 Comment Diego's lens is excellent - there's only a handful of ridiculously expensive T/S or Zeiss prime lenses that might offer even better image quality. Please don't forget the 5DSRs' sensor is very demanding and the resolution provided here is very generous. --Martin Falbisoner (talk) 08:46, 23 January 2017 (UTC)[reply]
- Agree. And that's exactly my point. A lens being good doesn't mean it will be a match for a (quoting yourself) very demanding sensor. And on top, it seems f/11 is not an optimum setting for sharpness for this lens. - Benh (talk) 09:07, 23 January 2017 (UTC)[reply]
- Wrt "optimum", the scene is not flat, so at f/5.6 most of the outer fifth of the scene would be blurred. Actually, looking at some reviews, f/11 is within 85-90% of the maximum sharpness so concerns about diffraction softening are often taken too strongly compared to concerns about DoF (or other factors such as fast shutter or low ISO). The "Photozone" website, for example, does not show its MFT graphs with the Y axis at 0. See Truncated graph on Wikipedia: "These graphs can create the impression of important change where there is relatively little change." -- Colin (talk) 09:26, 23 January 2017 (UTC)[reply]
- Comment The 5DS 50.6MP sensor has a pixel pitch of 4.14 microns compared to the 3.92 microns of my Sony A77ii's 24MP sensor. So the current crop of 24MP APS-C DSLRs still are more demanding of a lens than any full frame camera. They do, however, take their image from the sharper central portion. From what I have read, that's a great lens and 16mm on a full frame is a demanding wide angle. The f/11 will have reduced overall sharpness a little, balancing the need for DoF in this image where the scene is not flat, but a ceiling and upper walls. I'm not sure one could do better, and yes, this is a 43MP image so quite amazingly detailed overall. -- Colin (talk) 09:11, 23 January 2017 (UTC)[reply]
- @Colin: Did a quick check with a DOF calculator. Roughly I estimate the ceiling is 8 to 10m high and the border (frame) which can be seen must be at 4-5m high (I did check photos of the building). On a Nikon 810 (36mpix) at f/8, hyperfocal distance is 1,08m yielding everything from 54cm to infinity being in focus. Probably a bit less on a 5DS, but my bet is that the ceiling here fits. Even at f/5.6 it's very hard not to fit the whole ceiling in the "focused area". And to the concern of pitch, I'm not sure the "simple" thinking here is appropriate. It would apply if a lens designed for APS-C is a cropped version of a lens for FF, but I wonder if it's the case. - Benh (talk) 11:20, 23 January 2017 (UTC)[reply]
- I only mentioned APS-C wrt to the pixel pitch and the "very demanding" comment. The problem is that "DoF" calculators are still stuck with a mindset of an 8x10 inch print viewed from a comfortable distance, and they don't take MP into account because they are about viewing the image as a whole, rather than a magified portion of it -- they don't assume looking at a portion on a 27" 100DPI desktop monitor at 100% viewed from 50cm. I think the "pixel-peeper's hyperfocal distance" would be much further :-). Here is what a 24MP camera would have taken -- the image downsized to 73% and roughly 23MP (as it isn't quite 3:2 ratio). That's really sharp and I don't think anyone would question the lens choice or aperture choice. Unless your 100DPI monitor is two metres wide, Benh, you aren't actually looking at the whole image at 100% pixel size. So comments about DoF calculators, made at that high magification, aren't appropriate. If viewed on a 218DPI 5K monitor, it looks plenty sharp enough. -- Colin (talk) 11:37, 23 January 2017 (UTC)[reply]
- You're right, my mistake. Thanks. Just realised now that the DOF calculated is the same for a 5D and a 5DS, which makes no sens. Am looking for a DOF calculator which takes sensor res as a parameter (and this meets Code's issue which his church interior, below). And I'm not criticising the image quality as a whole (otherwise I wouldn't have supported it) but it's a fact that the sensor is too much for this lens. - Benh (talk) 12:27, 23 January 2017 (UTC)[reply]
- And btw, I'm not saying by any mean that Poco should downsample, even though it looks a tad soft at full size viewing. Just commented what came to my mind :) It's always better to have the biggest resolution of course. - Benh (talk) 12:38, 23 January 2017 (UTC)[reply]
