoshop for a month, I'm still paying for it. If Adobe were to go to metered access where I paid by the hour for the product I might be happier than with a subscription. At least I would be paying for what I use. Subscriptions are just a way to get money from people even if they are not doing anything with the product or don't need the new features.
You've probably heard this before, but please focus on photography! The time spent reviewing phone features could be better spent on lens reviews, editing software, camera tips, and other things we care about that are lacking both at dpreview and other places online. I wish dpreview were the go-to place for photographers rather than a me-too place where I can find just another tired cell phone review.
My ideal camera is something like the D800/EOS 5D with a sensor that tilts inside the camera body in both the vertical and horizontal axes. This would offer view-camera adjustments to focus that essentially make any lens into a tilt lens. Since these cameras have large numbers of AF sensors, it should even be possible for the camera to adjust the tilt itself for optimum depth of field in the scene. Build it and every landscape photographer will buy one.
I like the new design overall, especially the preview of replies feature. One thing I still want is an "expand all replies" feature to allow quick reading of all replies rather than having to click and wait for each one to display individually. It's always been a bother to wait for the server to respond and for that reason I seldom read all the responses to topics where I've wanted to do that.
Rachotilko: I would like to ask the informed ones here (people with actual experience with different sensor formats):
Herr Schulz talks about different image characteristics of the bigger formats compared to the 35mm FF. I take his claim seriously - at least because there are some supporting comments in this discussion.
My question is: what is the primary cause of the difference in image characteristics ? Herr Schulz used an analogy: S is better than FF the same way as FF is better than small sensor compact. Which I think is completelly wrong: excessive noise of compacts is sufficiently demonstrable in common print sizes. But D800E is capable of taking high ISO shots without noticeable drop in IQ.
The only part that can bring about advantage in terms of IQ to "above FF" (such as S or MF) systems is *optics*. There simply are limits on what FF lens can do (in terms sharpness, distortions, CA, diffraction), that can be overcome only by means of the format enlargement.
Is this assertion right ?
(continued) If one must stop a medium-format lens down more, doesn't that mean sharpness is reduced? Not for two reasons. First, the pixels are larger at the same resolution on larger formats. Second is that tilt lenses can provide an improvement in DOF so one doesn't need to stop down as much when the scene allows it. That's a reason Leica offers a tilt lens. Of course, so do Nikon and Canon. With the right optics on a D800, medium and large format can still produce a better image. But the differences are small, and what Schulz doesn't say is that the market for that difference is also small. I can produce a better image than Leica at less cost from my 4x5 camera, but at greater overall difficulty. The difference is that Leica produces a better image than the D800 at much greater cost and greater difficulty.
I shoot with everything from a 1/1.7" P/S compact through 35mm FF, 645, 6x7, and 4x5. To some extent there is truth in Schulz' opinion that medium format is different from 35mm at the same resolution. Two major factors come into play. First is that with a larger sensor there is shallower depth of field at equivalent focal length. Portrait photographers like medium format (and large format too) for that reason. The backgrounds and bokeh effects produce better portrait images.
The second factor is that diffraction effects are reduced per pixel on larger formats. This gives a photographer the ability to stop a lens down more using a larger sensor without compromising the image quality. That means that on a D800 one sees loss of sharpness around f/11-16 on most lenses but on medium format it doesn't happen until f/22-32, and large format more around f/32-45. This effect gives the photographer more range of sharpness with depth of field to work with on larger formats.
shifttilt: The Nikon 24mm PC lens is really awful (blurry in the corners and difficult to get full depth of field) not to mention overpriced compared to the Canon 24mm Tilt/Shift II, let's hope this lens functions better and works easily on a D800.
Corner blur can be a consequence of lens design constraints. Good tilt/shift lenses are not flat field, unlike fixed ones. See http://www.luminous-landscape.com/tutorials/focusing-ts.shtml. If the center is sharp, the corners will be blurry. This goes away when stopping down. The best focus is obtained by finding the near and far points in relation to the plane (actually, sphere) of focus and focusing halfway in between, then stopping down. It's not easy to do right, as any view camera user knows.
Government, get out of my camera bag!
I don't see anyone complaining about their cell phone or computer here, but modern digital cameras are really a computer and a lot of software with a big sensor chip wrapped around old-fashioned shutters, mirrors, and optics. Really, it's time to acknowledge that software errors like this are part of technology and stop bashing the manufacturer. We all want perfection, but sometimes problems don't appear until they are out "in the wild" where massive numbers of people exercise every feature of the product more than testers can. Please get out and shoot some photos instead of wasting time on a fact of life.