- This DoF calculator does let you specify a 2m wide print viewed from 50cm, though that still assumes a print rather than a pixel-based device. DoF is not dependent on the megapixels of the sensor, so a Canon D5S would have the same DoF as a Nikon 810 -- you won't find a calculator that takes MP as an input parameter. Provided the resolution of the sensor does not limit the resolution of the print, then it cannot play any significant role. For an 10" print, an 8000 pixel image like this corresponds to 800dpi which exceeds the best printer requirements. The standard DoF test assumes one does not see blur below 0.01 inches at 25cm viewing distance, but many people can in fact see finer detail. Given that 240 and 300dpi is acceptable for high quality printing, that would be the absolute limit of sharpness achievable. In other words, if this photo was only 3000 pixels long it would print at maximum resolution and be gloriously sharp everywhere in a 10" print. Benh, you are pixel peeping, not actually judging the sharpness of the image as a whole. The lens is fine, your expections for a single-frame photo of a scene with significant depth, is not IMO. According to ephotozine's MFT graphs, the difference in peak sharpness between f/8 and f/11 is 5-10% which I doubt very much you could see with a real-life scene, and only really something an instrument could measure. To be honest, applying some stronger sharpening to areas of the picture is probably going to improve it more than dreaming about a mythical Zeiss Otus 16mm prime, but not worth doing. Once we all start judging FP on 200+ DPI monitors, I think these concerns will fade. -- Colin (talk) 12:47, 23 January 2017 (UTC)[reply]
- Colin, Benh: after all this discussion, is there anything I could/should have done better than I did?. My modus operandi here (and always for this kind of shot) is the following: With autofous I focus on a spot that is aproximately 2 meters closer than the distance that I would like to focus on. I switch then the MF and (in this case) set the autobracketing function. I leave then the camera on the floor, release the 10 seconds and keep people away :) I checked and still believe that F/11 is the optimal f-number. So, any clue have to do it better? Poco2 18:28, 23 January 2017 (UTC) PD: I am happy with this lens, that is something I couldn't affirm from my former Canon EF 8-15mm L USM (this one couldn't keep up with the new sensor)[reply]
- Poco, you are far more experienced in photographing interiors than I am. One could use a DoF calculator but that assumes you know accurately the distances to the ceiling, etc, and have the time to faff around with that. But my point is that all standard DoF calculators are less demanding that a Commons reviewer, so you need to take those numbers with a pinch of salt. Perhaps your rule-of-thumb is working fine. Perhaps you could have got away with f/8 but perhaps not. Although f/8 is measurably a little sharper if the central portion is in focus (which is not the case here, if you focused 2m nearer), it will not be as sharp for areas further away from that focus plane. Only really your experience can be a guide. I do not find the degree of sharpness at 100% 100DPI to be troublesome in the slightest. I also have a 5K monitor which has 217DPI. If I look at this image full screen it is 52cm x 34cm which is larger than a double-spread in a photography magazine and larger than most "coffee-table" art photography books. It has sharpness to rival the finest colour print and this photo looks absolutely excellent with crisp detail all over. I think it is wonderful that we can capture such amazing images in a single frame and that you have offered it for free. For a single-frame photo, I'm really not sure you could have done much better. You should be very happy with this. -- Colin (talk) 22:19, 23 January 2017 (UTC)[reply]
- "DoF is not dependent on the megapixels of the sensor" Colin, this is only half true. If we assume a fixed print size with a fixed viewing distance it is true. However in the world of digital we gravitate more and more towards a fixed pixel count on a given area, which means that when you increase resolution you also increase the size of the circle of confusion as well (in essence print bigger and/or look closer). In that regard "depth of field" IS decreasing with the increasing megapixel count just as DoF at a certain f/ number is decreasing with the increase of the negative size. For digital we would need to replace print size with DPI to be able to do any meaningful DoF calculations. THAT would be truly independent of pixel count. — Preceding unsigned comment added by KennyOMG (talk • contribs)
- As with all these things, it depends on your definition. Per This DoF calculator. The circle of confusion is a physical area formed on the sensor. What matters is how big this is (aperture, focal length, focus distance) and how big that circle is then magnified when printing. That's why the calculator takes the print size and sensor size into account. It does not take megapixels into account because it assumes (rightly) that the dpi of the sensor will exceed the needed print dpi for modest sized prints. At 240dpi printing, a 6 megapixel camera is sufficient to produce good quality sharp A4 magazine photo. Any increase in megapixels beyond that will not alter the DoF when viewing a magazine photo, and decrease in megapixels below that will significantly impair the overall print quality [of course, there will be some benefits to exceeding any "minimum" by some margin]. You are right that increasing megapixels lets us print larger photos that are acceptably sharp. But it is the increased size of such prints that then places increased demand on DoF, not the MP of the camera sensor. With computer monitors, we have effectively a large rectangular magnifying glass, where we can blow up the image to such a degree that every grain making it up (pixel) is visible to the eye if we have decent eyesight. With a 200+DPI monitor, "retina" display, this is not possible at reasonable viewing distances. Monitors are likely to settle around the 240/300DPI level eventually, thus matching quality print for definition. Then we can all just view and review the images "full screen" rather than at 100% magnifying-glass view. -- Colin (talk) 22:19, 23 January 2017 (UTC)[reply]
- Colin, what I'm saying is that print size doesn't matter at all nowadays. Most images are viewed at 3 sizes, I think. 1) thumbnail, 2) resized to full window/screen, and 3) 1:1 pixel size (ie "100%"). Acceptable DoF in this setting depends on sensor pixel size and screen dpi instead on print size and viewing distance. We're basically trading a fixed print size for a fixed pixel pitch when viewing. In that regard even the screen dpi is irrelevant in terms of bigger resolution = shallower DoF, when viewed on the same screen. All I'm saying. :) KennyOMG (talk) 18:18, 24 January 2017 (UTC)[reply]
- The concept of DoF wrt "100% size" in a browser is totally meaningless. This discussion was about lens sharpness vs sensor pixel size and about optimal DoF. Whenever someone views an image at 100%, all bets are off wrt any parameter. It is a completely arbitrary magnification factor, nearly always involves looking at a tiny portion of the image, and is about as sensible way to judge whether an image is "among our finest" as an art gallery judging artworks while looking at a tiny square of one with a microscope. -- Colin (talk) 19:08, 24 January 2017 (UTC)[reply]
- Colin, when media is consumed 99.99% on screens with varying dpi latching on to some hundred-hundred and fifty years old definition of DoF does nothing for us. I agree that the world is moving towards 250-300 dpi screens (sadly), I think we should measure DoF based on that. Ie my current screen is ~170 dpi, if I resize a given image to ~60% that should be my guide whether said image has sufficient DoF or not for our current world. And yes, increasing the resolution will and should put more emphasis on correct bracket and focusing. KennyOMG (talk) 15:27, 26 January 2017 (UTC)[reply]
- KennyOMG, consider that the only place in the world that tiny rectangular portions of 40MP images are viewed at 100% size is .... Commons FP/QI. The rest of the world, who are not nearly so crazy, like to look at the whole of a photo and enjoy it that way. The best way to look at an digital image is full screen. Then the DoF is dependent on how big your screen is and how far you sit from it. In fact, most people look at digital photos on smartphones and tablets, with desktop monitors becoming a minority platform. So I maintain that DoF is somewhat meaningless and any idea that one might "measure it" just ignoring how photos are actually viewed. The one advantage of the old "8x10 inch photo viewed from 25cm" standard was that it was a standard and one can then compare the DoF effects of different lenses and sensor sizes and subject distances. But I agree that actually using the old DoF values is not really acceptable. The image-capture resolution of a full uncropped modern sensor currently exceeds the resolution of any common monitor (e.g. an HD monitor is only 2MP, a 4K monitor is about 8.3MP, a 5K monitor about 14.7Mp, the newly announced 8K monitors are 33MP, and these are 16:9 aspect ratios). Therefore our camera megapixels are not relevant when determining fullscreen sharpness, unless one starts cropping. -- Colin (talk) 16:25, 26 January 2017 (UTC)[reply]
- Let's just agree, Colin, that in my humblest of opinions you seriously underestimate the number of people that click/tap on a picture to view it 100%.. ;) KennyOMG (talk) 20:54, 26 January 2017 (UTC)[reply]
- I'm talking about photography in general. You seem to be basing your artistic choices as a photographer round the poor UI of MediaWiki's file description page. KennyOMG, I have lots of books on photography. Not one of them has ever suggested: "It is important, when setting your aperture and focus for appropriate depth of field to render your artistic vision in all its glory, to consider how the image looks when viewing a tiny portion of it at 100% on a desktop monitor". -- Colin (talk) 21:44, 26 January 2017 (UTC)[reply]
|
Info Ceiling in one of the rooms of Hasht Behesht, Isfahan, Iran. The palace, built in 1669, and which name means "Eight Heavens" is the only one left today out of forty mansions that existed during the rule of Safavids. All by me, Poco2 20:57, 22 January 2017 (UTC)
Support -- Poco2 20:57, 22 January 2017 (UTC)
